Eddie Reynolds

Over +20 years in sales, customer success and marketing, I’ve developed a passion for data-driven strategy and process. I’ve always been fascinated by the edge it can give an IC or an entire company. 

Ten years ago, I joined Salesforce as an AE covering high-growth, VC-backed B2B SaaS companies. I met with hundreds of executives and revenue leaders and they all wanted this edge. They wanted to hear how Salesforce and my other customers operated, not how the software worked. 

Our consulting partners, however, were solely focussed on the software, often missing the broader strategy. This inspired me to start a unique consulting firm, merging system implementation with data-driven GTM strategy. 

Since then, I’ve seen some of the deepest ingrained revenue growth problems plaguing B2B SaaS – and we’ve developed a system to solve them.

𝘾𝙍𝙊𝙨, 𝘾𝙈𝙊𝙨, 𝘼𝙉𝘿 𝙑𝙋𝙨:

Does this sound familiar?

💭“I don’t understand what my data is telling me. I’m not even sure it’s accurate.”
💭“We’re getting tons of leads but they’re not converting enough to outpace churn.”
💭“We’ve spent so much money on tech, marketing, training, etc. – but there’s no ROI.”

Or maybe even…

“I don’t want to end up like those guys ☝️ How do I get in front of this?”

𝙔𝙊𝙐’𝙍𝙀 𝙄𝙉 𝙏𝙃𝙀 𝙍𝙄𝙂𝙃𝙏 𝙋𝙇𝘼𝘾𝙀

I founded Union Square Consulting to help B2B recurring revenue brands increase profitability by targeting the source of their revenue engine issues (not just the symptoms).

𝙒𝙃𝘼𝙏 𝙒𝙀 𝘿𝙊

Fractional RevOps that delivers both systems implementation and go-to-market strategy.

𝙒𝙃𝙊 𝙒𝙀’𝙑𝙀 𝙃𝙀𝙇𝙋𝙀𝘿

High-growth brands from Series A to C and beyond like Techstars, Luminate, and Federato.

𝙂𝙀𝙏 𝘼 𝙁𝙍𝙀𝙀 𝘾𝙊𝘼𝘾𝙃𝙄𝙉𝙂 𝙎𝙀𝙎𝙎𝙄𝙊𝙉

Sit down with one of my Senior Revenue Strategists for a personalized Revenue Operations Coaching Session. Walk away with advice you can use immediately.

👉 CLICK HERE: unionsquareconsulting.com/workshop

𝙎𝙐𝘽𝙎𝘾𝙍𝙄𝘽𝙀 𝙏𝙊 𝙊𝙐𝙍 𝙍𝙀𝙑𝙊𝙋𝙎 𝙉𝙀𝙒𝙎𝙇𝙀𝙏𝙏𝙀𝙍

Get actionable and practical tips on optimizing your marketing, sales, and customer success engines.

👉 CLICK HERE: unionsquareconsulting.com/newsletter

Host

RevOps Corner is a podcast we produce at Union Square Consulting to share best practices for Strategic Revenue Operations. 

We do this by interviewing amazing guests, ranging from Revenue Operations Professionals to Revenue Leaders to Venture Capitalists and many other experts in the B2B SaaS community. 

We also share what we're seeing in the trenches every day in our work as Revenue Operations Consultants here at Union Square Consulting.

Listen to the Podcast on:
Spotify: https://open.spotify.com/show/1CM3rmelcdtm6gJVL0PtMK?si=00aae750ea7a4937
Apple: https://podcasts.apple.com/us/podcast/revops-corner/id1619571311

Just received the perfect cold call intro.

“Hi, I’m calling from Acme. This is a cold call. I’ll take 30 seconds to explain why I’m calling and you can decide if it’s worth discussing. Fair?”

I really can’t think of a better way to do this. 

As the recipient this felt:
 - Respectful
 - Honest
 - Efficient

What do you think?

CEO | GTM Strategy & Ops for B2B SaaS CROs

You're already spending $2 to generate $1 in ARR.
And you're asking about the ROI of GTM Ops?

𝐇𝐞𝐫𝐞'𝐬 𝐭𝐡𝐞 𝐚𝐯𝐞𝐫𝐚𝐠𝐞 𝐩𝐮𝐛𝐥𝐢𝐜 𝐒𝐚𝐚𝐒 𝐜𝐨𝐦𝐩𝐚𝐧𝐲
- Spends 33% of revenue on sales and marketing
- For a $100M company, that's $33M
- To generate $16M in new ARR
- Costs $2.00 to grow ARR by $1.00

𝐍𝐨𝐰 𝐥𝐞𝐭'𝐬 𝐬𝐚𝐲 𝐰𝐞 𝐬𝐩𝐞𝐧𝐝 $𝟐𝟓𝟎𝐊 𝐨𝐧 𝐆𝐓𝐌 𝐎𝐩𝐬
- To match our current ROI
- We'd need ARR to grow by just $125K
- That's it

You're telling me we can't get $125K in additional ARR?
From fixing broken processes across your entire funnel?

𝐇𝐞𝐫𝐞'𝐬 𝐰𝐡𝐚𝐭 𝐰𝐞'𝐫𝐞 𝐚𝐜𝐭𝐮𝐚𝐥𝐥𝐲 𝐚𝐬𝐤𝐢𝐧𝐠

Should I invest in fixing:
- Inbound lead routing and follow-up
- Pipeline qualification and management
- Sales handoff and onboarding
- Customer health tracking and renewals

Or keep spending $2 to make $1?

The answer should be obvious.

𝐁𝐮𝐭 𝐡𝐞𝐫𝐞'𝐬 𝐭𝐡𝐞 𝐫𝐞𝐚𝐥 𝐩𝐫𝐨𝐛𝐥𝐞𝐦
- Success isn't about financial investment
- It requires alignment
- Commitment from leadership
- Accountability across the GTM team

Without that, no amount of money helps.

With it, the ROI is obvious.

𝐈𝐧 𝐥𝐚𝐬𝐭 𝐰𝐞𝐞𝐤'𝐬 🎙𝙂𝙏𝙈 𝙎𝙘𝙞𝙚𝙣𝙘𝙚🎙 𝐞𝐩𝐢𝐬𝐨𝐝𝐞:
- Why the ROI question misses the point
- The real cost of broken GTM
- How small improvements compound
- What actually drives success

What has made GTM Ops an obvious investment for you?

🤔

Listen to the episode "The ROI of GTM Ops" on 🎙𝙂𝙏𝙈 𝙎𝙘𝙞𝙚𝙣𝙘𝙚🎙 here: https://bit.ly/4b28Nwn 

Get access to the below sheet in the Newsletter here: https://bit.ly/3M6IbQu     

✌️

Here's what no one wants to admit about GTM and RevOps

We're completely at the mercy of what's upstream.

1. Product Market Fit
 - We think of this as a startup thing
 - An inflection point around $1M in ARR
 - But it's a persistent problem at +$100M ARR
 - Because both products and markets are evolving
 - So GTM is left pushing inferior products to market
 - And expanding into new markets with poorer PMF

2. ICP/BP
 - By extension, ICP/BPs keep expanding
 - GTM sees diminishing returns by definition
 - The next expansion yields lower conversion
 - Making it harder and more expensive to close
 - And both ICP and BPs are evolving continuously

3. Channels
 - "Go where your customers hang out"
 - This is simple and highly effective advice
 - Picking the right channels makes all the difference

4. Messaging
 - It's not just Product Market Fit
 - It's Product Market Messaging Fit
 - It could eve be Product Market Channel Messaging Fit
 - Good messaging for cold email is VERY different from events

So, as CROs, CMOs, VPs of GTM Strategy and Operations.......

We're held accountable to drive results.

We're being told every day.... 
 - Go implement AI!
 - Hire more salespeople!
 - Improve marketing conversions!
 - Generate more pipeline and close more!

Often ignoring we're at the mercy of these upstream variables.

I'm not exactly complaining. 

This is the job. 

This is what we signed up for. 

But worth thinking about this more. 

✌️

I can't think of a bigger waste of time than CROs building AI agents. 

Can we just think about this through another lens. 

Imagine it's 2000. 
 - CRO manages GTM team of +100 people
 - Carves out 10 hrs per week to code
 - To build their own on-prem CRM
 - To build their own email tool
 - To build their own website

Would that make any sense?

Imagine it's 1970
 - CRO carves out 10 hrs per week
 - To learn the basics of IT
 - How to setup a mainframe
 - How to setup computers
 - How to wire telephone systems
 - IDK. Whatever IT did back then

Does this make any sense at all?

What did GTM leaders do back then?

1. Set the vision
2. Build the team
3. Outline the process
4. Drive execution
5. Hit targets

That's it!

In the 70s they had IT

In the 2000s they had Salesforce
 - And Gmail
 - And web designers
 - Etc. 

Why do I feel so strongly about this?

Because I did this!

I learned Salesforce
I learned to build websites
I learned Marketing Automation

It was a HUUUGE effing waste of time.

Because the second I met someone who actually knew what they were doing, they ran circles around me. 

So I hired them instead.

And focussed on my area of genius. 

That's what CROs should do with AI today. 

1. Set the vision for GTM
2. Build the right team
3. Outline the process
3B. Leverage the team to see where AI can help with the process
4. Drive execution (of humans + AI)
5. Hit targets

This requires.....
 - Picking very specific use cases for AI
 - In specific areas of the GTM process
 - Where the most opportunity lies
 - Knowing what good looks like
 - And driving the execution
 - To hit a specific target

That's it. 

There are plenty of crazy smart technologists that can build the thing. 

We've got some of those people on our team. 

And I am not going to try to be one of them!

Just like I'm not the Salesforce expert here!

✌️

Hype around AI GTM is at an all time high. 

I talk to ~5 VC/PE firms about this every single week. 
 - Firms you would know by name
 - People in those firms specialized in GTM
 - Mostly former CROs, CMOs, VPs of RevOps

Here's a secret:
They have no idea either. 

Everyone "has to implement AI"
No one knows how to see results from it. 

Here's what I'm hearing from top PE/VC firms:
 - "We need to implement AI in GTM"
 - But when I ask what's working right now...
 - They don't have many answers to this question

To be sure, there are exceptions. 
 - I'm posting more on them this week

But by and large
 - It's so much hype
 - So much "we have to do something"
 - And so little specifics
 - Like "this is the thing we want to do"

Why is that?

🤔

"What's the ROI of fixing our GTM Operations?"
I hear this question every week.

Right after they tell me everything is broken.

𝐇𝐞𝐫𝐞'𝐬 𝐰𝐡𝐚𝐭 𝐭𝐡𝐞𝐲 𝐣𝐮𝐬𝐭 𝐭𝐨𝐥𝐝 𝐦𝐞
- Forecasts are consistently off
- Inbound leads aren't getting followed up
- Pipeline is filled with deals that won't close
- NRR is slipping with no real process
- No one has visibility into anything

Then they ask if fixing it will pay off.
 🤯

𝐑𝐎𝐈 𝐢𝐬 𝐭𝐡𝐞 𝐰𝐫𝐨𝐧𝐠 𝐪𝐮𝐞𝐬𝐭𝐢𝐨𝐧
We can't measure ROI when metrics are broken

We could just look at ARR Growth
 - But that's taking too much credit

What we really WOULD need is conversions
 - How did our Close Rate improve? 
 - How did MQL conversions improve?
 - How did outbound prospecting improve?
 - How did NRR and GRR improve over the year?

𝗢𝐮𝐫 𝐩𝐢𝐩𝐞𝐥𝐢𝐧𝐞 𝐢𝐬 𝐚 𝐩𝐫𝐢𝐦𝐞 𝐞𝐱𝐚𝐦𝐩𝐥𝐞
- Some reps create opps for every meeting
- They have a 5% close rate
- Other reps wait until they get the verbal
- They have an 85% close rate
- Our blended rate is 25%

That 25% means absolutely nothing.

𝗪𝗲 𝐜𝐚𝐧'𝐭 𝐦𝐞𝐚𝐬𝐮𝐫𝐞 𝐑𝐎𝐈 𝐨𝐟 𝐟𝐢𝐱𝐢𝐧𝐠 𝐢𝐭
- Until we've actually fixed it
- Trained the team on the new process
- Waited months or quarters
- To see which qualified opps actually close

It's a chicken and egg problem.

𝐒𝐨 𝐡𝐞𝐫𝐞'𝐬 𝐚 𝐛𝐞𝐭𝐭𝐞𝐫 𝐚𝐩𝐩𝐫𝐨𝐚𝐜𝐡:

Look at the problem systemically.
- If you cleaned up the pipeline.
- Removed all unqualified opportunities
- Ensured reps only worked real opportunities.
- And also gave managers visibility and data to coach.

Do you believe that would result in more revenue?

Of course it would!

Is there really any doubt here?

𝐀𝐧𝐲 𝐢𝐦𝐩𝐫𝐨𝐯𝐞𝐦𝐞𝐧𝐭 𝐩𝐚𝐲𝐬 𝐟𝐨𝐫 𝐢𝐭𝐬𝐞𝐥𝐟.
- Imagine above work increases Close Rates just 1%
- From a Close Rate of 15% to a Close Rate of 16%
- On $100M of pipeline, that's $16M Closed Won
- That's an extra $1M of ARR just this year
- Plus $1M x Retention Rate next year
- We're not spending $1M on ops!

𝐌𝐮𝐥𝐭𝐢𝐩𝐥𝐲 𝐭𝐡𝐢𝐬 𝐚𝐜𝐫𝐨𝐬𝐬 𝐭𝐡𝐞 𝐟𝐮𝐧𝐧𝐞𝐥
- Fix inbound lead routing and response time
- Improve outbound territory design and focus
- Tighten sales handoff and onboarding
- Track customer health and manage renewals

Each improvement compounds with the others.
 - More qualified pipeline
 - Higher close rate
 - Better retention
 - More expansion

𝗕𝘂𝘁 𝘀𝘂𝗰𝗰𝗲𝘀𝘀 𝗿𝗲𝗾𝘂𝗶𝗿𝗲𝘀 𝗺𝗼𝗿𝗲 𝘁𝗵𝗮𝗻 𝗮 𝗳𝗶𝗻𝗮𝗻𝗰𝗶𝗮𝗹 𝗶𝗻𝘃𝗲𝘀𝘁𝗺𝗲𝗻𝘁
- It requires alignment
- Commitment from leadership
- Accountability across the GTM team

If everyone's committed to executing better?
These improvements are very achievable.

If not, they're impossible.

What's your take on the ROI of GTM Ops?

🤔

I went into this topic in much more depth in yesterday's 🎙𝙂𝙏𝙈 𝙎𝙘𝙞𝙚𝙣𝙘𝙚🎙Podcast.

Check comments for the link 👇

✌️

One of the biggest mistakes I see in GTM is trying to fix everything at once. 

When everything is a priority, nothing is a priority. 

𝐘𝐨𝐮 𝐜𝐨𝐦𝐞 𝐢𝐧 𝐨𝐧 𝐝𝐚𝐲 𝐨𝐧𝐞 𝐚𝐧𝐝 𝐬𝐞𝐞 𝐢𝐭 𝐚𝐥𝐥:
- Inbound leads aren't getting follow-up
- Pipeline management needs work
- Outbound isn't converting
- CS isn't tracking health
- The list goes on

So you try to show value quickly...

By fixing everything you see

Unfortunately, this is a recipe for failure

𝐇𝐞𝐫𝐞'𝐬 𝐰𝐡𝐚𝐭 𝐚𝐜𝐭𝐮𝐚𝐥𝐥𝐲 𝐰𝐨𝐫𝐤𝐬 𝐢𝐧 90 𝐝𝐚𝐲𝐬
- Pick 1-2 specific areas in GTM
- Focus on the fundamentals
- Design a repeatable process
- Drive the team to execute consistently

This is the best anyone can hope for
- It's also a monumental improvement to GTM

The first 90 days aren't about how much you build
- They're about proving you can drive real change
- You're not trying to optimize yet
- You're trying to get from broken to good enough

Because here's the reality
- You can build a pipeline report in 5 minutes
- But your goal is to get everyone to trust it
- You want to get there in 90 days

𝐈𝐧 𝐨𝐮𝐫 𝐥𝐚𝐬𝐭 📰𝙂𝙏𝙈 𝙎𝙘𝙞𝙚𝙣𝙘𝙚 𝘽𝙧𝙞𝙚𝙛 📰, 𝐰𝐞 𝐛𝐫𝐞𝐚𝐤 𝐝𝐨𝐰𝐧 𝐭𝐡𝐞 𝐩𝐥𝐚𝐲𝐛𝐨𝐨𝐤
- Step 1: Pick the one thing
- Step 2: Assess the fundamentals
- Step 3: Build a roadmap and get alignment
- Step 4: Execute with sprint planning

Why do you think it's so easy to fall into this trap?

🤔

More on this in our latest 📰𝙂𝙏𝙈 𝙎𝙘𝙞𝙚𝙣𝙘𝙚 𝘽𝙧𝙞𝙚𝙛 📰

We broke down exactly how to do this
 - How to pick the one thing
 - How to assess the fundamentals
 - How to build a roadmap and get alignment
 - And how to execute that roadmap and get traction

Link in comments below 👇

✌️

I spoke to a CMO in a +$100M B2B SaaS company the other day and she told me they have leads requesting a demo that never get a single response from sales!

How is this possible? 

One problem I see:
 - Mixing hand-raisers with score-based leads

Here's how this breaks down...

𝐌𝐚𝐫𝐤𝐞𝐭𝐢𝐧𝐠 𝐠𝐞𝐧𝐞𝐫𝐚𝐭𝐞𝐬 𝐡𝐚𝐧𝐝-𝐫𝐚𝐢𝐬𝐞𝐫 𝐥𝐞𝐚𝐝𝐬
- Someone fills out a demo request
- They want to talk to sales right now
- But they get routed with 9,000 other leads
- People who DON'T really want to talk to sales

𝐒𝐨 𝐡𝐚𝐧𝐝-𝐫𝐚𝐢𝐬𝐞𝐫𝐬 𝐝𝐨𝐧'𝐭 𝐠𝐞𝐭 𝐩𝐫𝐨𝐩𝐞𝐫 𝐟𝐨𝐥𝐥𝐨𝐰 𝐮𝐩
- They wait and wait and wait
- By the time an SDR calls them back
- They've already talked to your competitor

𝐘𝐨𝐮𝐫 𝐜𝐨𝐦𝐩𝐞𝐭𝐢𝐭𝐨𝐫 𝐫𝐞𝐬𝐩𝐨𝐧𝐝𝐞𝐝 𝐢𝐧 𝐣𝐮𝐬𝐭 5 𝐦𝐢𝐧𝐮𝐭𝐞𝐬
- They ran discovery and set the next meeting up
- Before you even responded to their first inquiry
- Industry studies show the first to respond wins
 - So you lose the deal before it even began

𝐌𝐞𝐚𝐧𝐰𝐡𝐢𝐥𝐞 𝐲𝐨𝐮'𝐫𝐞 𝐟𝐨𝐜𝐮𝐬𝐞𝐝 𝐨𝐧:
- Analyzing lead scores
- Optimizing campaigns 
- Building attribution models

But nobody's fixing the obvious problem
- Your hand-raisers are getting lost in the noise

𝐓𝐡𝐢𝐬 𝐢𝐬 𝐨𝐧𝐞 𝐨𝐟 𝐭𝐡𝐞 𝐭𝐰𝐨 𝐛𝐢𝐠𝐠𝐞𝐬𝐭 𝐰𝐚𝐲𝐬 𝐜𝐨𝐦𝐩𝐚𝐧𝐢𝐞𝐬 𝐰𝐚𝐬𝐭𝐞 𝐦𝐚𝐫𝐤𝐞𝐭𝐢𝐧𝐠 𝐛𝐮𝐝𝐠𝐞𝐭

The other?

𝐒𝐜𝐨𝐫𝐞-𝐛𝐚𝐬𝐞𝐝 𝐥𝐞𝐚𝐝𝐬 𝐭𝐡𝐚𝐭 𝐜𝐨𝐧𝐯𝐞𝐫𝐭 𝐥𝐞𝐬𝐬 𝐭𝐡𝐚𝐧 𝐨𝐮𝐭𝐛𝐨𝐮𝐧𝐝
- You're paying to generate those leads
- Then you're paying SDRs to chase them
- But those same SDRs could call cold prospects
- And those same SDRs could close way more deals

Why are we doing this?
- Most teams skip the basics
- Too busy chasing more demand
- Chasing more pipeline
- Chasing more leads
- Never asking if the process actually works

𝐈𝐧 𝐚 𝐫𝐞𝐜𝐞𝐧𝐭 🎙𝙂𝙏𝙈 𝙎𝙘𝙞𝙚𝙣𝙘𝙚🎙𝐞𝐩𝐢𝐬𝐨𝐝𝐞, 𝐰𝐞 𝐛𝐫𝐞𝐚𝐤 𝐝𝐨𝐰𝐧 𝐭𝐡𝐞 𝟗 𝐬𝐭𝐞𝐩𝐬
- To separate hand-raisers from scored leads
- To define your lead qual and routing process
- To track actual response time and follow-up
- To coach reps with data, creating a feedback loop

𝐓𝐡𝐢𝐬 𝐜𝐡𝐚𝐧𝐠𝐞 𝐡𝐞𝐥𝐩𝐞𝐝 𝐨𝐧𝐞 𝐨𝐟 𝐨𝐮𝐫 𝐜𝐮𝐬𝐭𝐨𝐦𝐞𝐫𝐬 𝐢𝐧𝐜𝐫𝐞𝐚𝐬𝐞 𝐥𝐞𝐚𝐝-𝐭𝐨-𝐜𝐥𝐨𝐬𝐞 𝐛𝐲 𝟐𝟓𝐱

25x increase!
- Not lead-to-pipeline.
- Not close rates on deals.
- Full lead-to-close by 25x.

If your lead conversion rate is terrible?
- The problem isn't just lead quality.
- It's operational.

Or am I missing something? 

🤔

Listen to the episode "The 9-Step Process for Improving Lead Conversion" on 🎙𝙂𝙏𝙈 𝙎𝙘𝙞𝙚𝙣𝙘𝙚🎙

See comment below for the link 👇

✌️

There's a lot of hype around AI in GTM right now.
But we've seen this pattern before.

At one time.....

𝗪𝗲 𝘁𝗵𝗼𝘂𝗴𝗵𝘁 𝗦𝗮𝗹𝗲𝘀𝗳𝗼𝗿𝗰𝗲 𝘄𝗮𝘀 𝗮 𝗺𝗮𝗴𝗶𝗰 𝗯𝘂𝗹𝗹𝗲𝘁
 - We left Excel or on-prem systems
 - We created one single source of truth
 - We had real-time visibility into everything
 - We expected to see immediate revenue growth

It didn't happen. 

Here's what did happen
 - GTM Team spend a lot of money
 - They spent even more time on it
 - And it was a massive improvement
 - But it reinforced existing GTM process
 - Some teams improved those processes
 - But many didn't change those processes
 - And many reps refused to use Salesforce
 - So they ended up doing all the same things
 - Not seeing any real improvement in revenue

Then a few years passed.

𝗪𝗲 𝘁𝗵𝗼𝘂𝗴𝗵𝘁 𝗢𝘂𝘁𝗿𝗲𝗮𝗰𝗵/𝗦𝗮𝗹𝗲𝘀𝗹𝗼𝗳𝘁 𝘄𝗮𝘀 𝗮 𝗺𝗮𝗴𝗶𝗰 𝗯𝘂𝗹𝗹𝗲𝘁
 - We all read Predictable Revenue
 - We built huge teams of hungry SDRs
 - We implemented Outreach/Salesloft
 - We automated our outbound sequences
 - And pressed "Send All" on email templates
 - Cranking activity to 10X what we we're doing
 - And we saw no improvement in actual pipeline
 - It was a race to the bottom of quality vs. quantity
  - At least for most teams. Some were the exception.
 - Some teams improved their process and saw results. 

𝐀𝐈 𝐢𝐬 𝐧𝐨 𝐝𝐢𝐟𝐟𝐞𝐫𝐞𝐧𝐭

Scale Venture Partners' recently released their State of GTM AI Report.

Craig led the year-long study
- 270+ survey responses
- 3-5 interviews per week
- AI labs and hackathons

We had him on the🎙𝙂𝙏𝙈 𝙎𝙘𝙞𝙚𝙣𝙘𝙚🎙Podcast to dig deeper.

𝐇𝐞𝐫𝐞'𝐬 𝐰𝐡𝐚𝐭 𝐚𝐜𝐭𝐮𝐚𝐥𝐥𝐲 𝐝𝐫𝐢𝐯𝐞𝐬 𝐀𝐈 𝐬𝐮𝐜𝐜𝐞𝐬𝐬
- Understand your current state first
- Figure out strategy, people, process
- Then decide where AI fits into process
- Know exactly what you want AI to produce
- Let RevOps lead the implementation of the AI
 
Successful teams don't say:
"Let's use AI."

They say:
"We'd like AI to create a QBR that's 4 pages long with this on page 1, this on page 2, pulling from these 6 sources."

They know what they want.
Then decide how AI helps.

𝐈𝐧 𝐲𝐞𝐬𝐭𝐞𝐫𝐝𝐚𝐲'𝐬 🎙𝙂𝙏𝙈 𝙎𝙘𝙞𝙚𝙣𝙘𝙚🎙𝗽𝗼𝗱𝗰𝗮𝘀𝘁
We dug into:

- Build vs buy (both work)
- Human in the loop requirements
- Why it's harder to do GTM than ever
- Phase 1 vs Phase 2 AI (3-5x difference)
- Why RevOps should lead (22% more impact)
- The "pacemaker" concept and taste in GTM
- Campaign analysis and micro-segmentation
- Why 85% say AI works but can't prove ROI

Listen to the episode "The State of GTM AI with Craig Rosenberg" on 🎙𝙂𝙏𝙈 𝙎𝙘𝙞𝙚𝙣𝙘𝙚🎙here: https://bit.ly/4b28Nwn

✌️

Pavilion built a microsegmentation engine.
Here's what worked and what didn't.

𝐓𝐡𝐞𝐲 𝐬𝐭𝐚𝐫𝐭𝐞𝐝 𝐰𝐢𝐭𝐡 𝐝𝐚𝐭𝐚, 𝐧𝐨𝐭 𝐠𝐮𝐞𝐬𝐬𝐞𝐬
- Member Success ran cohort analysis
- Who stayed, who contributed, who got value
- Regression on churners vs renewals
- Identified properties of highest LTV

𝐓𝐡𝐞𝐧 𝐭𝐡𝐞𝐲 𝐛𝐮𝐢𝐥𝐭 𝐭𝐡𝐞 𝐢𝐧𝐟𝐫𝐚𝐬𝐭𝐫𝐮𝐜𝐭𝐮𝐫𝐞
- Connected website visitor tracking
- Automated enrichment workflows
- Scored leads on fit criteria
- Generated personalized outreach
- Referenced similar existing members

This required tight collaboration.

𝐌𝐞𝐦𝐛𝐞𝐫 𝐒𝐮𝐜𝐜𝐞𝐬𝐬 𝐩𝐫𝐨𝐯𝐢𝐝𝐞𝐝 𝐜𝐫𝐢𝐭𝐞𝐫𝐢𝐚
- RevOps built infrastructure
- Content created assets
- Product Marketing crafted messaging

Someone has to own the vision.
 - At Pavilion, that was Josh Carter.
 - Without that champion?
 - It stays theoretical.

𝐁𝐮𝐭 𝐡𝐞𝐫𝐞'𝐬 𝐰𝐡𝐞𝐫𝐞 𝐢𝐭 𝐠𝐨𝐭 𝐭𝐫𝐢𝐜𝐤𝐲
- Technology can scrape everything
- Job postings, press releases, LinkedIn
- Just because you can, doesn't mean you should

𝐓𝐡𝐞 𝐥𝐢𝐧𝐞 𝐛𝐞𝐭𝐰𝐞𝐞𝐧 𝐡𝐞𝐥𝐩𝐟𝐮𝐥 𝐚𝐧𝐝 𝐜𝐫𝐞𝐞𝐩𝐲
- Reference something that changes priorities? 
- Helpful
- Reference something that goes too far? 
- Creepy

Focus on what prospects care about right now.
Everything else is noise.

𝐓𝐡𝐞𝐲 𝐭𝐞𝐬𝐭𝐞𝐝 𝐬𝐦𝐚𝐥𝐥
- Target 10, 50, or 100 people
- Test different motivators
- Learn what works
- Roll out to other populations

The cost of a miss is much lower.

𝐓𝐡𝐞 𝐫𝐞𝐬𝐮𝐥𝐭𝐬
- Open rates went up
- Response rates climbed
- Lower churn, higher LTV
- Conversions increased significantly

But the real win?

𝐁𝐫𝐢𝐧𝐠𝐢𝐧𝐠 𝐢𝐧 𝐭𝐡𝐞 𝐫𝐢𝐠𝐡𝐭 𝐦𝐞𝐦𝐛𝐞𝐫𝐬
- People who stayed longer
- Contributed more
- Made the community stronger

Volume metrics look good in reviews.
But wrong prospects create downstream problems.

𝐈𝐧 𝐲𝐞𝐬𝐭𝐞𝐫𝐝𝐚𝐲'𝐬 📰𝙏𝙝𝙚 𝙂𝙏𝙈 𝙎𝙘𝙞𝙚𝙣𝙘𝙚 𝘽𝙧𝙞𝙚𝙛 📰 we covered:
- The operational microsegmentation playbook
- How to build the infrastructure
- How to test and scale
- What this means for CROs, CMOs, RevOps

What segmentation play are you seeing win the most these days?

🤔

More on this in tomorrow’s 📰𝙏𝙝𝙚 𝙂𝙏𝙈 𝙎𝙘𝙞𝙚𝙣𝙘𝙚 𝘽𝙧𝙞𝙚𝙛📰
Check comments for the link 👇

✌️

There's a better way to segment in marketing. 
Not by demographics but by situations. 

Not "CMO at Series B SaaS"
But "CMO preparing for first board meeting"

Kathleen Booth built this at Pavilion.
Conversions increased significantly.

Here's how they did it:

𝐒𝐭𝐚𝐫𝐭 𝐰𝐢𝐭𝐡 𝐰𝐡𝐨 𝐚𝐜𝐭𝐮𝐚𝐥𝐥𝐲 𝐰𝐢𝐧𝐬
- Run cohort analysis on best customers
- Who stayed, contributed, got value
- Regression on churners vs renewals
- Only then build your segments

𝐅𝐢𝐧𝐝 𝐭𝐡𝐞 𝐦𝐨𝐦𝐞𝐧𝐭𝐬 𝐭𝐡𝐚𝐭 𝐦𝐚𝐭𝐭𝐞𝐫
- Just raised funding
- Got a new job
- Lost a job
- Preparing for board meeting

𝐁𝐮𝐢𝐥𝐝 𝐜𝐨𝐧𝐭𝐞𝐧𝐭 𝐟𝐨𝐫 𝐭𝐡𝐨𝐬𝐞 𝐦𝐨𝐦𝐞𝐧𝐭𝐬
- Board prep guides with sample decks
- First 90 days checklists
- Main content ungated
- Practical tools required email

𝐁𝐮𝐢𝐥𝐝 𝐭𝐡𝐞 𝐞𝐧𝐠𝐢𝐧𝐞 𝐟𝐨𝐫 𝐬𝐜𝐚𝐥𝐞
- De-anonymize website visitors
- Clay enriches automatically
- Scores and customizes outreach
- Copy AI generates messaging
- References similar happy customers

𝐅𝐨𝐜𝐮𝐬 𝐨𝐧 𝐰𝐡𝐚𝐭 𝐭𝐡𝐞𝐲 𝐜𝐚𝐫𝐞 𝐚𝐛𝐨𝐮𝐭 𝐧𝐨𝐰
- The pain they're feeling
- The goals they're chasing
- Challenges specific to their moment

𝐓𝐞𝐬𝐭 𝐬𝐦𝐚𝐥𝐥, 𝐥𝐞𝐚𝐫𝐧 𝐟𝐚𝐬𝐭
- Target smaller populations
- Maybe 10, maybe 50, maybe 100
- Test different motivators
- Roll out what works

𝐓𝐡𝐞 𝐫𝐞𝐬𝐮𝐥𝐭𝐬
- Open rates went up
- Conversions increased significantly
- Response rates climbed
- Churn decreased
- Lifetime values increased

They brought in members who stayed longer.
Contributed more. Made the community stronger.

How often do you receive outbound this specific?

🤔

More on this in tomorrow’s 📰𝙏𝙝𝙚 𝙂𝙏𝙈 𝙎𝙘𝙞𝙚𝙣𝙘𝙚 𝘽𝙧𝙞𝙚𝙛 📰

Check comments for the link 👇

✌️

You're 5x more likely to improve conversion using AI for campaign analysis and scoring.

According to Scale Venture Partners' "State of AI Report"

We interviewed Craig Rosenberg:
- To dig deeper on the report findings
- To see what winning teams did differently
- And why other teams couldn't measure AI ROI

Tons of gold in that interview
Yesterday's newsletter focused on one bit:

𝐓𝐞𝐚𝐦𝐬 𝐮𝐬𝐞 𝐀𝐈 𝐢𝐧 𝐭𝐰𝐨 𝐝𝐢𝐬𝐭𝐢𝐧𝐜𝐭 𝐩𝐡𝐚𝐬𝐞𝐬
- Phase one: speed and productivity
- Phase two: revenue impact

The teams seeing 3-5x better results?
They're in phase two

𝐖𝐡𝐚𝐭 𝐩𝐡𝐚𝐬𝐞 𝐭𝐰𝐨 𝐭𝐞𝐚𝐦𝐬 𝐝𝐨 𝘁𝗼 𝘀𝗲𝗲 𝗿𝗲𝘀𝘂𝗹𝘁𝘀:

𝗧𝗵𝗲𝘆 𝗮𝗻𝗮𝗹𝘆𝘇𝗲 𝗯𝗲𝗳𝗼𝗿𝗲 𝘀𝗰𝗮𝗹𝗶𝗻𝗴
- Which campaigns worked?
- What touchpoints converted?
- What messaging resonated?
- What timing mattered?

𝗧𝗵𝗲𝘆 𝘁𝗮𝗿𝗴𝗲𝘁 𝗯𝗲𝘁𝘁𝗲𝗿
- Use AI to score accounts
- Build tighter segments
- Sometimes 100 vs thousands

Not just using AI to work faster
Using it to drive better outcomes
𝗧𝗼 𝘀𝗼𝗹𝘃𝗲 𝗳𝘂𝗻𝗱𝗮𝗺𝗲𝗻𝘁𝗮𝗹𝗹𝘆 𝗱𝗶𝗳𝗳𝗲𝗿𝗲𝗻𝘁 𝗽𝗿𝗼𝗯𝗹𝗲𝗺𝘀

Where are you seeing AI have the most impact on revenue growth?

🤔

More on this in 📰𝙏𝙝𝙚 𝙂𝙏𝙈 𝙎𝙘𝙞𝙚𝙣𝙘𝙚 𝘽𝙧𝙞𝙚𝙛📰
Read it here: https://bit.ly/45v38vv

You can also listen to the episode "The State of GTM AI with Craig Rosenberg" on 🎙𝙂𝙏𝙈 𝙎𝙘𝙞𝙚𝙣𝙘𝙚🎙here: https://bit.ly/4b28Nwn

✌️

People talk about ICP and PMF like it's a one-time task. 
 - We're in our 10th year in business still iterating on it. 

We're still asking "Who's our ICP?"

"Remember Acme, the customer we won 5 years ago, retained for 3 years, made good profit on, the one we have a solid customer story.......?"

"Do we want more customers like that?"

"Does this inhibit our abilities to better serve other types of customers?"

"Will we burn cycles improving our product and service to accomodate their needs that are different from the needs of other types of customers?"

"Will we waste energy in sales and marketing trying to build content, channels, sales motions, etc. to attract these companies at the cost of attracting the customers we think we can help most?"

"When someone comes to us offering us money, that is explicitly NOT in our ICP, that we have literally said right on our website is not our ICP, do we take their money and try our best to serve them?"

This defines everything we do, from how we improve our product and service, to how to attract, retain, and grow customers. 

It get's as granular as.....
 - What content will we produce this week?
 - How will we improve our pitch deck and brochures?
 - When a prospect asks for something odd, do we go fetch or just say no?

These are questions we're still asking, 10 years in, and I'm surprised so many companies think about this once and then just move on. 

Why is that?

🤔

Two thirds of B2B teams are using AI in GTM.
Most can't prove it's working (impacting revenue)

Scale Venture Partners wanted to know why.
- Craig Rosenberg spent a year on it
- Surveyed hundreds of teams
- Ran AI labs and hackathons

Here's what their data revealed:

Teams use AI in two distinct phases.

𝗣𝗵𝗮𝘀𝗲 𝗼𝗻𝗲 𝘁𝗲𝗮𝗺𝘀 𝘂𝘀𝗲 𝗔𝗜 𝗳𝗼𝗿 𝘀𝗽𝗲𝗲𝗱:
- Draft emails faster
- Research accounts quicker
- Save time on existing activities

This is just doing the same things faster.
It doesn't directly impact revenue production.

𝗣𝗵𝗮𝘀𝗲 𝘁𝘄𝗼 𝘁𝗲𝗮𝗺𝘀 𝗱𝗿𝗶𝘃𝗲 𝗯𝘂𝘀𝗶𝗻𝗲𝘀𝘀 𝗼𝘂𝘁𝗰𝗼𝗺𝗲𝘀:
- Account scoring
- Campaign analysis
- Micro-segmentation
- Skills coaching for reps

This is doing things better.
This is directly impacting revenue.

𝐏𝐡𝐚𝐬𝐞 𝐭𝐰𝐨 𝐭𝐞𝐚𝐦𝐬 𝘀𝗮𝘄 𝐛𝐞𝐭𝐭𝐞𝐫 𝐫𝐞𝐬𝐮𝐥𝐭𝐬
- 3x more likely to see pipeline growth
- 3x more likely to see faster velocity
- 5x more likely to see conversions

That's from the actual report data.

We had some questions for Craig about this
- So we interviewed him on the🎙𝙂𝙏𝙈 𝙎𝙘𝙞𝙚𝙣𝙘𝙚🎙Podcast.
- Then distilled the key details in our newsletter

𝐈𝐧 𝘁𝗼𝗺𝗼𝗿𝗿𝗼𝘄'𝐬 📰 𝙂𝙏𝙈 𝙎𝙘𝙞𝙚𝙣𝙘𝙚 𝘽𝙧𝙞𝙚𝙛 📰 ....

𝗪𝗲 𝗯𝗿𝗲𝗮𝗸 𝗱𝗼𝘄𝗻:
- Why phase 1 AI implementation can't prove ROI
- What phase 2 teams actually do operationally
- Who should be leading these AI initiatives
- What this means for revenue leaders

What are you seeing in your work on GTM AI?

🤔

More in tomorrow’s 📰 𝙂𝙏𝙈 𝙎𝙘𝙞𝙚𝙣𝙘𝙚 𝘽𝙧𝙞𝙚𝙛 📰
Subscribe to get it here: https://bit.ly/49RCm0h 

You can also listen to the episode "𝗧𝗵𝗲 𝗦𝘁𝗮𝘁𝗲 𝗼𝗳 𝗚𝗧𝗠 𝗔𝗜 𝘄𝗶𝘁𝗵 𝗖𝗿𝗮𝗶𝗴 𝗥𝗼𝘀𝗲𝗻𝗯𝗲𝗿𝗴" on the🎙𝙂𝙏𝙈 𝙎𝙘𝙞𝙚𝙣𝙘𝙚🎙Podcast here: https://bit.ly/4b28Nwn

✌️

How can we use AI to "personalize at scale" while staying human?

We interviewed Kathleen Booth on the🎙𝙂𝙏𝙈 𝙎𝙘𝙞𝙚𝙣𝙘𝙚🎙Podcast and she shared some great tips. 

𝐇𝐞𝐫 𝐭𝐞𝐚𝐦 𝐚𝐭 𝐏𝐚𝐯𝐢𝐥𝐢𝐨𝐧:
- Used AI to scrape press releases
- Found companies’ budget plans
- Referenced it in outreach

Not like "Congrats on your recent funding."
- Everyone's doing that

𝐈𝐧𝐬𝐭𝐞𝐚𝐝, 𝐭𝐡𝐞𝐲 𝐰𝐫𝐨𝐭𝐞 𝐭𝐡𝐢𝐬:
- "I see you're using the funding to hire engineering talent."
- "Did you know Company X did the same thing?"

That's personalization.

But here's what most people miss...

𝐘𝐨𝐮 𝐜𝐚𝐧 𝐝𝐨 𝐭𝐡𝐢𝐬 𝐚𝐮𝐭𝐨𝐦𝐚𝐭𝐢𝐜𝐚𝐥𝐥𝐲 𝐧𝐨𝐰
- Tools like Clay scrape websites
- They Identify similar companies
- Reference your happy customers
- Analyze job postings
- Read press releases

𝐓𝐡𝐞𝐧 𝐀𝐈 𝐰𝐫𝐢𝐭𝐞𝐬 𝐭𝐡𝐞 𝐞𝐦𝐚𝐢𝐥
- Tailored to that specific person
- Based on everything it knows
- You just review it
- Add a personal note
- Send it

This used to take hours per email
- Now it takes minutes

𝐁𝐮𝐭 𝐭𝐡𝐞𝐫𝐞'𝐬 𝐚 𝐥𝐢𝐧𝐞 𝐛𝐞𝐭𝐰𝐞𝐞𝐧 𝐡𝐞𝐥𝐩𝐟𝐮𝐥 𝐚𝐧𝐝 𝐜𝐫𝐞𝐞𝐩𝐲
- Just because you can
- Doesn't mean you should

The art isn't in the tools

𝐈𝐭'𝐬 𝐢𝐧 𝐭𝐡𝐞 𝐣𝐮𝐝𝐠𝐦𝐞𝐧𝐭
- What do you reference?
- What do you leave alone?
- How do you stay human?
- How do you keep it relevant?

𝐈𝐧 𝐨𝐮𝐫 𝐥𝐚𝐭𝐞𝐬𝐭 𝐞𝐩𝐢𝐬𝐨𝐝𝐞, 𝐊𝐚𝐭𝐡𝐥𝐞𝐞𝐧 𝐞𝐱𝐩𝐥𝐚𝐢𝐧𝐬:
- Where to draw the line
- How to test without being creepy
- Why smaller segments let you fail safely
- How to build prompts that feel human

This is just one part in our broader topic:
𝐁𝐮𝐢𝐥𝐝𝐢𝐧𝐠 𝐚 𝐌𝐢𝐜𝐫𝐨𝐬𝐞𝐠𝐦𝐞𝐧𝐭𝐚𝐭𝐢𝐨𝐧 𝐄𝐧𝐠𝐢𝐧𝐞

The tools are available to everyone.
The edge is how you use them.

Does your inbox have any "AI personalization" horror stories?

🤔

More on this in the🎙𝙂𝙏𝙈 𝙎𝙘𝙞𝙚𝙣𝙘𝙚🎙podcast. 

Link in comments below. 👇

✌️

Building perfect metrics can take months. 
But we need visibility today. What do we do?

Accurate pipeline measurement requires us to:
- Define our ICP and buyer personas
- Document qualification criteria
- Build our sales process
- Implement it all into Salesforce
- Build reporting and dashboards
- Train our entire team
- Set up recurring pipeline reviews
- Remind reps constantly to follow it

That's months of work.
Just to answer "how much pipeline did we generate?"

𝐀𝐧𝐝 𝐭𝐡𝐞𝐧 𝐰𝐞 𝐡𝐚𝐯𝐞 𝐭𝐨 𝐰𝐚𝐢𝐭
- For at least one full sales cycle
- To see real Close Rate, Sales Cycle
- To see if our process actually works
- Then iterate when it doesn't
- Then wait another full cycle

For enterprise deals?
That could be years.

𝐌𝐞𝐚𝐧𝐰𝐡𝐢𝐥𝐞, 𝐰𝐞'𝐫𝐞 𝐟𝐥𝐲𝐢𝐧𝐠 𝐛𝐥𝐢𝐧𝐝
- Can't tell the board how we're performing
- Can't make decisions about what to fix
- Can't allocate resources effectively
- Can't see what's working and what's not

So what do we do?
𝐀𝐬𝐤 𝐚 𝐝𝐢𝐟𝐟𝐞𝐫𝐞𝐧𝐭 𝐪𝐮𝐞𝐬𝐭𝐢𝐨𝐧 

Not "what metrics should we have?"

"𝐖𝐡𝐚𝐭 𝐜𝐚𝐧 𝐰𝐞 𝐦𝐞𝐚𝐬𝐮𝐫𝐞 𝐭𝐨𝐝𝐚𝐲?"
- Demo requests from our website
- Demos actually delivered
- Form fills on website
 - Sales emails sent
- Meetings booked
- Closed won deals

Not perfect.
But available right now.

𝐓𝐡𝐢𝐬 𝐢𝐬 𝐭𝐡𝐞 𝐪𝐮𝐢𝐜𝐤 𝐚𝐧𝐝 𝐝𝐢𝐫𝐭𝐲
- Get visibility today with imperfect metrics
- Make decisions based on real data
- Build the perfect foundation tomorrow
- Stop flying blind while we wait

We did this with a +$100M B2B SaaS client.

𝐓𝐡𝐞𝐲 𝐜𝐨𝐮𝐥𝐝𝐧'𝐭 𝐦𝐞𝐚𝐬𝐮𝐫𝐞 𝐚𝐧𝐲𝐭𝐡𝐢𝐧𝐠
- No visibility into leads or opps
- Their PE firm needed answers fast
- We couldn't wait six months for perfect
- So we measured what was available today
- While building the right foundation in tandem

𝗪𝗵𝗮𝘁'𝘀 𝘁𝗵𝗲 𝗺𝗼𝘀𝘁 𝘂𝘀𝗲𝗳𝘂𝗹 "𝗾𝘂𝗶𝗰𝗸 𝗮𝗻𝗱 𝗱𝗶𝗿𝘁𝘆" 𝗺𝗲𝘁𝗿𝗶𝗰 𝘆𝗼𝘂'𝘃𝗲 𝘂𝘀𝗲𝗱?

🤔

We dove much deeper into how to do this on the 🎙𝙂𝙏𝙈 𝙎𝙘𝙞𝙚𝙣𝙘𝙚🎙 Podcast. 

Link to the Podcast in my comment below 👇

✌️

What do we do if we have bigger revenue targets in marketing this year and less resources?

𝐀𝐬𝐬𝐮𝐦𝐞 𝐰𝐞 𝐡𝐚𝐯𝐞 𝐭𝐡𝐞 𝐟𝐮𝐧𝐝𝐚𝐦𝐞𝐧𝐭𝐚𝐥𝐬 𝐢𝐧 𝐩𝐥𝐚𝐜𝐞
- Sales team is following up consistently
- Conversion rates are decent
- Leads are flowing in

𝐖𝐡𝐞𝐫𝐞 𝐝𝐨 𝐰𝐞 𝐠𝐨 𝐟𝐫𝐨𝐦 𝐡𝐞𝐫𝐞?

𝐈𝐧𝐜𝐫𝐞𝐚𝐬𝐞 𝐭𝐡𝐞 𝐦𝐚𝐫𝐤𝐞𝐭𝐢𝐧𝐠 𝐛𝐮𝐝𝐠𝐞𝐭?
- More ad spend?
- More headcount?
- Increase production?
- Launch more channels?

What if we don’t have the budget?
Or what if that’s just not enough?

𝗪𝗲 𝗰𝗮𝗻 𝗮𝗹𝘀𝗼 𝗶𝗺𝗽𝗿𝗼𝘃𝗲 𝗲𝗳𝗳𝗶𝗰𝗶𝗲𝗻𝗰𝘆
- Some lead segments convert at 3x others
- We're giving them all the same resources
- We've never calculated lead value by segment
- Or which channels have best CAC payback

How do good GTM engines become exceptional?

𝐁𝐲 𝐚𝐬𝐤𝐢𝐧𝐠 𝐭𝐡𝐞 𝐝𝐞𝐞𝐩𝐞𝐫 𝐪𝐮𝐞𝐬𝐭𝐢𝐨𝐧𝐬:
- Which segments actually drive revenue?
- Where are we wasting sales effort?
- Are enterprise leads getting SMB playbooks?
- Should we reallocate spend between channels?

𝐈𝐧𝐬𝐭𝐞𝐚𝐝, 𝐜𝐨𝐦𝐩𝐚𝐧𝐢𝐞𝐬 𝐤𝐞𝐞𝐩 𝐠𝐞𝐭𝐭𝐢𝐧𝐠 𝐭𝐡𝐢𝐬 𝐰𝐫𝐨𝐧𝐠:

Enterprise leads get SMB follow-up
- Six-figure opps get 3 automated emails
- When they need executive engagement
- And multi-threading across buyers
- That's not a lead quality problem
- That's a process problem

Some segments are worth 10x others
- But we're spending the same to acquire them
- Following up with the same intensity
- Giving them the same resources
- When CAC payback varies wildly

How do we figure this all out?

𝐌𝐚𝐬𝐭𝐞𝐫 𝐞𝐱𝐞𝐜𝐮𝐭𝐢𝐨𝐧 𝐛𝐞𝐟𝐨𝐫𝐞 𝐚𝐭𝐭𝐫𝐢𝐛𝐮𝐭𝐢𝐨𝐧
- Perfect attribution can wait
- Perfect lead follow up is better
- It tells us which leads are truly valuable
- Then we can track where leads come from
- A simple "how did you hear about us" works
- To see what's really driving revenue and not

But the biggest challenge isn't the analysis.

It's getting the team aligned on priorities

𝐖𝐢𝐭𝐡𝐨𝐮𝐭 𝐂𝐑𝐎 𝐛𝐮𝐲-𝐢𝐧, 𝐧𝐨𝐭𝐡𝐢𝐧𝐠 𝐡𝐚𝐩𝐩𝐞𝐧𝐬
- RevOps analyzes data
- Makes recommendations
- Then insights sit in a deck somewhere
- Instead of being actioned by the GTM team


𝗧𝗵𝗶𝘀 𝗶𝘀 𝘄𝗵𝗲𝗿𝗲 𝗮 𝗣𝗶𝗽𝗲𝗹𝗶𝗻𝗲 𝗖𝗼𝘂𝗻𝗰𝗶𝗹 𝗰𝗮𝗻 𝗵𝗲𝗹𝗽
- RevOps presents analysis with recommendations
- Leadership decides where to cut, improve, invest
- Turns insights into concrete initiatives
- With owners, timelines, and success metrics


𝐈𝐧 𝐭𝐨𝐦𝐨𝐫𝐫𝐨𝐰'𝐬 📰𝙏𝙝𝙚 𝙂𝙏𝙈 𝙎𝙘𝙞𝙚𝙣𝙘𝙚 𝘽𝙧𝙞𝙚𝙛 📰 𝐰𝐞 𝐭𝐚𝐥𝐤 𝐚𝐛𝐨𝐮𝐭 𝐡𝐨𝐰 𝐭𝐨:
- Compare channel performance objectively
- Calculate lead value by segment
- Fix process vs lead quality issues
- Run Pipeline Councils that drive action
- Turn analysis into revenue impact

𝐄𝐯𝐞𝐫𝐲 𝐦𝐨𝐧𝐭𝐡 𝐰𝐞 𝐝𝐞𝐥𝐚𝐲 𝐢𝐬 𝐚 𝐦𝐨𝐧𝐭𝐡 𝐥𝐨𝐬𝐭

The foundation gets us in the game.

Optimization is how we win it.

✌️

More on this in tomorrow’s 📰𝙏𝙝𝙚 𝙂𝙏𝙈 𝙎𝙘𝙞𝙚𝙣𝙘𝙚 𝘽𝙧𝙞𝙚𝙛📰
Subscribe to get it here: https://bit.ly/49RCm0h

✌️

A while back we took on a +$100M B2B SaaS co that couldn't even tell you how many leads or opportunities they had. 

Their systems and processes were so broken they had zero visibility.

Reports were junk in/junk out. 

ELT, Board, Investors were flying blind!

So, first, this blew my mind!

They blew past $100M in ARR despite not having any GTM Ops in place. 
 - Amazing product
 - Even better Product Market Fit
 - Sheer hustle and drive to win their market
 - But this was obviously unsustainable.

So.........

𝐓𝐡𝐞 𝐏𝐄 𝐟𝐢𝐫𝐦 𝐜𝐚𝐥𝐥𝐞𝐝 𝐮𝐬
- Couldn't wrap their head around the numbers
- Couldn't get visibility into what's going on
- Couldn't wait six months for a diagnostic
- Wanted to see an immediate solution
- In tandem w/ a long-term foundation

So here's what we did.

𝐖𝐞 𝐚𝐬𝐤𝐞𝐝: 𝐖𝐡𝐚𝐭 𝐜𝐚𝐧 𝐰𝐞 𝐦𝐞𝐚𝐬𝐮𝐫𝐞 𝐭𝐨𝐝𝐚𝐲?
- Number of demo requests on the website
- Number of demos actually delivered
- Form fills, emails sent and received
- Meetings booked and meetings held
- Deals closed won

Not perfect metrics.
But better than waiting six months.

𝐖𝐞 𝐜𝐚𝐥𝐥𝐞𝐝 𝐝𝐞𝐦𝐨 𝐫𝐞𝐪𝐮𝐞𝐬𝐭𝐬 "𝐌𝐐𝐋𝐬"
- And demos delivered "SQOs"
- A quick and dirty solution
- While we built the real foundation

Because here's the reality:

𝐓𝐨 𝐦𝐞𝐚𝐬𝐮𝐫𝐞 𝐩𝐢𝐩𝐞𝐥𝐢𝐧𝐞 𝐚𝐜𝐜𝐮𝐫𝐚𝐭𝐞𝐥𝐲, 𝐲𝐨𝐮 𝐧𝐞𝐞𝐝:
- Clear ICP and buyer personas
- Defined qualification criteria
- Strong sales process documented
- Process implemented into Salesforce
- Reporting and dashboards built
- Team trained on the process
- Recurring pipeline review process
- Management reinforcing it constantly

That's a tremendous amount of work.

Just to answer "how much pipeline are we generating?"

𝐀𝐧𝐝 𝐢𝐭 𝐜𝐚𝐧 𝐭𝐚𝐤𝐞 𝐦𝐨𝐧𝐭𝐡𝐬 𝐨𝐫 𝐲𝐞𝐚𝐫𝐬
- Depending on your sales cycle
- Especially for enterprise deals
- You have to wait to see if it works
- Then iterate if it doesn't

This is why I've learned to ask:

𝐖𝐡𝐚𝐭'𝐬 𝐭𝐡𝐞 𝐪𝐮𝐢𝐜𝐤 𝐚𝐧𝐝 𝐝𝐢𝐫𝐭𝐲?
- What can we measure today?
- While building the foundation for tomorrow?
- Get visibility now with imperfect metrics
- Build sustainable measurement for the future

𝐈𝐧 𝐲𝐞𝐬𝐭𝐞𝐫𝐝𝐚𝐲'𝐬 🎙𝙂𝙏𝙈 𝙎𝙘𝙞𝙚𝙣𝙘𝙚🎙 𝐞𝐩𝐢𝐬𝐨𝐝𝐞, 𝐰𝐞 𝐭𝐚𝐥𝐤𝐞𝐝 𝐚𝐛𝐨𝐮𝐭:
- How to get visibility into metrics fast
- What you can measure today vs tomorrow
- How to use the GTM Decision Tree to prioritize
- Why this is a never-ending journey

How would you have approached this?

🤔

More on this in the🎙𝙂𝙏𝙈 𝙎𝙘𝙞𝙚𝙣𝙘𝙚🎙podcast.

Check comments for the link  👇

✌️

Could you eliminate 2/3rds of your MQLs and see more revenue?

Scott Sutton did exactly this at ZoomInfo
 - He shared the story on our Podcast

We wanted to take this a step further
 - So we broke down the steps in our Newsletter
 - What we've seen working with out clients

Here's how you optimize marketing revenue:

𝐒𝐭𝐞𝐩 𝟏: 𝐒𝐞𝐭 𝐭𝐡𝐞 𝐩𝐞𝐫𝐟𝐨𝐫𝐦𝐚𝐧𝐜𝐞 𝐛𝐚𝐫
- Track activity-based returns
- Calculate CAC payback by channel
- Compare inbound vs outbound conversion
- If outbound consistently beats inbound?
- We have a problem

Why would any rep or company call inbound leads if they could just run outbound and generate more revenue?

𝐒𝐭𝐞𝐩 𝟐: 𝐒𝐞𝐠𝐦𝐞𝐧𝐭 𝐚𝐧𝐝 𝐜𝐚𝐥𝐜𝐮𝐥𝐚𝐭𝐞 𝐥𝐞𝐚𝐝 𝐯𝐚𝐥𝐮𝐞
- Break down by company size, product, geo. etc
- Calculate actual value per lead by segment
- Enterprise leads might be worth $4K each
- SMB leads might be worth $2k
- That changes everything about follow-up

𝐒𝐭𝐞𝐩 𝟑: 𝐑𝐞𝐯𝐢𝐬𝐢𝐭 𝐥𝐞𝐚𝐝 𝐬𝐜𝐨𝐫𝐢𝐧𝐠
- SMB leads with score of 100 might convert well
- But enterprise leads with same score don't
- Maybe we're scoring white paper downloads
- In enterprise, that's junior people
- Who rarely have authority to buy

𝐒𝐭𝐞𝐩 𝟒: 𝐃𝐢𝐚𝐠𝐧𝐨𝐬𝐞 𝐩𝐫𝐨𝐜𝐞𝐬𝐬 𝐢𝐬𝐬𝐮𝐞𝐬
- Are enterprise leads getting SMB playbooks?
- Do we need different messaging by segment?
- Should we multi-thread for certain deals?
- Is our follow-up adequate for segment value?

𝐒𝐭𝐞𝐩 𝟓: 𝐑𝐮𝐧 𝐚 𝐏𝐢𝐩𝐞𝐥𝐢𝐧𝐞 𝐂𝐨𝐮𝐧𝐜𝐢𝐥
- RevOps presents analysis with recommendations
- Leadership decides where to cut, improve, invest
- Turns insights into concrete initiatives
- With owners, timelines, success metrics

𝐒𝐭𝐞𝐩 𝟔: 𝐓𝐮𝐫𝐧 𝐚𝐧𝐚𝐥𝐲𝐬𝐢𝐬 𝐢𝐧𝐭𝐨 𝐚𝐜𝐭𝐢𝐨𝐧
- Modified follow-up for high-value segments
- Budget reallocation from low to high performers
- Messaging changes for each segment
- Increased spend in top-performing channels
- Cut segments that are dead weight

𝐇𝐞𝐫𝐞'𝐬 𝐰𝐡𝐚𝐭 𝐦𝐨𝐬𝐭 𝐜𝐨𝐦𝐩𝐚𝐧𝐢𝐞𝐬 𝐠𝐞𝐭 𝐰𝐫𝐨𝐧𝐠:
- RevOps does all the analysis
- Makes the recommendations
- Presents to leadership
- Then nothing happens.

𝐖𝐢𝐭𝐡𝐨𝐮𝐭 𝐂𝐑𝐎 𝐛𝐮𝐲-𝐢𝐧 𝐚𝐧𝐝 𝐟𝐨𝐥𝐥𝐨𝐰-𝐭𝐡𝐫𝐨𝐮𝐠𝐡:
- Sales managers don't change processes
- Marketing doesn't reallocate budget
- Insights sit in a deck
- The inefficient engine keeps running

In yesterday's 📰𝙏𝙝𝙚 𝙂𝙏𝙈 𝙎𝙘𝙞𝙚𝙣𝙘𝙚 𝘽𝙧𝙞𝙚𝙛 📰 we broke down:
- Each step in detail
- How to calculate lead value
- When attribution actually matters
- How to run Pipeline Councils that execute
- How to turn insights into revenue

Which step do you think most companies skip?

🤔

More on this in yesterday’s 📰𝙏𝙝𝙚 𝙂𝙏𝙈 𝙎𝙘𝙞𝙚𝙣𝙘𝙚 𝘽𝙧𝙞𝙚𝙛 📰
Get it here:  https://bit.ly/4sn0FwQ

✌️

It's harder to do GTM today than ever.
Even veterans with proven playbooks struggle.

Craig Rosenberg built Scale Venture Partners' State of GTM AI Report to see AI's impact.

We had him on 🎙𝙂𝙏𝙈 𝙎𝙘𝙞𝙚𝙣𝙘𝙚🎙to discuss it.

𝐇𝐞𝐫𝐞'𝐬 𝐰𝐡𝐲 𝐢𝐭'𝐬 𝐡𝐚𝐫𝐝𝐞𝐫 𝐧𝐨𝐰
- Emails don't get read
- SEO is dying
- More crowded
- More competitive

Old tactics that worked for years:
- Outbound SDR emails
- High-pressure sales cycles
Marketing digital/SEO bringing 40-60% of pipeline

They still work. But not like they used to.

𝐀𝐈 𝐢𝐬𝐧'𝐭 𝐚 𝐦𝐚𝐠𝐢𝐜 𝐟𝐢𝐱
- But it does let you rethink everything
- Try different approaches
- Solve problems you couldn't solve before

𝐒𝐨𝐦𝐞 𝐜𝐨𝐦𝐩𝐚𝐧𝐢𝐞𝐬 𝐚𝐫𝐞 𝐮𝐬𝐢𝐧𝐠 𝐢𝐭 𝐝𝐢𝐟𝐟𝐞𝐫𝐞𝐧𝐭𝐥𝐲
- Analyzing campaigns before scaling
- Scoring accounts with multiple signals
- Building micro-segments of 100 vs thousands
- Creating automated pipeline reviews

𝗧𝗵𝗶𝘀 𝗶𝘀 𝘂𝘀𝗶𝗻𝗴 𝗔𝗜 𝘁𝗼 𝗰𝗿𝗲𝗮𝘁𝗲 𝗯𝘂𝘀𝗶𝗻𝗲𝘀𝘀 𝗼𝘂𝘁𝗰𝗼𝗺𝗲𝘀.
Not just to work faster.

What did the report say about these companies?

They're 3-5x more likely to see ROI improvement.

𝗖𝗼𝘃𝗲𝗿𝗲𝗱 𝗶𝐧 𝗹𝗮𝘀𝘁 𝘄𝗲𝗲𝗸'𝐬 🎙𝙂𝙏𝙈 𝙎𝙘𝙞𝙚𝙣𝙘𝙚🎙𝗲𝗽𝗶𝘀𝗼𝗱𝗲:
- Phase 1 vs phase 2 AI implementation
- Data quality requirements for AI
- Tools that are driving better targeting
- Why RevOps should lead (22% more impact)
- Why you can't expect exponential win rate gains

AI won't fix a bad GTM strategy.
But it gives you new options.

What creative uses of AI have you seen?

🤔

Listen to the episode "The State of GTM AI with Craig Rosenberg" on 🎙𝙂𝙏𝙈 𝙎𝙘𝙞𝙚𝙣𝙘𝙚🎙here: https://bit.ly/4b28Nwn

✌️

Head of RevOps | Recently helped SaaS organisation 19% faster growth, +10-20% sales productivity, and 30% lower GTM costs with Revenue Operations

My Name is Ali and I love building things. Whether that be data solutions, my business, or helping other organizations at the end what I'm doing is building myself into the person I want to become. 

Our thesis is that data is the lifeblood of any organization and will only work if it's made custom to your goals, needs and priorities at any stage of your growth.

Unfortunately, many businesses don't have that as they 

-  Rely on off-the-shelf solutions that are not custom to their needs and don't adapt 

- Their data is all over the place resulting in not able to make any data driven decisions.

- Can't rely on their data to make rational logical decisions. 

- They track surface-level KPIs and not actionable insights that would give them a competitive advantage 

- They don't know how to bridge the gap between business and data expertise.

Well at aztela we are that bridge that connects business and data together to develop a moat inside of your business resulting in: 

- One of our clients generated over 20x ROI since starting working with us resulting in over $10m in revenue attributed 

- Increasing profits due to seeing everything at one place which would normally take you 100h a month

- Decreased churn and increased LTV able to see bottlenecks, and root cause as opposed then spinning in circles 

- Fixing inefficient operations due to removing data silos 

We do that by utilizing data you already have and turning it into actionable insights. 

We implement custom analytics dashboards (BI) and advanced solutions to achieve business results just by utilizing data you already have.  

If you want to create a moat, competitive edege in your market click on this link to speak to our team- > 
https://calendly.com/alizu/15-minute-meeting-aztela?month=2024-04I

Ali Z

I recently met a manufacturer with 500 employees and 800 dashboards. And they were still flying blind. Your production run is still late, you still bleeding margin and your inventory is still wrong. 

If your data doesn't trigger an action, it's just noise. 

Most manufacturers, asset heavy orgs are drowning in dashboards.
Every kpi looks great but you’ve got 50 dashboards for each.

But the same issues keep showing up quarter after quarter:

Late production runs nobody catches until it’s too late.
Maintenance firefighting that kills uptime.
Inventory volatility eating your margins.
Leadership meetings where nobody trusts the numbers.

The truth?

Dashboards don’t fix problems.
They just make them prettier.

You’ve built systems of 'reporting' when what you need are systems of action.

Dashboards show you what happened.
But dashboards are just the medium, the format. 

The old model of “data → dashboard → decision” is too slow.
By the time you ask for data, ask for a change via Slack or Email, gets back to you in 2-weeks priorites already changed.   

There is no time for bureaucracy . 

Because  your data is only valuable if you can perceive, react, and re optimize  

Without it you’re not “data-driven.”

Here is what need to do.

1. Start with problems not tools. 

Don’t start with data or tools. 
Start with the decisions that move the P&L called data strategy. 

Examples:

When to switch production lines.
What is our profitability to save time on reconciliation . 
When to reorder raw materials.
Speed up drug releases  

If a dashboard doesn’t directly impact a decision or an action, kill it.
Don't built it, even if someone asked for it. 

2. Merge IT and OT for Real-Time Feedback.

Your ERP, MES, SCADA, and shop floor systems are all talking,  but not to each other.

That’s why you’re still operating at “email or Slack speed.”
IT+OT so insights can trigger changes instantly.

You are drowning in data but, using it 10% 
That’s not future tech, that’s just modern integration done right.

3. Build a Data Fabric, Not Another Pipeline.

Factories don’t need more dashboards.
They need a nervous system.

An industrial data fabric connects every machine, application, and metric through one unified data layer.

That’s how you don't need to rebuild dashboards 50 times.

4. Shift from Dashboards to Decision Systems

Don't get stuck in 'dashboard la la land' only.  
I seen orgs have more dashboards then FTEs but still can't act or trust it.

Your goal is to operationalize data to act. 

Example: 
Automated alerts when downtime exceeds thresholds
Real-time material variance notifications 
Providing users access  data_marts.

Data only valuable when it's used to built data products. 
Dashboards, ML models, AI platforms are tip of the iceberg. 
The real work is below surface.

The companies winning in 2026 aren’t the ones visualizing data better.
They’re the ones who leverage data to business impact that detect, decide, and act at scale.

Ali Šifrar

CEO @ aztela | Fractional data leader helping scaleups+ mid-market, gain control, value from their data to accelerate their data, digital & ai journey.

Recently met a manufacturer having over 1,000 connected machines. Terabytes of data. Modern data tools. But when asked COO: “Can you see your cost per unit now?". He went silent. You don't have visibility. You are collecting expensive noise.

Every manufacturing exec says they’re “data-driven.”
They’ve connected every factory, sensor, and system.
Petabytes of data streaming in real time.

Then you ask:

> “Can you tell me which line will miss its delivery target tomorrow?” 
> “Can you see cost per unit in real time?”
> “Do all your dashboards agree?”

Silence.

Manufacturers have never had more data, and never trusted it less. 
Scale has become noise.

One manufacturer in healthcare we worked had +1000 machines operating worldwide.
Over 1000s are connected and streaming data.

And even their CIO said it plainly:
“We needed a foundation of data to make insights meaningful.”

That’s the reality every manufacturing CIO is facing right now.

The last decade was about connecting everything.

The next decade is about making sense of it.

Most buy all the tools, get all the data just in case and think ‘problem solved’

But you’ve got:
-MES systems calculating OEE five different ways.
-IoT platforms pumping out unread sensor logs.
-“Digital threads” that look more like spaghetti.
-AI pilots choking on inconsistent data.

Everyone scaled data collection.
Almost no one scaled context.

You don’t have a visibility problem.

You drowning it 

1. List out inventory

Count every dataset you have.
Now count how many are actually used in decisions.
In most orgs, it’s under 10%.

The rest? 
Digital exhaust.

If your team can’t tie a dataset to a decision, kill it.

2. Kill Redundant Data Streams

Every plant collects the same metrics three times, MES, historian, IoT.
Pick the source closest to reality.
Kill the copies.
Redundant data doesn’t create trust; it kills it.

3. Align on a Single Definition of Truth

If your OEE in Plant A isn’t the same as Plant B, you’re not measuring performance.
You’re measuring chaos.
Build a semantic layer that defines metrics once.

Let every dashboard pull from it.

Truth is not a dashboard; it’s a definition.

4. Contextualize OT + IT data.

OT tells you how something happened (sensors, machines, PLCs).
IT tells you what happened (ERP, MES).
Combine both, and you can finally answer why it happened, and fix it.

5.Move from dashboards to data products.

Dashboards show you data.
Data products act on it.
Example: automatically alert engineering when defect rate exceeds design tolerance. That’s impact.

Manufacturers spent the last 10 years connecting machines.
The next 10 will be about connecting meaning.

More data won’t make you smarter.

Better context will.
Before you fund another “data lake” or “smart factory” initiative, ask one question:

“Does this make the signal stronger or just the noise louder?”

That’s the manufacturing data paradox.
And most of your competitors are drowning in it.

CEO @ aztela | We help manufacturers, logistics and asset-heavy organization mid-market build scalable AI ready data foundation so they leverage data to decrees cost and gain edge in the market.

The winners in 2026 won't be the manufacturers or asset-heavy org with the best AI models. They will be the ones who leverage their Proprietary Data that major LLM hasn't seen. 
 
Everyone’s talking about AI.
Most manufacturers and asset heavy companies aren’t remotely ready for it.
They have a data architecture problem.

Walk into any org today and you’ll see the same story on repeat:

17 PLCs all speaking different languages
ERP that only talks to itself
A historian that logs numbers no one looks at
Excel reports sitting in SharePoint folders named “Final_v3_Updated”

And yet, the board is asking, “When are we deploying predictive maintenance or generative AI?”

Here’s the truth no one likes to say out loud:

You can’t train AI on data you don’t understand.

AI doesn’t learn from isolated points.
It learns from relationships, between machines, jobs, operators, and conditions.

Here is simplified 3-Step AI Readiness Path for manufactures, nd asset heavy companies

We’ve helped dozens to go from “we have tons of data” to “edge with propitery.”

1. CONNECT,  Get the Data Moving

Every AI story starts with the same pain: the data is trapped.
In PLCs. In historians. In MES systems. In vendor clouds.

Actionable Moves:
-Connect sensors and ohters using modern, lightweight protocols.
-Stop dumping data into yet another proprietary historian. 
-Standardize the basics, timestamp, source ID, event type, value. 

If your data isn’t moving, nothing else matters.

2. CONTEXTUALIZE, Give the Data Meaning

Here’s where most companies fail.

They connect everything and celebrate- but all they’ve done is move chaos from on-prem to the cloud.

AI doesn’t care that you have 10 million data points.
It cares whether those points have meaning.

Actionable Moves:
-Build a (UNS) - a single, logical data model that organizes information by plant, line, machine, product, and property.
-Capture who, what, when, and why -  not just “Temp001 = 78°F.”
-Map relationships -  e.g., Machine A feeds Machine B; Job 1005 ran with Operator X using Material Y. 

That’s contextualization.
It’s what separates “data movement” from “data understanding.”

When your data starts speaking a common language, AI starts listening.

3. CORRELATE, Turn Context into Action

This is where it all pays off.
Once you’ve connected and contextualized your data, you can finally see cause and effect.

Actionable Moves:
-Merge OT data and IT data (from your data lake) into one model.
-Correlate across systems — production → quality → energy → supplier → environment.

Start small. Choose a use case where correlation drives ROI. 

Every manufacturer says they “want AI.”
Few are willing to do the groundwork that makes AI possible.

Winners won’t be those with the best LLMs or tools, but those who exploit their proprietary data that major models aren’t trained on

Comment me “AI Readiness” and I’ll send our 3-Step AI Readiness Playbook

Recently met a COO who wanted to scale data across multiple sources, and 12 plants. But their integration is running on a Python script written by "Joe" three years ago. Every time you want to onboard a new source, it takes months and it breaks. 

Every manufacturer says the same thing:
“We want to scale data, AI across all our plants.”

But here’s what they don’t say out loud:

“We can’t even connect half our sources, machines to the same system.”

And that’s the real reason AI isn’t scaling.
Not the algorithms.
Not the data scientists.
It’s the integration debt.

Every new line, every new machine, every new system another one-off integration, another script, another point of failure.

By the time your new data source connects the first line, the business has already moved on.

If you want scalable AI, you need scalable connectivity.
And that means one thing: Connectivity Factories.

Manufacturers understand scale better than anyone.
You don’t build one widget at a time you design a production line.

So why is your data integration still a custom craft?

Every company we talk to treats integration as a project.

-New MES? Write new connectors.
-New ERP? Rewrite the integration.
-New site? Rebuild it all again.

Each project “solves” one local pain but it creates global chaos.
Now, you’ve got hundreds of one-off integrations, no documentation, and no idea why your AI use case takes six months to go live.

Even worse if “Joe” who built this in ops/IT leaves and now you just in mess. 

The outcome?

A pile of “AI pilots” that never make it past the POC stage.
Data teams chasing protocols and APIs instead of insights.
Executives wondering where the ROI went.

We’ve seen this movie.
And the only companies escaping it are doing one thing differently they’re industrializing connectivity itself.

The companies that are scaling aren’t doing more integrations, they’re manufacturing integrations.

They’ve built standardized, repeatable systems that make connecting machines, systems, and data sources..

- Standardized integrations templates
- Use AI to assisted with mapping
- Automate deployment across plants (Use AI hint)
- Reuse logic across every integration
- Every plant “installs” connectivity, not rebuilds it

Integration becomes scalable, measurable, and fast, like any other industrial process.

The results 

-Connectivity setup time drops from weeks to hours
-Integration stops being a bottleneck.
-Data becomes reliable, consistent, and scalable
-Data and AI products finally speed up 10x.

Here’s the truth no one wants to say:
Integration is the bottleneck of every data or it team. 

Until you treat it like a scalable system not a project you’ll stay stuck in pilot purgatory.

Stop building integrations.
Start building clear process to move faster. 

We broke down the full playbook in our latest newsletter.
Check it out below ↓

Dashboards are dead. By the time your teams reacts, already lost +100k and the COO is already screaming. This is the reality for 90% of manufacturers. You don't need another dashboard, you need action. 

The line goes off spec or route delivery is late. 
The dashboard shows it three hours later.
By the time maintenance reacts, the scrap’s already in the bin and the CFO is on the floor asking what went wrong.

That’s data-delayed.

One manufacturer and supply-chain org we worked decided to fix that.

And in doing so, we quietly built what every manufacturer will eventually need:
an ‘operational’ data warehouse that runs their factory floor like a reflex not a ‘dashboard’.

After that was talking to VP of Manufacturing and COO, when one of them said this:

“We used to respond to issues in days. Now we respond in minutes. That’s the difference between a steady plant and a profitable one.”

They didn’t add more dashboards.
They redesigned how data moves, reacts, and drives action.

Here’s what changed:

-Signals that used to sit in siloed PLCs now stream continuously into a unified time-series system.

-Machine data gets contextualized instantly which product, shift, material lot, operator.

-Analytic models push recommendations back to the domain experts, not just to dashboards that are opened once a month in meeting when it’s to late.  

So when a temperature, route delay, stockouts or vibration reading goes off spec the operator sees it in seconds, not hours.

They didn’t “modernize BI.” lol 
They operationalized their warehouse.

Or what we call it used the propitery data and turn it into intelligence

Here’s the framework any asset heavy org can copy. 

1. Identify Where Latency Costs Money
Ask one question: Where does a 15-minute delay cost us $50K?
Start there. That’s your ROI case for real-time

2.Connect Machine + Business Data
A temperature spike means nothing unless you know what batch, material, and shift it belongs to.
Fuse your PLC or historian data with ERP/MES context in near-real-time

3. Stream, Don't Wait
Move from nightly ETL to continuous micro-batches.
Every minute counts. You don’t need Kafka you need freshness.
Ask yourself always “What does it cost us” 

4.Close the Loop
If your data ends at a dashboard, it’s dead.
Data → Decision → Action must be one continuous loop.
Example:
If temp deviates 2°C, the system notifies maintenance instantly.

5.Make Operators Data Domains 
When the person who is domain expert, in front line sees the data in real time they act faster than any dashboard ever could.

It doesn’t matter if you have best data stack if you only look from hindsight. 
Data has to be used to be valuable, not reported only on. 

This manufacturer and other supply chain firm didn’t get ahead by chasing buzzwords.
They got ahead by closing the latency gap between data collection → action

That’s what true data-driven company looks like
Not another BI tool

99% of Your factory Data Is Useless. You have 80,000 sensor tags. But  can't answer basic question "Why did throughput drop 4% last week?" without a 3-hour meeting arguing about who is right. You don't need 80,000 tags, you only need 4 schemas

Every manufacturing leader says the same thing:

“We’re swimming in data.”

But when you ask a simple question like

“Can you tell me, right now, why your throughput dropped 4% last week?”

You get ten different answers… and none of them are right.

That’s because 99% of factory data is completely unusable.
And it has nothing to do with AI, tools, or dashboards.

Manufacturers have a data usability problem and it’s killing trust, and ROI.

Manufacturers generate more data than almost any other sector
But less than 1–5% of that data is ever used.

Each machine speaks its own “language.”
The same tag name means different things across lines.
Quality data arrives eight hours late.

You’ve got 80,000 tags, but nobody knows what machine half of them belong to.

Meanwhile, leadership spends millions chasing "AI” or some "vertical platform " that’s built on this mess and wonders why nothing scales beyond a single pilot.

Before you can do AI, predictive maintenance, or digital twins
you need your factory data to speak a common language.

That means automating how you collect, clean, and standardize data from the shop floor.

The manufacturers who are actually scaling AI (think Coca-Cola’s operations) all did the same thing first

They implemented a data-first architecture that structures everything into just four schemas:

1.Cycle Schema: Every discrete event every few seconds, across every machine.
2.Downtime Schema: When stopped, why it stopped, and how long.
3.Part Schema: How materials move and transform across.
4.Defect Schema: Where quality issues occur and where they’re detected.

That’s it. Four schemas.
Not 80,000 tags. Not 15 disconnected systems.

Once structured, data becomes instantly usable for analytics, AI, and even GenAI copilots that let operators ask questions in natural language

“Why did Line 4 slow down yesterday?”
“Which shift has the highest defect rate?”

That’s when “data-driven” actually means something.

Here’s the Playbook That Works

1. Automate Data Engineering: Stop manually mapping tags. Use streaming pipelines that structure everything in real time.
2. Standardize Once, Scale Everywhere: One model across all plants build once, reuse everywhere.
3. Empower the Frontline: Data is worthless if it doesn’t help operators act in real time.
4. Measure ROI in Productivity, Not Dashboards: 1% improvement in OEE = millions in recovered value.

If your data team is still debating which "ai tool" to buy, you’ve already lost.

Real question isn’t “Which model should we use?”
It’s “Can our data even be used?”

99% of your factory data is sitting idle, because it’s not, trusted, or usable.

Fix that, and every AI pilot, dashboard, predictive model suddenly starts working

Recent manufacturer spent +$1millions on "Data Transformation." But you operators still trust a $2 notebook more than your dashboard or predictive model. You can't buy trust. 

Walk into any “digitally transformed” asset heavy company and you’ll see the same thing:

Robots. Dashboards. Sensors. AI pilots.

But behind the curtain?
Spreadsheets.
Email chains.
People saying, “We don’t trust the data.”

Let’s be clear you can’t automate what you don’t trust.

For the past decade, manufacturers have been told that digital transformation means buying more tech.
So they bought robots, IoT, and “predictive AI.”

And yet:

-Operators still take notes on paper.
-Finance and Ops can’t agree on OEE.
-Dashboards are beautiful and ignored.

That’s not transformation.
That’s expensive stagnation.

Because automation without trust just means you’re accelerating bad decisions.

Any Transformation Starts With Trust

Every manufacturer that’s actually scaled AI, reduced downtime, or unified global KPIs they didn’t start with tech.

They started with data trust.

Data trust means:
- People believe the numbers before they act.
- Systems agree on one version of truth.
- Decisions can be traced and defended.

If you skip that foundation, you’ll keep “digitizing chaos.”

If your “digital transformation” isn’t moving, stop buying tools, failed AI POCs and start fixing trust.

1.Fix Visibility Before Velocity
You can’t optimize what you can’t see.
Connect your systems, clean the data, and make performance visible before you automate anything.

2.Audit Where Data Trust Breaks Down
If two systems show two answers, no one acts.
Map your data flows. Document the disconnects.

You either have:
Inconsistent data - No standardization across
Delayed data - Stale and outdated
Or Opaque data- No one knows where the numbers came

You’ll find 80% of your problems are definition-process problems, prior data hitting the cloud not system problems.

3.Standardize Language Before Systems
Create a data dictionary units, KPIs, naming rules.
Make everyone speak the same language before they plug anything in.
Boring? Yes.
But it’s the single biggest unlock for scaling automation.

4.Bridge IT + OT
Transformation fails in the gap between IT and the plant floor.
Build one governance team that owns both.
Data doesn’t care who your org chart says reports to whom.

5.Build a Culture of Verification, Not Blame
If people get punished for “bad data,” they’ll hide it.
If they’re rewarded for fixing it, trust compounds.

Digital transformation isn’t about robots.
It’s about confidence in your data, in your people, in your decisions.

You can automate lines and connect machines all you want.
But if your people don’t believe the data behind it?
You’re not transforming. You’re decorating.

Automation amplifies whatever data you feed it.
If your data is wrong, automation just makes the wrong faster.

Start with trust.
Everything else scales from there.

"We’ve got data everywhere, but no answers." A COO of a $100M manufacture told me this last week. They had an $1m ERP, 400 dashboards and 5,000 sensors BUT still couldn't answer why the production line went down on Tuesday or drop in OEE across.  

Let’s be blunt most manufacturers and asset-heavy companies don’t have a “data strategy.”

They have 20 years of data chaos.

ERP, MES, SCADA, spreadsheets, PLCs none of them talk.
Every team has their own version of the truth.
And every “digital initiative” dies in pilot purgatory.

Yet when executives hear “data strategy,” they imagine a two-year consulting project full of diagrams and jargon.

That’s not what this is.
An industrial data strategy is not a report.
It’s the operating system of your factory.

A mid-market manufacturer we met had spent years investing in tech new MES, shiny dashboards, “AI readiness.”

But they couldn’t answer three simple questions:

- What’s our real-time OEE across lines?
- Why did that line go down last Tuesday?
- Which supplier issue is driving our scrap rate?

The COO summed it up perfectly:

“We’ve got data everywhere, but no answers anywhere.”

That’s an architecture problem.

The most successful mid-market manufacturers the ones scaling AI and digital aren’t the ones buying more software.

They’re the ones that finally said:

“Enough chaos. Let’s fix the foundation.”

They follow a 4-layer data blueprint that actually works:
Ingest → Store → Contextualize → Act

1.Ingest Capture What Matters, Not Everything

Stop streaming every sensor tag “just in case.”
Decide what data actually drives a business decision.

Start by connecting:
PLCs (machine states, alarms)
MES & ERP (production orders, downtime)
Quality systems (defects, batch IDs)
Transform and filter at the edge before it ever hits your cloud.
Collect less. Use more.

2. Store tructure Your Data or Drown in It

You don’t need a “data lake.” You need data discipline.

Use the Bronze–Silver–Gold model. 
With clear constitutional  layer. 
Your "predictive model"  is only as good as the structure beneath it.

3.Contextualize Turn Data Into Meaning

Most manufacturers already have the data.
They just can’t relate machine → operator → supplier → part → defect.

That’s where contextualization comes in:

Use a Unified Namespace (UNS) to standardize naming and structure.
Build a Graph to link data across systems.
You can’t run predictive maintenance or AI until your data understands itself.

4. Act Stop Reporting. 
Dashboards are fine. Closed-loop action is better. 
Empower domain experts. Delayed or not used data is useless  
Data → Transformation → Trigger

Because in 2026, the manufacturers winning aren’t the ones with more sensors
They’re the ones who know how to turn proprietary data into intelligence and action.

→ Comment “Blueprint” 

We’ve built a free 90-Day “Manufacturing Data Strategy” the exact 4-layer plan mid-market manufacturers, supply chains & asset heavy organization.

8 out of 10 "Predictive Maintenance" projects fail. Recent manufacturer wasted $400k and 12-months on model nobody used. Not because the AI is broken. But because of your data architecture. 

Every slide deck every pitch sounds the same:

“We’ll predict when your machines will fail before they fail, inventory or delivery delays” 

But here’s the truth:
8 out of 10 predictive maintenance projects fail before they ever deliver value.

Not because AI doesn’t work.
Because your data doesn’t.

A $400K Lesson in Skipping Basics

A mid-market manufacturer spent 18 months, $400K, and hired a small data science team to predict bearing failures on a critical asset.

They had sensors.
They had dashboards.
They had an ML model running in the cloud.

But they didn’t have context.

Maintenance data lived in SAP.
Telemetry lived in the PLCs, IoTs, WMS and POS.
Failure logs were Excel sheets in someone’s inbox.

The model’s predictions were useless.
Technicians ignored them.
The system was shut down six months later.

The post-mortem was simple:

“We tried to predict the future without understanding the present.”

Why Most Predictive Maintenance Projects Fail

Let’s cut through the consulting jargon the real reasons are painfully simple:

Data Silos Everywhere – You can’t train an ML model if your maintenance, MES, and sensor data don’t talk to each other.

No Context Layer – You have data points, but no lineage no way to connect a failure to operator, supplier, or process.

Jumping Straight to AI – You skipped the foundational stages: OEE visibility, diagnostic analytics, and process understanding.

Bad Edge–Cloud Strategy – You’re streaming every millisecond of data into your lake turning it into a swamp.

No Trust from the Floor – Maintenance teams don’t act on alerts they don’t believe.

The Playbook: What Smart Manufacturers Do Instead

Smart manufacturers don’t start with AI.
They start with data discipline.

Here’s the exact playbook they follow:

Step 1: Build Real-Time Visibility (Descriptive Analytics)

Connect assets. Build OEE visibility. Fix your real-time awareness before predicting anything.
Start with least amount of variables. 

Step 2: Contextualize Your Data (Diagnostic Analytics)

You are lacking why. 
Map maintenance, MES, and IoT data into a unified model often a Unified Namespace or Graph.

Step 3: Standardize and Clean Early

Set a single schema for units, tags, and statuses. Enforce it at data ingestion, not reporting time.
This ensures your data is usable for analytics from day one.
You can’t have one source temperature in Celsius and another in Fahrenheit.

Step 4: Push Small One Asset

Don’t boil the ocean. Pick one high-impact thing, route, machine, inventory, train a model, and prove ROI in 8–12 weeks.

Step 5: Close the Loop

If can’t act on it, it’s usless. 
Automate the workflow. 

If your predictive maintenance project is failing, it’s not an AI problem.
It’s a data architecture problem.

The worst way to find out your data is broken is from an angry COO. Global manufacturer 5+ plants, 10,000 suppliers, 5TB of data. Hundreds of automated pipelines and dashboards. If your VP finds the bug before you do, you, it's to already to late. 

Not because of an outage.
Not because of a cyberattack.
Because of one dropped column.

A single schema change in a production system one field someone thought “wasn’t needed anymore” silently disappeared.

By the time anyone noticed, it had already cascaded:

Dashboards went blank.
Planning models failed.
Forecasts froze.
Supply chain teams reverted to manual SAP transactions.

Factories couldn’t see what to produce next.
Managers couldn’t trust their data.
And the global ops team spent days firefighting instead of running the business.

That’s how one schema change broke the business.

This isn’t a data story. It’s a reliability story.

Every company thinks their data problems are “technical.”
They’re not.
They’re operational.

If your business depends on ERP, BI, or AI systems you’re one schema change away from chaos.

Why?
Because every data pipeline is a set of unwritten contracts.
When those break, everything downstream does too.

The field that disappeared wasn’t the problem.
The lack of enforcement was.

The real lesson: Every schema is a contract.

You’d never let a supplier change a part spec without notice.
So why do you let your systems change data structures without validation?

Your data is your supply chain.

Schemas are your supplier specs.
Pipelines are your logistics network.
Dashboards are your distribution.

And when your “data supplier” changes specs without a contract your entire operation stalls.

That’s what happened here.
 How to Stop This From Happening to You

Here’s exactly how we prevent this chaos for manufacturers, and asset heavy orgs. 

1.Enforce Data Contracts

Define exactly what data producers deliver fields, types, update frequency, ownership.
In DBT, that means schema + test definitions.
In business, it means accountability.
No contract, no data.

2.Shift Testing Left

Don’t wait for dashboards to fail.
Validate schema changes before deployment.
Treat data pipelines like code versioned, reviewed, and tested before release.

3.Automate Drift Detection

Use DBT, Great Expectations, or even AI (LLMs trained on metadata) to flag schema drift automatically.
If a column changes, you should know before your planners do.

4.Make Ownership Non-Negotiable

No one should say “that’s not my data.”
Ops, Finance, and IT all own reliability.
Because when data fails, everyone loses.

The next major business outage won’t come from a ransomware attack or a system crash.

It’ll come from a schema change no one noticed.

That’s the quiet risk every executive is carrying right now.

Your dashboards won’t crash they’ll just lie.
Your forecasts won’t stop they’ll just be wrong.
And your teams won’t notice until they can’t deliver.

Your supply chain is broken. They are automated, AI-enabled but completely blind. Ask a COO why they’re out of stock. You’ll get 14 different excuses from 5 different systems. The problem with modern supply chains isn't logistics or tools. It's data. 

You don’t need more sensors or $2m digital twin.

You need to see your data move.

One manufacturing firm we worked figured this out 

Their CIO and COO told a story that stuck with me:

“We can trace diaper from the store shelf all the way back to the exact microsecond it hit the glue gun in the factory.”

That’s not a metaphor.

It’s data lineage every transaction, every handoff, every timestamp linked from the consumer back through manufacturing, distribution, and finance.

Here’s how it works for them

When a parent scans a product pack for loyalty points, that simple scan connects:

-the store it came from,
-how long it sat in distribution,
-what plant produced it,
-whether that batch had issues,
-and what it cost to get there.

That’s real visibility.
That’s how you prevent supply chain fires before they happen.

And no you don’t need to be giant conglomerate like P&G or spent $5m on tools. 
Wrong approach. 

You just need to stop treating data as IT’s job

We’ve helped manufacturers build the same visibility using the systems they already have 

1.Start with a single pain point.

Don’t “launch a dashboard or project.” 
Start with urgent problem, not tools, costing you. 

“Why do we keep running out of SKU 123?”
“Why does lead time swing by 40% region to region?”

If that question crosses multiple functions, perfect. 

2. Map your “data DNA.”

Don't go into built mode yet.
List every system and entity that touches that process.
ERP → MRP → WMS → TMS → Finance.

Who owns each? Where does the data change? What rules apply? Where used downstream?
This gives you visibility into how your supply chain actually runs behind scenes before building it.

3.Instrument the flow.

Add timestamps, anomaly checks, and ownership.
Every data handoff should say: who, when, accuracy, and next hop.

4.Tie lineage to data products  

Example 
Problem Repeated out-of-stock in key stores
Use Case: Trace product flow from DC → shelf → sales
ROI: +15% inventory accuracy 

or facing 

Customer complaints on product defects. You can trace SKU to manufacturing batch & time stamp and cut issue-resolution time from weeks to hours 

That’s the business case. 

5. Assign business ownership

IT can’t own data lineage
If everyone owns their data node, the lineage keeps itself healthy.

6. Build modular, not monolithic. 

Start with urgent problems like “order to delivery”
Prove it in 4–6 weeks. Then add

Most supply chains don’t fail because of bad logistics
they fail because leaders can’t trace their data

Future-ready supply chains aren’t the ones that predict the future or best tools.
They’re the ones that leverage data

Comment "Supply Chain", and I’ll send you  the exact 6-week playbook manufacturers are using to fix this

Until IT and OT speak the same language, your $1M dashboards are just expensive "useless digital decorations" Recent COO asked "Why did yield drop 4%?" and nobody could tell him. 

A few months ago, I sat in a manufacturing  conference room watching two teams argue about the same problem.

Your digital transformation didn’t fail because you picked the wrong tool.
It failed because your IT and OT teams don’t speak the same data language. 

They just didn’t know it yet.

IT is building data lakes, deploying Snowflake, and talking about dashboards and OPEX. 
OT is wiring sensors, logging PLC tags, and talking about machine uptime.

IT thinks in tables and transactions
OT thinks in tags and tolerances.

They were both convinced the other side didn’t understand the business.

And they were both right.

Then COO walks in, looks around, and asks one question:
“Why did yield drop 4% on Line 3 last quarter?”

Silence.

IT pulled up reports showing production volumes, OEE, and downtime trends.
OT pointed at machine logs and said, “Temperature spikes, but we don’t know when.”

No one could connect the two.

The data existed.
The insight didn’t.

That’s when it hit me, they weren’t arguing about systems. They were speaking different languages.

Here’s the truth no one likes to say out loud:

Digital transformation doesn’t fail because of your tech stack.
It fails because your teams don’t share a common data language.

IT speaks “business.”
OT speaks “machines.”

They both produce data, just in completely different dialects.

IT

What: Orders, materials, costs
When: Daily jobs
Where: Cloud systems
Why: Visibility

OT

What: Sensors, cycles, tolerances
When: Millisecond updates
Where: On the shop floor
Why: Uptime

You can’t unify that by buying another platform.
You have to translate it into a shared language, one built on context, not tools.

We started small.
No buzzwords. No 2-year roadmap.

Just one question:

“What’s the one decision you can’t make because your data doesn’t connect?”
We picked yield analysis.

Here’s what we did in 60 days:

1. Mapped the language
Defined what “batch,” “machine,” and “tolerance” actually meant — across MES, ERP, and IoT.

2. Unified identifiers
Synced Part IDs, Batch IDs, and Timestamps so systems could finally talk to each other.

3. Built one contextual model
Merged design tolerances from PLM with live machine data  so when a deviation occurred, we knew which part, which shift, which batch.

4. Closed the loop
When a tolerance breach hit, it didn’t go to a dashboard — it triggered alert and created a feedback ticket to engineering.

After a while COO came back.
Asked the same question:

“Why did yield drop 4% on Line 3?”
This time, now were able to see full picture.

“Machine 12 ran outside tolerance for 2 hours on Batch 431A, same issue as the supplier lot from March.”

No debate. Just facts.

That’s what happens when IT and OT stop reporting separately and start understanding each other’s data

Recently talked to CIO of an F1 team on why they win. He didn't say due "perfect data architecture." But due speed. Most burn $500k, 18-month designing the "perfect" architecture that delivers zero impact. By the time they finish, the business has moved on.

You don’t win by building the most complex car. 
You win by iterating faster than your competitors.

Every team has access to the same tech, data, and talent pool.
But the fastest ones don’t out-engineer they out-learn.

Meanwhile, most data teams are doing the opposite.

They spend 12–18 months designing “perfect architecture” multi-cloud diagrams, vendor evaluations, governance committees and by the time it’s ready,
the business has already changed direction.

Architecture matters. Data foundation matters. 
But only as a means to speed, not an end in itself.

Most data programs die because they optimize for elegance instead of execution.

I’ve seen companies burn $500k+ on replatforming projects that produced zero usable impact because no one asked:

“Are we moving faster? Are we decreasing delivery times?

The F1 world understands this better than anyone.
Every race weekend, teams run thousands of experiments.
They test, fail, learn, and push updates in real time.

Not once a year.
Every week.

That’s the mindset Data and CIOs need to steal

1.Launch fast, learn faster.

Stop over-architecting.
In F1, no car is “finished.” They deploy updates constantly.
Do the same deploy data products in 2–4 week sprints, not quarters.
Build, test, iterate. Treat every release as an experiment.

Measure success in adoption, trust and results. 

2.Reduce drag first.

The slower a race car, the more drag it carries.
Same with your data stack.

Cut tools. Kill redundancy.
Don't mesure success in tool count.
Most CIOs think more platforms = more capability.

Wrong. 

It’s more integration overhead, more latency, more failure points.

3. Optimize for marginal gains.

F1 wins come from 0.2% improvements done 1,000 times.
Your data team should operate the same way.
Small wins compound into trust, adoption, and ROI.

4. Train like a pit crew.

Pit stop takes <3 seconds because everyone knows their role.
That’s what your data operations should look like.
Small, cross-functional teams aligned to outcomes not tickets.
Reward precision under pressure.

5. Balance short-term with long-term.

F1 teams design next year’s car while racing this one.
CIOs must do the same: ship value weekly while evolving the foundation.
You are not “strategic”, if you’re not delivering in real time.

Architecture wins awards. 
Agility wins championships.

Your competitors aren’t beating because they have better tools.
They’re beating you because they learn, adapt, and ship faster.

If you’re leading a data team, ask yourself:

Are we optimizing for speed to value?

Business doesn’t care how elegant your data model is.
They care how fast it helps them win.

Stop building like an enterprise IT department.
Start building like an F1 team.

Your Supply Chain is running on a 12-person email thread. By the time you figure it out, it's to late  Your nice dashboard show you what happened But not what to do next. 

Every ops leader knows the moment:
It’s 2:13 a.m.

Your phone buzzes “URGENT: shipment delayed.”

You scroll through a 10-person email chain, 12 replies deep.
Everyone’s copied, nobody’s deciding.

By the time you re-route, the production line’s idle, the freight’s expedited, and the cost has already hit the P&L.

Here’s the truth:
You don’t have a delay problem.
You have a manual decision problem.

Supply Chain Delays Are Inevitable. Manual Reactions Aren’t.

Most manufacturers and distributors are still reacting to delays like it’s 2015.

Planners manually reschedule orders.
Logistics teams call carriers and “wait for updates.”
Managers decide expedites in a Teams thread.
Finance finds out three days later when the bill hits.

Everyone’s working hard. Nobody’s working smart.

Your systems can tell you a shipment’s late.
But they can’t tell you what to do about it.
You have nice dashboards but they are usless. 

Don't get me started on IBP or other parts. 

That’s the missing layer:
Automated decisioning.

What Automated Decisioning Actually Means

It’s not “AI.”
It’s not a magic dashboard.
It’s simply this:

When an event happens (a delay, shortage, or disruption)…
-It’s detected in real time.
-Your rules or models evaluate the impact.
-The next best action happens automatically.

No waiting for planners. No “what should we do?” meetings.

The human becomes the exception handler, not the bottleneck.

1. Map Where Decisions Actually Happen

Don’t start with tech. Start with truth.
List every time someone manually reacts to a delay:

“If supplier X is late, we expedite with Y.”
“If truck misses cutoff, alert customer service.”

2. Build a Real-Time Event Backbone

You can’t automate what you can’t detect.

Stream your key operational events (shipment, supplier, production) into a central.

No delay should be trapped in a spreadsheet or silo.If data doesn’t move in real time, your decisions never will. 
That’s your “nervous system.”

No more email alerts. Real-time triggers instead.

3. Codify the Decision Logic

Turn tribal knowledge into rules

Don’t start with some complex AI or ML model. 
Turn common sense turned into code. 
Don't force AI where it makes no sense. 

3.Connect Decisions to Action Systems

A decision without execution is just another alert.

Connect your logic from central source of truth ERP, TMS, MES, or CRM so the system can actually do something:

Reschedule production.
Reroute shipment.
Notify customer automatically.

That’s how you close the gap between insight and impact.

Here’s what happens in real life:
Delay event detected instantly.
System calculates risk.
Auto-reroute triggered.

A Tier-1 auto supplier we worked with cut reaction time from 12 hours to under 10 minutes.

Every delay costs money.
But slow decisions cost more

Your 2026 AI strategy is already dead. You are missing the one thing that actually makes AI work. This year, the winners won't be companies with the best AI model. But those who have context.

“We’re going all in on AI this year.”
Except 9 out of 10 of those “AI strategies” are already dead.

Not because of bad models.
Not because of missing talent.
Because your data doesn’t know how to talk to itself.

Your CRM calls it “customer.”
Finance calls it “account.”
Ops calls it “site.”
And your AI? It just calls it confused.

For years, companies believed if they centralized data, they’d create intelligence.
So they built warehouses.
Then lakes.
Then lakehouses.

Now, they’re realizing:

→ Centralized ≠ Connected.

Your data warehouse tells you what you have.
Your knowledge graph tells you how it all connects.

Without that connective tissue, your AI models are operating blind:

Forecasting demand without understanding product hierarchies.
Predicting churn without linking customer interactions.
Detecting anomalies without connecting to operational events.

That’s why AI pilots look great in demos, and fail in production.
Because AI isn’t about bigger models.
It’s about connected meaning.

If you want AI that works, stop chasing models.
Start teaching your data to speak your business language.

Here’s the 5-step playbook:

1. Build a Business Ontology (Not Another Dashboard)

Map how your business actually thinks: customers, orders, assets, suppliers, SKUs.

“Product” → linked to “Manufacturing Batch” → linked to “Quality Event”

This isn’t IT work. It’s strategy work.
Run a 2-day ontology sprint with your functional leaders.

2. Link Your Structured + Unstructured Data

AI doesn’t just learn from tables.
It learns from emails, PDFs, call notes, logs.

Maintenance logs mention “compressor 14,” which is tagged in OT data as “C14.” 

Connect them through entity tagging and relationships.

3. Create a Semantic Layer (Your AI Translator)

Your existing data catalog describes data. Knowledge graph interprets it. 
This layer turns relationships into queryable context.

So when you ask, “Which suppliers caused downtime in the last 6 months?",  it actually answers correctly.

4. Anchor Every Data Product in the Graph

Your “Customer 360” dashboard isn’t 360 unless it’s connected across domains.
Graphs make products dynamic and self-healing when structures change.

Like 
Dynamic supplier risk scoring (linked across purchase orders, logistics) 

5. Leverage proprietary data 

Those who gone win are not with best LLM models or AI tools. 
But those who leverage their proprietary data. 
Every data point, variable is a edge in your market. 

If your AI strategy doesn’t include a knowledge graph, it’s not a strategy.

You can’t build intelligence on top of disconnection.

The future of AI won’t belong to the companies with the best prompts.
It’ll belong to the ones whose data knows how it relates to itself

90% of manufacturers are jumping straight to "AI". But But takes them 5 days and 20 meetings to explain why a number changed. Half of operators are still in excel. You are not 'data-driven' and here is how to change that in 2026

You’ve got Power BI reports, a shiny Snowflake instance, maybe an AI pilot running in a corner of one plant.

But when I ask execs one simple question

“Can you trace one KPI from source to decision, end to end?”

90% of the time, the answer is no.

Or it takes 5 days, 5 people and 20 meetings. 

That’s not a technology failure.
That’s a maturity gap.

Everyone wants to jump straight to “predictive maintenance” or “AI optimization.”

But manufacturing, logistics orgs data maturity doesn’t work like that.
You don’t go from chaos to AI in one quarter.

You pass through five unavoidable stages each one with different pain, ownership, and ROI.

Here’s how to know where you actually are 

Stage 1: The Dark Ages The Blind Factory

Data exists everywhere, but no one uses it.
It’s in SCADA, MES, Excel sheets, and people’s heads.
Operators rely on intuition, not information.

Symptoms:

Endless firefighting
Paper logbooks
Reports created manually once a week
manufacturing or supply chain. 
Your job is to assign ownership and Inventory your systems.  

Stage 2: The Conflicting Era

You can see data now… sort of.

Dashboards pop up everywhere, but they all tell different stories.
Finance shows 92% OEE. Ops shows 78%. Nobody knows who’s right.

Symptoms:

“Multiple versions of the truth"
Dashboard fatigue
No agreed KPI definitions

Your job is to standardize definitions, resurface and redundant dashboards or sheets. 

Stage 3: Diagnose The Context Stage

This is the turning point.
You finally understand why things happen not just what happened.

Symptoms:

You can link downtime to shifts, batches, or recipes
Teams start trusting reports
OT and IT actually talk to each other

Your job here is to build your context layer.
Context isn’t metadata it’s meaning.
It’s how you connect sensors → equipment → process → cost.
Without it, you’ll never scale.

Stage 4: Predictive Stage

Now you can see ahead.
Predictive maintenance, quality forecasting, energy optimization they’re real.

But this is where most teams stall.

Why?
Because they forgot to govern context.
Models degrade. Data drifts. Trust erodes.

Your job here is version your context.
Treat it like code. Keep feedback loops tight.

Stage 5: Operationalize 

This is the endgame. Closed-loop optimization.
Data drives action automatically, end users able to access the data with no bureaucracy.

Not everything is a 'dashboard'

It’s about culture, governance, and compounding context

Data maturity is leadership work, not technical work.
And once you know your stage, you can stop chasing hype and start building capabilities that last and impact today and set you for long term success .

Last week COO of a $250M manufacturer said to me: “We’ve got SAP, ERP, WMS, IoT but when something breaks, can’t see it.” It took them 7 meetings and 20 spreadsheets  to uncover root cause or trace a defect. Here is what most manufacturers and supply chain miss 

He wasn’t exaggerating.
His team had more dashboards than decisions.
And they still couldn’t trace a defect without six people, three spreadsheets, and a week of digging in meetings. 

Your problem isn’t visibility.
It’s traceability. 

Manufacturers or supply chain orgs don’t fail because they lack data.
They fail because their data lives in isolation.

MES knows what happened.
ERP knows what order it was for.
SAP for you fleet optimizations. 
QMS knows which batch failed.
Maintenance knows when the line went down.
IoT knows the temperature and vibration.

But no one, and I mean no one can tell you how those things connect.
That’s why your teams spend hours pulling CSVs, merging spreadsheets, and arguing over “whose data is right.” 

You’ve built a data warehouse. You’ve visualized KPIs.
But you never built a knowledge graph.

Everyone’s obsessed with collecting more data.
No one’s focused on connecting it.

You don’t need another system of record.
You need a system of relationships.


Think of it as your factory’s digital DNA:
Machines → Batches → Materials → Suppliers → Quality → Customers in one connected truth.

Forget OT + IT. It’s bigger then that. 

The best manufacturers and supply chain organizations, retailers we work with don’t chase “more data.”

They chase connected data.

Example:

“Show me all defects in the last 90 days tied to Supplier X’s batch, produced on Line 3 during night shift.”

If you can’t answer that in seconds, you don’t have traceability.
You have data chaos with prettier charts.


You stop asking “what happened?” and start answering “why.” 

We built a knowledge graph for a precision parts manufacturer last year.
Before that, root cause investigations took 7 days on average.
Seven.

After 90 days, they could trace a defect to a supplier in 30 minutes.

They didn’t buy a new platform.
They just connected MES, ERP, and QMS data into a graph model.

That one change saved $1.2M in rework and scrap in the first quarter.
Not because of AI.
Because of context.

The smart ones don’t start with tools.
They start with relationships.

1. Map Entities and Relationships .
2.Connect Existing Systems into Graph
3. Tie to urgent business needs and questions. 
4. Data has to act. Stop reporting only dashboards.

Everyone in manufacturing is chasing AI.

But AI is useless when your data can’t explain itself.
If your systems can’t talk to each other, your machine learning model will just automate confusion.

Because until your data knows how it connects, it doesn’t matter how much you have, it’s just noise.

If your team still needs a week to find the cause of a root cause, you don’t have a visibility problem.

You lack context

Over 80% of manufacturing data modernization's fail. The moment you add a new source, tool, plant everything breaks. You built architecture for stability. But your business is built for growth. 

When your company doubles in plants, products, or markets- your data foundation collapses under its own weight.

Pipelines break. Metrics drift. Dashboards contradict each other.
And AI pilots stall because nobody trusts the inputs.

It’s because your architecture was built for stability, not growth.

You start your “modernization” journey with a clear plan:

- Move to the cloud.
- Centralize ERP, MES, and historian data
- Build a few “data products
- Launch shiny dashboards and AI PoCs.

Then growth hits.

-You acquire two plants with different historians.
-One uses OPC-UA, the other Modbus.
-Every site has its own definition of “OEE.”
-Finance calls it “downtime.” Ops calls it “line changeover.”
-Suddenly, your “single source of truth” has 10 dialects.

This constantly happens in asset heavy industries. 

And now your so-called “modern data stack” becomes the same mess just with higher AWS costs.

This is where 80% of manufacturing modernization efforts stall.
Because the foundation wasn’t designed to flex with growth.

What the top-performing manufacturers do differently:

They don’t chase tools.
They build systems that absorb chaos.

Here’s how they do it:

1. Design for expansion by default.

Don’t hard-code your integrations or models.
Build metadata-driven ingestion and transformation frameworks that can adapt as new systems, schemas, and sites come online.

If your pipelines can’t onboard a new plant without rewriting code, you’re not modernized- you’re fragile.

2. Harmonize semantics before harmonizing systems.

If “OEE,” and “margin” mean different things in every source or team no data lake can fix that.

Start with a business dictionary.
Run cross-site metric harmonization meetings.
Document it in your catalog.
Make the data mean the same thing before you centralize it.

3. Integrate OT early, not later.

Stop treating historian and sensor data as a “Phase 3.”
That’s the data that drives every AI and predictive use case. 
If it’s not integrated, you’re modernizing for reporting not intelligence.
Of course align it to business impact. 

4. Scale ownership, not just engineering.

Central data teams can’t keep up with hyper-growth .
Assign data stewards and champions by domain (Ops, Quality, Supply Chain).
Give them visibility into data trust in format they like. 
Get data in domain expertise hand, without them needing to email or slack anyone. 

That’s how you scale governance without bureaucracy.

5. Treat modernization as a capability, not a project.

There’s no “done” state.
Your platform, models, and governance must evolve with every product line, plant, and AI use case.

If your foundation can’t handle new systems, or new product lines- you haven’t modernized anything.

You’ve just moved legacy to the cloud.

taplio

The Battle

Eddie Reynolds

Ali Z

Followers

Posts

Engagements

Likes

Performance ScoreEddie Reynolds's LinkedIn Performance

Performance ScoreAli Z's LinkedIn Performance

Ready to grow your LinkedIn presence?

taplio

The Battle

Eddie Reynolds

Ali Z

Followers

Posts

Engagements

Likes

Performance Score.css-goub6s{font-size:var(--chakra-fontSizes-md);font-weight:500;color:var(--chakra-colors-gray-600);display:block;margin-top:var(--chakra-space-1);}Eddie Reynolds's LinkedIn Performance

Performance ScoreAli Z's LinkedIn Performance

Ready to grow your LinkedIn presence?

Performance ScoreEddie Reynolds's LinkedIn Performance