I was early Marketing at Uber spending 5 years building Brand, Performance, and Lifecycle Marketing Data Science teams. I got to work with & learn from world-class marketing, data science, and martech leaders. Over the course of my career there I've measured over $1Bn in Marketing Spend and estimate that I've helped drive $200M+ in incremental revenue and cost savings . 

Then I moved on to a few early stage startups and scaleups (both as a FTE and Consultant) where I learned how to be even scrappier when building Marketing. 

Now I help Marketers access those same playbooks top Consumer Tech Marketers use to prove, optimize, and scale ROI by sharing frameworks, processes, and analyses. 

I'm doing this because I love the Marketing stories and people behind it in a world where Product seems to get most of the credit. There's so much magic in Marketing and I want to share more about it. 

Check it out here → https://experimental.beehiiv.com/

Outside of my newsletter and podcast, I love spending time with my 2 beautiful kids and my incredible wife in Amsterdam. We used to love traveling and eating out but you know... kids ...  so now I love cooking and exploring all the fun things little kids can do!

Sundar Swaminathan

CEO: We're using 7-day click attribution. 

New CMO: What's your average purchase cycle?

CEO: Probably 30 days from first awareness to purchase.

CMO: Then our attribution window is cutting off 75% of the customer journey.

CEO: Wait, what?

CMO: We're only giving credit to touchpoints in the last 7 days. Everything that happened before that gets ignored.

CEO: So we should use a 30-day window?

CMO: Maybe.

CEO: This is confusing. How do we pick the right window?

CMO: There is no right window. There's the window that matches your business reality, but more importantly reflects incrementality. 

CEO: What does that mean?

CMO: If you're selling impulse purchases, 1-day windows probably work. If you're selling considered purchases, you need longer windows but also need to acknowledge the attribution is fuzzy.

CEO: So what should we do?

CMO: Test different windows and see how channel performance changes. If Facebook looks amazing at 7 days but terrible at 28 days, that tells you something.

CEO: What does it tell me?

CMO: That Facebook is getting credit for conversions close to the purchase, but it might not be the thing that started the journey.

CEO: How do other companies handle this?

CMO: The sophisticated ones stop relying on attribution windows alone and run incrementality tests instead.

CEO: So attribution is useless?

CMO: It's useful for directional insights. Just don't make million-dollar decisions based on which arbitrary window you picked. 

–
👋 I’m a former Uber Head of Marketing Science where I optimized $1Bn+ across Brand, Performance, and Lifecycle Marketing

🔔 Follow for Marketing Strategy & Science 
♻️ Repost because sharing is caring
⬇️ Save to send to your CEO later

Former Head of Marketing Science at Uber who optimized $1Bn+ across Brand, Performance, and Lifecycle → Sharing weekly playbooks used by top Marketers to measure and scale ROI.

CEO: Need to cut Brand Marketing budget by 30%.

CMO: Why Brand Marketing?

CEO: We need to preserve cash now. Brand Marketing feels like the safest place to cut.

CMO: Why does it feel safe?

CEO: Because we can turn it back on when things improve. Engineering headcount we can't.

CMO: You're right that we can turn it back on. But you're wrong that it's safe.

CEO: Explain.

CMO: When you cut Perf Marketing, you see the impact immediately. Revenue drops in weeks. And when you cut Brand Marketing? You won't feel it for months, but it'll come.

CEO: So we should cut Performance instead of Brand?

CMO: I'm saying both have consequences. Performance cuts hit revenue fast. Brand cuts hit revenue slowly, which makes them feel safer but they're actually riskier.

CEO: How is slow worse?

CMO: Because by the time you realize brand cuts hurt, your competitors have filled the void. Customers forgot about you. And it takes twice as long to rebuild than it did to build initially.

CEO: We're not cutting all of marketing. Just 30%.

CMO: Where's the 30% coming from?

CEO: I was hoping you'd tell me.

CMO: Well, it's not coming from me and it's not what I'd recommend, but I know I'm not going to win this fight, but I do have a recommendation. 

CEO: What do you recommend?

CMO: Cut performance by 15% and brand by 15%. Spread the pain. Don't kill either completely.

CEO: Why not just cut the one that's less important?

CMO: Because Brand Marketing is not less important! They work together. Brand makes performance cheaper. Performance capitalizes on brand equity. Kill one and the other suffers too.

CEO: How do other companies handle this?

CMO: The smart ones actually do the opposite of what you're saying. They protect Brand Marketing during downturns because that's when competitors are cutting. You can gain share cheaply. The desperate ones cut everything and wonder why recovery takes years.

CEO: That's a tough call.

CMO: It is. But cutting Marketing always feels safe because the damage is delayed. That's what makes it dangerous. 

–
👋 I’m a former Uber Head of Marketing Science where I optimized $1Bn+ across Brand, Performance, and Lifecycle Marketing

🔔 Follow for Marketing Strategy & Science 
♻️ Repost because sharing is caring
⬇️ Save to send to your CEO later

CEO: Brand campaign doesn't look it's working! 

CMO: What metrics are you looking at?

CEO: The usual. New customer acquisition, conversion rate, revenue growth. All flat.

CMO: Good :) 

CEO: What do you mean? You said this campaign would drive growth.

CMO: It will. In two to three months. 

CEO: Two or three months? Why so long?

CMO: Because of how our customers actually buy. Walk me through how you decided to buy those shoes you're wearing. 

CEO: Hmm...I saw ads for it, ignored them for a while, then a colleague mentioned it, then I researched it, then maybe a month later I finally bought it.

CMO: Why'd you buy it a month later?

CEO: I needed shoes and I saw another ad for them which .... OH! I get it! 

CMO: Exactly! That's what's happening here. Our product isn't something people impulse buy. We cant snap our fingers and make people want our product.

CEO: So we're just... waiting for them to be ready?

CMO: We're making sure that when they're ready, they think of us first instead of our competitors.

CEO: But how do we know it's working if we can't see results for months?

CMO: We track leading indicators like search volume for our brand terms. Those move first. Revenue moves later.

CEO: What are those showing right now?

CMO: Search volume for our name is climbing. 

CEO: Why didn't you lead with that?

CMO: Because you asked about revenue. But revenue is a lagging indicator. It's like planting a tree and asking why there's no fruit after six weeks.

CEO: That's a good analogy.

CMO: Thanks. Now, stop texting me and go back to bed. It's 3:30 AM.

--
🔔 Follow me for Marketing Strategy & Science 
♻️ Repost because sharing is caring
⬇️ Save so you don't lose it

CMO: Our MMM is crazy. It changes all the time. 
.
Me: How often are you refreshing the model?

CMO: Monthly. We want the latest insights.

Me: That's your problem.

CMO: Why? Shouldn't we use the most recent data?

Me: MMM needs long time horizons to detect patterns. Monthly refreshes are too frequent. You're picking up noise, not signal.

CMO: How long should we wait?

Me: Quarterly at minimum. Ideally every six months. You need enough data to see through seasonal fluctuations and campaign variability.

CMO: Okay, what else could be causing the instability?

Me: Are you changing variables between runs?

CMO: Sometimes. We add new channels or remove ones we stopped spending on.

Me: Every time you change the model structure, you're asking a different question. The answers will change because the question changed.

CMO: So we need to keep the structure consistent?

Me: As consistent as possible. If you must add variables, retrain the full history with the new structure. Don't just append.

CMO: This is frustrating. I thought MMM would give us definitive answers.

Me: MMM gives you directional guidance, not precision. It's better at telling you "brand is underinvested" than "spend exactly $47K more on brand."

CMO: Then why did we spend six figures on this?

Me: Because it's still valuable for strategic allocation decisions. Just don't expect it to be as precise as an A/B test.

CMO: How should we be using it then?

Me: Use incrementality tests for precision. MMM for strategy. Attribution for daily steering. 

--
🔔 Follow me for Marketing Strategy & Science 
♻️ Repost because sharing is caring
⬇️ Save it so you don't lose it

CEO: Our ROAS is at 8x! Best it's ever been.

New CMO: Why do you celebrate ROAS?

CEO: Because it's a good metric! 

CMO: But ROAS doesn't tell you if you're growing. It tells you how efficient your spending is.

CEO: Aren't those the same thing?

CMO: Not even close. We could have 20x ROAS and be shrinking as a business.

CEO: Give me an example.

CMO: We're spending less on acquisition and only targeting our highest-converting audiences. 
ROAS looks amazing because we're cherry-picking easy wins.

CEO: But we're being more efficient.

CMO: We're being more efficient at reaching fewer people. Our total customer volume is down because we've narrowed our focus so much.

CEO: So high ROAS is bad?

CMO: Not always, but ROAS alone doesn't always tell the whole story and if we optimize to just ROAS, we might be missing the picture. 

CEO: What should we be optimizing for?

CMO: Well like most things in Marketing, it's triangulating across a few pictures:
1. Total new customers
2. ROAS (ideally incremental if we have it)
3. LTV / CAC 

CEO: Why do all the dashboards show ROAS then?

CMO: Because platforms want you focused on efficiency, not growth. It makes their reporting look good even when your business isn't growing. ROAS is useful for comparing high level channel efficiency, but it shouldn't be the only goal.

CEO: How do we fix this?

CMO: Stop celebrating ROAS improvements in isolation. Always pair it with growth metrics. If ROAS is up but then new customer revenue  is down, we're headed in the wrong direction. It just means we likely cut ad spend or perhaps focused on retargeting already warm leads. 

CEO: That's counterintuitive.

CMO: Most good marketing metrics are. 

--
🔔 Follow for Marketing Strategy & Science 
♻️ Repost because sharing is caring
⬇️ Save it so you don't lose it

CMO: Test showed zero lift. Let's cut the channel.

Me: Hold on. Let me see the test setup first.

CMO: We turned off the ads for a month. Revenue didn't change. Seems pretty clear.

Me: Did you account for delayed conversions?

CMO: What do you mean?

Me: Buyers don't convert immediately. They hear about you, research later, maybe buy weeks after exposure.

CMO: We ran the test for a full month though.

Me: Right, but if your average time-to-conversion from exposure is 6 weeks, a one-month test won't capture the full effect.

CMO: So we should've run it longer?

Me: At least 8 weeks for a product like yours, but really 12 at the minimum. Better would be me to actually use sample sizes. 

CMO: So the test is useless?

Me: Not useless, but potentially not representative. We should also check if the impact shows up in brand search lift instead of direct conversions.

CMO: Why would it show up there?

Me: Because that channel can build awareness. People hear about you, then search for you later. If branded search went up during the test period, that's a signal ads are working even if direct conversions didn't move.

CMO: We didn't track that.

Me: Then we're not capturing the full impact. Let me redesign this test with the right metrics and timeframe before we kill the channel.

CMO: How much will that cost us?

Me: Way less than cutting a working channel because we measured it wrong. 

–
👋 I’m a former Uber Head of Marketing Science where I optimized $1 Billion across Brand, Performance, and Lifecycle Marketing

🔔 Follow for Marketing Strategy & Science 
♻️ Repost because sharing is caring
⬇️ Save to send to the C-suite later

CEO: Just saw that CAC dropped 25%! Nice!

CMO: Meh, I am not happy. 

CEO: Why? Lower CAC is always better.

CMO: Not if we're acquiring fewer customers overall.

CEO: Wait, are we?

CMO: New customer volume is down 15% versus last quarter.

CEO: But our CAC improved.

CMO: Right. We got more efficient at a smaller pool. It's like bragging about catching fish faster while fishing in a shrinking pond.

CEO: I don't understand. How is lower CAC bad?

CMO: CAC is an efficiency metric. It tells you how well you're converting. It doesn't tell you if you're growing.

CEO: So what should I be looking at?

CMO: CAC and total new customers. I want new customer growth to outpace CAC growth. If CAC drops but we're acquiring fewer customers, we're optimizing ourselves into irrelevance.

CEO: What's driving this?

CMO: We tightened our targeting so much that we're only reaching people ready to buy right now. We're ignoring the 95% of the market that isn't in-market yet.

CEO: But those people have terrible conversion rates.

CMO: Today, yes. But if we don't reach them now, our competitors will. And when they're ready to buy, we won't be on their radar.

CEO: So, what are we going to do?

CMO: We need to balance efficiency and reach. Some campaigns for high-intent, low CAC conversions. Others for building future pipeline, even if CAC looks worse.

CEO: How do other companies think about this?

CMO: The smart ones track both CAC and total customer growth. If both are improving, you're winning. If only CAC improves while volume drops, you're in trouble.

CEO: Got it. Thanks for explaining that. I feel much better. 

CMO: Good. Now stop reading reports and slacking me at 2:30 AM.

--
🔔 Follow me for Marketing Strategy & Science 
♻️ Repost because sharing is caring
⬇️ Save it so you don't lose it

CEO: We need to be sending more emails! 

New CMO: Why? What data are you seeing?

CEO: Email open rates correlate perfectly with revenue.

CMO: I believe you, but that doesn't mean email is driving revenue.

CEO: What else would be driving it?

CMO: Could be anything. Seasonality, product improvements, word of mouth, other marketing channels.

CEO: But they move together. That has to mean something.

CMO: Likely, but that might not be what's happening. 

CEO: So what's actually happening with email?

CMO: My guess? When people are already interested in buying, they open more emails. Email isn't creating demand. It's capturing attention from people who already have intent.

CEO: How do we know for sure?

CMO: Run a holdout test. Stop sending email to a random part of the audience for a month. Then we can split it by segment and If revenue drops in that segment, email is causal. If it doesn't, it's just correlated.

CEO: What if revenue does drop?

CMO: Then we know email matters and we can invest more confidently.

CEO: And if it doesn't drop?

CMO: Then we just saved a ton of money on a channel that wasn't actually working. We can reinvest those resources into more testing and be more aggressive at iterating on the campaigns we thought were working. 

CEO: This seems like a lot of work to validate something that looks obvious.

CMO: The expensive mistakes are the ones that look obvious but aren't true. Would you rather spend $500K scaling email based on a correlation or spend $20K on a test to know for sure?

CEO: When you put it that way, the test seems cheap.

CMO: It is cheap. Way cheaper than scaling the wrong thing. 

–
👋 I’m a former Uber Head of Marketing Science where I optimized $1Bn+ across Brand, Performance, and Lifecycle Marketing

🔔 Follow for Marketing Strategy & Science 
♻️ Repost because sharing is caring
⬇️ Save to send to your CEO later

CEO: What's our marketing efficiency?

CMO: What do you mean by efficiency?

CEO: You know, how well is marketing performing?

CMO: That depends. Are you asking about ROAS? CAC payback? LTV ratios?

CEO: Aren't those all the same thing?

CMO: Not even close. They measure completely different aspects of marketing performance.

CEO: Can't you just give me one number that tells me if marketing is efficient?

CMO: I could, but it would be meaningless.

CEO: Why?

CMO: Because different channels work on different timelines. Performance marketing pays back in weeks. Brand takes months. Combining them into one number hides everything useful.

CEO: So what should I be asking for?

CMO: Depends on what decision we're trying to make.

CEO: I want to know if we should increase marketing budget.

CMO: Then you need incremental ROAS. 

CEO: Why not use Blended ROAS? 

CMO: Blended ROAS might be 5x. Looks great. But if we're maxed out on our best channels, incremental ROAS might only be 2x. Adding budget would actually hurt efficiency.

CEO: So I can't just ask "what's our marketing efficiency?"

CMO: You can. But I need to know what you're trying to decide first. Then I'll give you the right metric.

CEO: This is way more complicated than I thought.

CMO: Marketing measurement is. That's why one-number answers are usually wrong. 

–
👋 I’m a former Uber Head of Marketing Science where I optimized $1Bn+ across Brand, Performance, and Lifecycle Marketing

🔔 Follow for Marketing Strategy & Science 
♻️ Repost because sharing is caring
⬇️ Save to send to your CEO later

CEO: Finance says you drove $2M. You say $2.3M?

CMO: We're both right. Here's why...

CEO: It's revenue. How can there be different #s?

CMO: Finance measures when cash hits the bank. I measure when the customer converts.

CEO: Why does that matter?

CMO: I bring in signups, and often they buy the same day. But sometimes, they buy that week and sometimes a few week laters, but the revenue is attributed to my signups.

CEO: Can't you just align on one definition?

CMO: We could, but then one of us is optimizing for the wrong thing.

CEO: Explain.

CMO: If I optimize for recognized revenue, I'll ignore user behaviors where they are still valuable but might purchase later. We'd be optimizing towards same day purchase events. 

CEO: And if Finance uses your number?

CMO: They count based on dates so signups that happened this week but haven't shown revenue yet will not count for them. Their job is different than mine.

CEO: So we just accept having two different numbers?

CMO: We accept that different functions need different metrics. The key is we both understand what the other is measuring and why. 
--
🔔 Follow me for Marketing Strategy & Science 
♻️ Repost because sharing is caring
⬇️ Save it so you don't lose it

CEO: AI is my #1 priority!

New CMO: Yeah... it's like my 7th...

CEO: What?! We need to be AI first and fast! 

CMO: Here's whats more important for me :
1. Our Marketing data is an insane mess
2. We have ZERO organic social presence
3. Our landing pages tell a disjointed story
4. I have no idea what channels are incremental
5. Our users don't know about half of what we do
6. Oh and retention is 30% below where it should be

CEO: Okay, great, but our competitors are using AI! 

CMO: Let them. I know for a fact they're building on a foundation of sh*t. It'll come tumbling down. I promise you that I will make AI my H2 priority but you need to let me focus on H1 on getting things right. And that's an accelerated timeline. I'm not even adding the fact that I just joined and need to ramp up.

CEO: Okay, I see your 6 months timeline and I raise you 2.

CMO: No! Give me 6 months. Let me do what you hired me for.

CEO: Okay, 4 months?!

CMO: ** closes Zoom meeting **

–
👋 I’m a former Uber Head of Marketing Science where I optimized $1Bn+ across Brand, Performance, and Lifecycle Marketing

🔔 Follow for Marketing Strategy & Science 
♻️ Repost because sharing is caring
⬇️ Save to send to the C-suite later

CEO: Do we need an MMM?

CMO: That's not the question we should be asking...

CEO: Ohhh, what is?!

CMO: Are we ready for one?

CEO: What do you mean?

CMO: First, an MMM is not a holy grail. It's one part of a holistic measurement strategy.

CEO: That doesn't sound so bad...

CMO: Agree. The problem is that it might tell us something we weren't expecting. That channel A isn't contributing as much. 

CEO: Okay... still not seeing an issue...

CMO: Will you let me finish!

CEO: Haha oops... too excited.

CMO: Okay, so it might tell us something we don't know and we have to be prepared to then test that and validate it. That could mean we turn off 50% or even 100% of our "best" channel.

CEO: Oh boy... really?!

CMO: Yes. That's why the question we should be asking is "Are we ready for one?". Are we culturally and operationally ready to act on one or are you just asking because you heard that's what we need.

CEO: Definitely the latter, but this is really helpful.

CMO: Good. Don't worry, you'll ask me this 10x more times and I'll say the same thing 😞 

--
🔔 Follow me for Marketing Strategy & Science 
♻️ Repost because sharing is caring
⬇️ Save it so you don't lose it

Marketing alone should NOT own CAC. 

That's the equivalent of saying:

- Product alone owns LTV 
- Ops alone owns conversion rate
- Finance alone owns payback period 

Sure, it's easy to silo these metrics and "assign" an owner but we do that to create an illusion of control. 

The only time CAC should be considered a Marketing metric is when all of the levers to drive CAC fall under Marketing.

This means:
1. Activation experience
2. Landing pages
3. Signup flow
4. Pricing

Otherwise, Marketing can really only influence CPM, CTR, and traffic which hamstrings their impact and influence. 

There's no perfect solution but the best orgs understand 3 things:

1. The customer doesn't care who owns what
2. You can't keep top talent this way 
3. Silos just lead to inefficiencies

It's why so many Marketers get frustrated when they're asked to reduce CAC but hit a ceiling in terms of what they can deliver.

If you want Marketing to own a metric, give them ownership.

It's pretty straightforward. 

–
👋 I'm a former Uber Head of Marketing Science where I optimized $1Bn+ across Brand, Performance, and Lifecycle Marketing

🔔 Follow for Marketing Strategy & Science
♻️ Repost because sharing is caring
⬇️ Save to send to the C-suite later

My next guest raised $6M to fix Marketing Measurement after being frustrated with the tools he had available at Dropbox, Pilot, and Adobe 🤑  

Pranav Piyush, CEO & Founder at Paramark (and Reforge's course instructor on Marketing Measurement) joined experiMENTAL Newsletter & Podcast to break down why marketing feels like a "foreign language" to the C-suite and how to fix it.

We chatted in depth about:
⮑ Why MMM is now table takes
⮑ How to make marketing metrics CFO-friendly 
⮑ Why brand and performance teams are dysfunctional 

My favorite discussions:
⭐ Most marketing leaders feel anxious before board meetings because there's a massive communication gap between CMOs and the C-suite. It feels like they're speaking a different language.

⭐ If you're doing $10-50M in ads stop wasting time on holdout tests. You likely don't have enough data and should focus on pulse-up testing and new channels. 

⭐ Pranav wishes he could remove all decimals from marketing dashboards. Saying "17% incrementality" creates false precision. The reality is that It's probably somewhere between 13-19%, and that's okay.

If you're a marketing leader struggling to prove ROI to your CFO or CEO, or you're tired of attribution tools that don't actually help you make better decisions, this conversation is for you!

Pranav brings an engineer's rigor and a marketer's empathy to one of the most broken parts of modern marketing. 

Tune in wherever you love listening to podcasts using the links in the comment!

CEO: Are we measuring ROI of our Super Bowl Ads?

CMO: No, for 3 important reasons.

CEO: Oh boy, I'm nervous. 

CMO: Haha I figured, but trust that it's not so bad:
1. We can't measure ROI in a traditional sense

It's a national TV campaign and the US is our largest market so there's no real opportunity of a holdout

2. If we try to measure business impact around the time of our ad, we'll be concentrating all the "impact" into a short window and it'll make the ROI look poor

3. All other proxy metrics like search volume, traffic, etc. are going to be metrics you think are "soft" and don't represent business value.

CEO: Okay all good points, but we're spending a sh*tload

CMO: Yes, I agree, and to be clear there is an ROI. I just won't see it for months and so I can't share the ROI with you.

CEO: Okay, fair. What ROI are you expecting.

CMO: Great question. Let's get into this!

1. It's a launch of our new product. We're getting in front of hundreds of millions of our target demographic. Many of them won't be ready to buy today but they will be in a few months and we'll be there because...

2. It's the launch of our new campaign which is rolling out across 8 different channels over the next few months. Everyone is going to see us everywhere!

3. It's a competitive bet too. Our competitors are also showing up and obviously I think our team is better than theirs so people will compare us and we're going to win. Plus if we don't show up, that's just another void they'll fill 

For the campaigns we do rollout over the next couple months, we will be using our standard measurement framework of Attribution + Incrementality tests + MMM to triangulate ROI. 

Overall, there's an ROI and I believe it's huge but it'll take time to materialize

CEO: Okay. We have trusted you this far and you have not led us astray. Forgive me. I was wrong to despair.

CMO: Did you just quote Lord of the Rings?

CEO: Did we just become besties?!

–
👋 I’m a former Uber Head of Marketing Science where I optimized $1Bn+ across Brand, Performance, and Lifecycle Marketing

🔔 Follow for Marketing Strategy & Science 
♻️ Repost because sharing is caring
⬇️ Save to send to your CEO later

CEO: CAC keeps climbing. Should we shift to brand?

CMO: It's not an either/or decision.

CEO: But, we have a fixed budget. If we spend more on brand, we need to spend less on performance.

CMO: True. But the real tradeoff isn't "performance" versus "brand". We need to decide on whether we prioritize today's revenue or tomorrow's efficiency.

CEO: Explain.

CMO: Performance marketing captures current demand. Brand Marketing creates it and at higher volumes. When we only do Performance, we're competing for the same pool of high-intent customers as everyone else.

CEO: Why is that bad?

CMO: Well, you run out of high intent customers eventually. By layering in Brand, we increase the pool of higher intent customers.

CEO: But brand takes FOREVER to show results.

CMO: Six to twelve months to show business results but much sooner to show that we're moving in the right direction. But once it kicks in, our performance channels get more efficient without changing anything.

CEO: So Brand makes performance cheaper?

CMO: Exactly. Strong brand means higher consideration, which means more people searching for you, which means lower CPCs and better conversion rates.

CEO: So we need both?

CMO: You need both working together. Performance capitalizes on brand equity. Brand reduces performance costs over time. They're complementary, not competitive.

CEO: How do other companies balance this?

CMO: The good ones don't see it as balance. They see it as sequencing and investing. 
Scale performance → Build Brand → Scale performance → Build Brand in ongoing loop. 

Eventually there's a cap, but we're not even close. 

But by then we'll have much better unit economics.

A rare few can survive on just performance marketing but they'll have insane AOVs and LTV / CACs which is not realistic. Also, they're in a market where they might not be looking to win market share and are happy with their current growth rates. 

CEO: So, we don't want to make the same mistake most companies do? 

CMO: Right. They wait until performance stops working to start Brand Marketing. By then, they're desperate and impatient. Brand Marketing needs time. We should start before we need it. 

--
🔔 Follow Sundar for Marketing Strategy & Science
👉 Join 2.6K+ Marketers learning how to measure and scale ROI: https://lnkd.in/gxR5c4Xd

CMO: Hey, engagement is up but revenue is flat. 

Me: What's your hypothesis on why?

CMO: Honestly? I'm stumped. The funnel looks healthy. Everything we're tracking is improving.

Me: Can I ask you a different question? What drives someone to purchase from you?

CMO: Product quality, pricing, trust in the brand. The usual stuff.

Me: Right. Now, how closely do your engagement metrics correlate with any of those?

CMO: Hmmm... not sure. 

Me: That's the issue. We're measuring what's easy to track, not what predicts outcomes.

CMO: So what should we be doing differently?

Me: Start by working backwards. What's the customer action that leads directly to revenue for you?

CMO: Completing their first transaction. Once someone buys our retention is strong.

Me: Perfect. So what predicts first transaction? Not correlates. Predicts. 

CMO: How quickly they experience value in the product. And whether they trust us enough to enter payment info.

Me: Now we're getting somewhere. Are any of your current metrics measuring speed to value or trust?

CMO: No. We're measuring clicks and time spent.

Me: Which is why engagement is up but revenue isn't. Those metrics aren't connected to what drives purchasing behavior.

CMO: This makes sense, but how do we figure out what actually matters?

Me: If you've got the data, then look at it and see what it says. If not, then another test you can do is change your optimization metric. Play around with what happens if you use X instead of Y to optimize. 

CMO: This is going to be uncomfortable. The team loves celebrating engagement wins.

Me: They'll love celebrating revenue wins more. And once you show them the connection between the right metrics and business impact, they'll never go back.

CMO: Alright. Let's do this. Where do we start?

Me: Make a list of every metric you track. Then next to each one, write down why you believe it predicts revenue. If you can't articulate it, cut it.

--
🔔 Follow me for Marketing Strategy & Science 
♻️ Repost because sharing is caring
⬇️ Save so you don't lose it

In 2025, I wrote 38 articles on Marketing. 

Here are my top 10 by open rate (60%+):

Oh and they're all free to read 😃 

1. How to write a measurement brief 
https://lnkd.in/gPrr-tAF

2. How I build Uber’s Brand Science Team
https://lnkd.in/evJDr2wc

3. Why analyzing Marketing feels impossible
https://lnkd.in/gqNagEBr

4. How to calculate LTV
https://lnkd.in/gxUqcxcw

5. 7 Brutal truths about CAC
https://lnkd.in/gVjh-YRK

6. How to measure when you can’t A/B test
https://lnkd.in/gcHsSpqy

7. How to build better Marketing targets
https://lnkd.in/gmH8VV8T

8. How to create more impactful analyses
https://lnkd.in/esmGEWmk

9. Why startups shouldn’t spend a $ on Paid Ads
https://lnkd.in/e8kFYYU5

10. How Data Scientist should use AI
https://lnkd.in/expdRfFZ

Let me know which one was your favorite!

Every startup at every stage needs a data leader, but the engagement model evolves:

Pre-product market fit → Advisor

Especially if you're a marketplace or E-com, they can help you with what to analyze, track, and prioritize.  

Even just an hour or two a week is huge! 

Scaling → fractional head of data 

Assuming you have 1-4 data folks (engineers, analysts etc), a Fractional data leader can be a game changer. 

You’ll need someone to mentor your data team and help the executive team scale. 

Scaling is about building systems and having the right data leader will help you build that.

Beyond → Coach / Consultant

Once you’re bigger, you’ll likely have a full time Head of Data but the truth is they’ll need help. 

Going from scale to “corporate” has different demands and most data leaders aren’t prepared for it. 

Bring in advisors and coaches to help your leader scale. 

Everything from how to ask for budget to how to promote people are fundamental skills that can be coached. 

Follow Sundar for Marketing Strategy & Science

li_at

JSESSIONID

LinkedIn App Challenge

Talks about #solopreneurship and #experimentation

CEO: I need to present marketing ROI to the board. 

CMO: How are you defining ROI? 

CEO: The ROI. How much revenue are we generating per dollar spent?

CMO: On which channels? Which timeframe? Are we including brand or just performance?

CEO: All of it. Just give me one number.

CMO: I can't do that responsibly.

CEO: Why not? Finance gives me one number for profitability.

CMO: Did Finance give you the ROI for Product?

CEO: Well... no...

CMO: Okay, exactly. My job is to help you here but it's important that we're aligned on the right story and narrative. Storytelling is kind of my job. What does the board want?

CEO: The board wants to know if Marketing is working.

CMO: That's not specific enough. Are they asking if we should increase budget? Cut budget? Shift budget somewhere else?

CEO: Probably whether we should increase it.

CMO: Then you'll want to show incremental ROI not blended ROI.

CEO: Do we know our incremental ROI?

CMO: On performance channels, yes. On our Brand Marketing efforts, we'd need to run tests to find out.

CEO: The board isn't going to like "we need to run tests."

CMO: Then we give them blended ROI with a massive caveat that it doesn't predict what happens if we scale. Their choice.

CEO: Why is this so complicated?

CMO: Because you're asking for simple answers to complex questions. Marketing isn't accounting. 

CEO: What do other CMOs do?

CMO: The good ones educate their board on what's measurable versus what's theater. The desperate ones give clean numbers that look good but don't mean anything.

CEO: I don't want to be desperate. Help me educate them. 

CMO: Let's set up a session tomorrow so I can help you understand it better first. Also, read some of the articles from the newsletter that I DM'ed you about. 

CEO: experiMENTAL? Done. 

--
🔔 Follow me for Marketing Strategy & Science 
♻️ Repost because sharing is caring
⬇️ Save it so you don't lose it

I love Data. And I love writing. I love writing about Data.
Open for brand collab.

(still figuring out the rest)

Amney Mounir

I went through this cycle too many times in my career.

1. You get excited about this new project.
2. Then you find out that the data is so bad.
3. You work your way to make some fixes.
4. You then like: "Hmm, it's not THAT bad".
5. You find some pattern & share them to the team.
6. And then nothing happens.

This is why YOU always need to 𝗺𝗮𝗿𝗸𝗲𝘁 your analysis.

You can do that by:
- Sharing your work more broadly (email or slack)
- Building relationship to have ppl spreading the word
- Having clear next steps/recommendations

So that all the work you do have: ACTUAL IMPACT.
And doesn't get thrown to the trash.

———
Btw Sai Kumar Bysani (200k+ followers) is hosting a FREE workshop on how to ACE your next data interview, that you should absolutely join!

📕 Data Workshop: https://lnkd.in/gbR_mCJW

———
♻️ Repost this if you found it insightful
👋 Follow Amney for more Daily Tips

Most Data Analysts fail interviews for the wrong reasons.

Not because they don’t know SQL.
Not because they can’t build charts.
Not because they’re “not technical enough”.

But because 𝘁𝗵𝗲𝘆 𝗱𝗼𝗻’𝘁 𝘂𝗻𝗱𝗲𝗿𝘀𝘁𝗮𝗻𝗱 𝘁𝗵𝗲 𝗶𝗻𝘁𝗲𝗿𝘃𝗶𝗲𝘄 𝗴𝗮𝗺𝗲!

I put together this Data Analyst interview process
That breaks down every step:

1) What each round is really testing
2) How to prepare for each one
3) What interviewers actually care about

If you’re preparing for interviews in 2026,
𝘔𝘢𝘬𝘦 𝘴𝘶𝘳𝘦 𝘵𝘰 𝘳𝘦𝘱𝘰𝘴𝘵 𝘢𝘯𝘥 𝘴𝘢𝘷𝘦 𝘵𝘩𝘪𝘴 𝘱𝘰𝘴𝘵!

———

♻️ Repost this if you found it useful!
👋 Follow Amney for more Daily Tips

When I worked as the 𝘰𝘯𝘭𝘺 Data Analyst, I hit a ceiling fast.

I was the "smartest" data person in the room, but that was actually a problem. I had no Lead Analyst to review my code. No mentor to challenge my assumptions. No one to help me level up.

So I started studying what other companies were doing.

Here are my top 5 technical case studies that leveled me up:

𝟭. 𝗧𝗵𝗲 𝗣𝗼𝘄𝗲𝗿 𝗼𝗳 𝗮 𝗣𝗶𝗰𝘁𝘂𝗿𝗲 𝗮𝘁 𝗡𝗲𝘁𝗳𝗹𝗶𝘅: 
https://lnkd.in/gvHY9w-n

𝟮. 𝗣𝗿𝗲𝗱𝗶𝗰𝘁𝗶𝗻𝗴 𝗮 𝗟𝗶𝘀𝘁𝗶𝗻𝗴'𝘀 𝗣𝗿𝗶𝗰𝗲 𝗮𝘁 𝗔𝗶𝗿𝗯𝗻𝗯: 
https://lnkd.in/gRQmyWw7

𝟯. 𝗙𝗼𝗿𝗲𝗰𝗮𝘀𝘁𝗶𝗻𝗴 𝗗𝗲𝗺𝗮𝗻𝗱 𝗮𝘁 𝗨𝗯𝗲𝗿: https://lnkd.in/gPegbSxY

𝟰. 𝗛𝗼𝘄 𝗗𝗮𝘁𝗮 𝗦𝗰𝗶𝗲𝗻𝘁𝗶𝘀𝘁𝘀 𝗟𝗲𝗮𝗱 𝗮𝘁 𝗠𝗲𝘁𝗮: https://lnkd.in/gFafCDi4

𝟱. 𝗨𝘀𝗶𝗻𝗴 𝗣𝗿𝗼𝗽𝗲𝗻𝘀𝗶𝘁𝘆 𝗦𝗰𝗼𝗿𝗲 𝗮𝘁 𝗦𝗽𝗼𝘁𝗶𝗳𝘆: https://lnkd.in/g4wZZuHA

———

♻️ Repost this if you found it useful! 
👋 Follow me, Amney for more Daily Tips

Prompting will become a major skill for Data Analysts.

Not SQL.
Not Python.
Not building dashboard.

𝗣𝗿𝗼𝗺𝗽𝘁𝗶𝗻𝗴.

I don’t believe AI will replace Data Analysts.
But Data Analysts with poor prompting skills will be left behind.
𝘣𝘺 𝘵𝘩𝘰𝘴𝘦 𝘸𝘩𝘰 𝘬𝘯𝘰𝘸 𝘩𝘰𝘸 𝘵𝘰 𝘶𝘴𝘦 𝘪𝘵 𝘸𝘦𝘭𝘭.

𝗜 𝗽𝘂𝘁 𝘁𝗼𝗴𝗲𝘁𝗵𝗲𝗿 𝘁𝗵𝗶𝘀 𝗔𝗜 𝗣𝗿𝗼𝗺𝗽𝘁 𝗰𝗵𝗲𝗮𝘁𝘀𝗵𝗲𝗲𝘁 𝗳𝗼𝗿 𝘆𝗼𝘂.
Bad prompts vs Great prompts.
With Real work use cases.

AI will become a core skill in 2026.
The earlier you learn how to use it properly,
the bigger your edge will be.

———

♻️ Repost this if you found it useful! 
👋 Follow Amney for more Daily Tips

There is NO excuse in 2026 to say "I have no experience".

Here are 7 volunteering websites you can gain experience as a Data Analyst 👇:

1) VolunteerMatch
2) Catchafire
3) Techfleet 
4) Datakind
5) Statistics without Borders
6) United Nations Volunteers
7) Code for America

𝘠𝘦𝘴, 𝘪𝘵'𝘴 𝘶𝘯𝘱𝘢𝘪𝘥 𝘣𝘶𝘵...
Volunteering experience does 3 great things:

1) Gives you purpose while you wait
2) Keeps your skills sharp and relevant
3) Adds real experience to your resume

And that will get you closer to 𝘆𝗼𝘂𝗿 𝗳𝗶𝗿𝘀𝘁 𝘆𝗲𝘀!

———

♻️ Repost this if you found it useful!
👋 Follow Amney for more Daily Tips

SQL is 𝘴𝘵𝘪𝘭𝘭 the king of data in 2026.

People keep saying “SQL is dead" but the reality is:

- Every interview still has a SQL round
- SQL is still the main tool to pull data
- SQL is still the most reliable language
- SQL isn’t a “nice-to-have", it's a must
- SQL is still good to know even with AI

Here are few resources for you to get started:
📗 Free Workshop: https://lnkd.in/gbR_mCJW
📕 Interview Book: https://lnkd.in/ei8Sp6Qw
📘 Roadmap: https://lnkd.in/gVaKx-tv

———
♻️ Repost this if you find it insightful!
👋 Follow Amney for more Daily Tips

Here is 𝘮𝘺 roadmap to master SQL in 14 days:

When I first started learning, I got stuck 
Because I didn't know where to start
What I needed was a clear path.

𝗗𝗮𝘆𝘀 𝟭–𝟰: 𝗕𝘂𝗶𝗹𝗱 𝘆𝗼𝘂𝗿 𝗦𝗤𝗟 𝗙𝗼𝘂𝗻𝗱𝗮𝘁𝗶𝗼𝗻𝘀

- Write your first queries: SELECT, FROM, WHERE
- Filter with LIKE, IN, BETWEEN, IS NULL
- Use ORDER BY, LIMIT, OFFSET
- Use Aggregate functions

𝗗𝗮𝘆𝘀 𝟱–𝟳: 𝗟𝗲𝗮𝗿𝗻 𝘁𝗼 𝘂𝘀𝗲 𝗝𝗢𝗜𝗡𝗦

- Start by learning INNER and LEFT JOINs
- Then go for FULL OUTER, SELF JOINS
- As a Bonus, learn CROSS joins

𝗗𝗮𝘆𝘀 𝟴-𝟭𝟯: 𝗟𝗲𝘃𝗲𝗹 𝘂𝗽

- Use CASE statements
- CTEs and Subqueries
- Window Functions
- Text Analysis

Day 14 and beyond: Practice!

Here are some resources for you to get started:

📗 Data Workshop: https://lnkd.in/gbR_mCJW
📕 SQL Questions: https://lnkd.in/g62bfHF6

———
♻️ Repost this if you found it insightful!
👋 Follow Amney for more Daily Tips

Please stop taking 1000s courses if you’re still entry level. 

Companies that wanna hire you 𝘳𝘦𝘢𝘭𝘭𝘺 don't care. 

They care about: 𝘞𝘩𝘢𝘵 𝘺𝘰𝘶’𝘷𝘦 𝘢𝘤𝘵𝘶𝘢𝘭𝘭𝘺 𝘣𝘶𝘪𝘭𝘵.

Here are 5 projects to make your portfolio stand out 👇

1. 𝐄-𝐜𝐨𝐦𝐦𝐞𝐫𝐜𝐞 𝐅𝐮𝐧𝐧𝐞𝐥 𝐀𝐧𝐚𝐥𝐲𝐬𝐢𝐬

 Track how users move: view → cart → purchase.
 Spot drop-offs. Recommend fixes.

Link: https://lnkd.in/eDcHXdy2

2. 𝐀/𝐁 𝐓𝐞𝐬𝐭 𝐄𝐯𝐚𝐥𝐮𝐚𝐭𝐢𝐨𝐧
Run a real Experiment.
 Check if the change worked. Back it with stats.

Link: https://lnkd.in/eZ2ZDr7z

3. 𝐂𝐨𝐡𝐨𝐫𝐭 & 𝐑𝐞𝐭𝐞𝐧𝐭𝐢𝐨𝐧 𝐀𝐧𝐚𝐥𝐲𝐬𝐢𝐬
 See who sticks around.
 Plot retention by month or region.

Link: https://lnkd.in/ePFvMKkx

4. 𝐊𝐏𝐈 𝐃𝐚𝐬𝐡𝐛𝐨𝐚𝐫𝐝
Build a live dashboard: Sales, Profit, Users.
Add filters

Link: https://lnkd.in/e4ncGEf5

5. 𝐀𝐢𝐫𝐛𝐧𝐛 𝐂𝐚𝐬𝐞 𝐒𝐭𝐮𝐝𝐲
 Where should Airbnb expand?
 Analyze price, reviews, and trends.

Link: https://lnkd.in/etiyDarf

———

♻️ Repost this if you found it useful!
👋 Follow Amney for more Daily Tips

Don't limit your job search to Data Analyst in 2026.

To increase your chances to land your next role, I highly recommend to expand your search to more Data Analyst roles.

There are so many, and here are a few for your reference:

𝐁𝐮𝐬𝐢𝐧𝐞𝐬𝐬 𝐅𝐨𝐜𝐮𝐬𝐞𝐝 𝐀𝐧𝐚𝐥𝐲𝐬𝐭 𝐑𝐨𝐥𝐞𝐬:
↳ Customer Insight Analyst
↳ Market Research Analyst
↳ Operation Analyst
↳ Business Analyst
↳ Strategy Analyst

𝐑𝐞𝐩𝐨𝐫𝐭𝐢𝐧𝐠 𝐀𝐧𝐚𝐥𝐲𝐬𝐭 𝐑𝐨𝐥𝐞𝐬:
↳ Analytics Associate
↳ Analytics Engineer
↳ Reporting Analyst
↳ BI Engineer
↳ BI Analyst

𝐏𝐫𝐨𝐝𝐮𝐜𝐭 & 𝐄𝐱𝐩𝐞𝐫𝐢𝐦𝐞𝐧𝐭𝐚𝐭𝐢𝐨𝐧 𝐑𝐨𝐥𝐞𝐬:
↳ Causal Inference Analyst
↳ Data Scientist, Analytics
↳ Experimentation Analyst
↳ Product Analyst
↳ Growth Analyst

𝐀𝐝𝐯𝐚𝐧𝐜𝐞𝐝 𝐀𝐧𝐚𝐥𝐲𝐭𝐢𝐜𝐬 & 𝐌𝐋 𝐑𝐨𝐥𝐞𝐬:
↳ Machine Learning Engineer
↳ Forecasting Analyst
↳ Research Scientist
↳ Decision Scientist
↳ ML/AI Analyst

———

♻️ Repost (please) if you found it useful!
👋 Follow me, Amney for more Daily Tips

Results don’t happen overnight, they take time.

Here is my Data career for the last 8 years:

- 2019: Junior Analyst in an e-commerce start-up
- 2020: Promoted to Product Analyst
- 2021: Joined a start-up as a Founding Analyst
- 2021 (3 months later): Laid off from that company
- 2021: Joined Poshmark as Sr. Product Analyst 
- 2023: Promoted to Lead Data Analyst
- 2024: Joined Meta as a Product Growth Analyst

And I don't even know what will happen in 2026.
And even beyond that.

But when I take a step back, I realize I've done a lot.

So be consistent in what you do, always take a step back to realize and enjoy all the great things you've accomplished.

And keep pushing for more! 

———

♻️ Repost this if you found it insightful!
👋 Follow Amney for more Daily Tips

The life of a Data Analyst looks like this every day.

“Can you pull this really quick?”
“Can you fix this dashboard?”
“Can you split it one more way?”

Early in my career, 𝗜 𝘀𝗮𝗶𝗱 𝘆𝗲𝘀 𝘁𝗼 𝗲𝘃𝗲𝗿𝘆𝘁𝗵𝗶𝗻𝗴.

What I didn’t realize:
Saying yes to everything often means spending time on work that doesn’t actually matter.

The best analysts aren’t the ones who answer more requests.

They’re the ones who push back and learn to say NO.

———

♻️ Repost this to your friends if you agree
👋 Follow Amney for more Daily Tips

Here are 4 things I'd master to become a Data Scientist:
𝘙𝘦𝘢𝘥 𝘶𝘯𝘵𝘪𝘭 𝘵𝘩𝘦 𝘦𝘯𝘥, 𝘴𝘰𝘮𝘦𝘵𝘩𝘪𝘯𝘨 𝘤𝘰𝘰𝘭 𝘵𝘰 𝘴𝘩𝘢𝘳𝘦...

𝟭/ 𝗣𝗿𝗼𝗱𝘂𝗰𝘁 𝗦𝗲𝗻𝘀𝗲

- Metric Definition (North Star vs.Guardrail Metrics)
- Diagnosing why a metric dropped 
- Trade-off (𝘸𝘰𝘶𝘭𝘥 𝘺𝘰𝘶 𝘭𝘢𝘶𝘯𝘤𝘩 𝘪𝘧 𝘹 𝘨𝘰𝘦𝘴 𝘶𝘱 𝘣𝘶𝘵 𝘺 𝘨𝘰𝘦𝘴 𝘥𝘰𝘸𝘯)

𝟮/ 𝗧𝗲𝗰𝗵𝗻𝗶𝗰𝗮𝗹 𝗦𝗸𝗶𝗹𝗹𝘀 

- SQL (Joins, Self-Joins, Window Functions)
- Pandas (aggregations, cleaning, merging)

𝟯/ 𝗘𝘅𝗽𝗲𝗿𝗶𝗺𝗲𝗻𝘁𝗮𝘁𝗶𝗼𝗻 

- A/B Testing (Sample size, Power analysis)
- Causal Inferences (Propensity Score, Diff in diff)

𝟰/ 𝗕𝗲𝗵𝗮𝘃𝗶𝗼𝗿𝗮𝗹 

- Using STAR Method
- Storytelling (explain results to any audience)

Btw, Banani, 𝗗𝗦 𝗠𝗮𝗻𝗮𝗴𝗲𝗿 𝗮𝘁 𝗪𝗮𝗹𝗺𝗮𝗿𝘁, 𝗶𝘀 𝗴𝗼𝗶𝗻𝗴 𝗹𝗶𝘃𝗲 𝘁𝗼 𝗰𝗼𝘃𝗲𝗿 𝗲𝘃𝗲𝗿𝘆𝘁𝗵𝗶𝗻𝗴 𝘆𝗼𝘂 𝗻𝗲𝗲𝗱 𝘁𝗼 𝗸𝗻𝗼𝘄 𝘁𝗼 𝗺𝗮𝘀𝘁𝗲𝗿 𝘆𝗼𝘂𝗿 𝗻𝗲𝘅𝘁 𝗘𝘅𝗽𝗲𝗿𝗶𝗺𝗲𝗻𝘁𝗮𝘁𝗶𝗼𝗻 𝗶𝗻𝘁𝗲𝗿𝘃𝗶𝗲𝘄. 

Make sure join, 𝘪𝘵'𝘴 𝘧𝘳𝘦𝘦 𝘢𝘯𝘥 100% 𝘸𝘰𝘳𝘵𝘩 𝘪𝘵.
📕 A/B Test Workshop: https://luma.com/t5xh23i3

———

♻️ Repost this if you found it useful!
👋 Follow Amney for more Daily Tips

Most Data Analysts prepare for the wrong things in 2026.

So I tested something cool to learn about it:

I used 𝗖𝗼𝘂𝗻𝘁, an AI Agent to analyze 10k Data Analyst job postings from 2025 and 2026. 

Here’s what actually stood out:

𝟭) 𝗦𝗤𝗟 𝗶𝘀 𝘀𝘁𝗶𝗹𝗹 𝗞𝗜𝗡𝗚
It appears in 77.8% of Data Analyst roles.
(𝘠𝘦𝘴, 𝘴𝘵𝘪𝘭𝘭 𝘮𝘰𝘳𝘦 𝘪𝘮𝘱𝘰𝘳𝘵𝘢𝘯𝘵 𝘵𝘩𝘢𝘯 𝘗𝘺𝘵𝘩𝘰𝘯.)

𝟮) 𝗦𝗤𝗟 + 𝗣𝘆𝘁𝗵𝗼𝗻 𝗶𝘀 𝘁𝗵𝗲 𝗯𝗲𝘀𝘁 𝘀𝗸𝗶𝗹𝗹 𝗰𝗼𝗺𝗯𝗼
Shows up in 50% of analyst jobs.

𝟯) 𝗪𝗵𝗮𝘁 𝗿𝗲𝗮𝗹𝗹𝘆 𝗺𝗮𝗸𝗲𝘀 𝘆𝗼𝘂 𝘀𝘁𝗮𝗻𝗱 𝗼𝘂𝘁:
- Analytical thinking (87.8%)
- Team work (82.9%)
- Business acumen (72.9%)

And the cool part is I was able to analyze all these jobs,
without writing a single line of SQL.

I just:
- Uploaded a CSV
- Asked Count AI agent my questions
- Got insights + visuals in few minutes

That's the kind of work that will take me few hours as a Data Analyst to pull, and by asking the right questions, I was able to save so much time!

𝗜𝗳 𝘆𝗼𝘂 𝘄𝗮𝗻𝘁 𝘁𝗼 𝘁𝗿𝘆 𝗶𝘁 𝗳𝗼𝗿 𝗳𝗿𝗲𝗲, 𝗵𝗲𝗿𝗲’𝘀 𝘁𝗵𝗲 𝗹𝗶𝗻𝗸:
https://lnkd.in/gtj3_u-K

—

♻️ Repost if this was helpful!
👋 Follow me, Amney for more 𝗥𝗘𝗔𝗟 Data tips

Insights to become a Data Analyst using an AI agent 

This is what your Data Analytics interview looks like:

𝗦𝘁𝗲𝗽 𝟭: 𝗥𝗲𝗰𝗿𝘂𝗶𝘁𝗲𝗿 𝗖𝗮𝗹𝗹 (𝟯𝟬 𝗺𝗶𝗻)
You’ll be asked about your experience, skills, and fit.

How to prepare:
- A 30-second pitch
- 3-4 relevant stories
- Mention skills like SQL, BI tools

𝗦𝘁𝗲𝗽 𝟮: 𝗧𝗲𝗰𝗵𝗻𝗶𝗰𝗮𝗹 𝗥𝗼𝘂𝗻𝗱 (𝟭𝗵)
Your SQL will be tested, often paired with product or business questions.

How to prepare:
- SQL (Basic functions, Joins, Window functions)
- How to improve a feature
- Investigate a metric drop
- Define success metrics

𝗙𝗶𝗻𝗮𝗹 𝗥𝗼𝘂𝗻𝗱 (𝟰-𝟱𝗵)

It's a deeper version of the technical round plus behavioral questions. You can expect to talk about a you're proud of, how you handle pressure and deal with multiple tasks. 

———

♻️ Repost this if you found it insightful
👋 Follow Amney for more Daily Tips

I use less SQL at Meta than I did 6 month ago.

And I’m more 𝘪𝘮𝘱𝘢𝘤𝘵𝘧𝘶𝘭 than before.

Not because SQL doesn’t matter.
It does.

But because AI changed where my value shows up.

Today, I spend more time:
- Prompting instead of querying
- Framing the problem instead of ad-hocs
- Iterating on insights instead of debugging

AI help me handle the 𝘢𝘯𝘯𝘰𝘺𝘪𝘯𝘨 job, which is writing the SQL from scratch, instead I just take time reviewing and validating queries.

That freed up so much time time for thinking about strategic work which then drives so much impact for the team.

———

♻️ Repost this if you found it insightful
👋 Follow Amney for more Daily Tips

Interviewers don’t tell you 𝙬𝙝𝙮 you failed.

And here is why:

If companies told candidates exactly how they were evaluated, interviews would stop working.

𝗘𝘃𝗲𝗿𝘆𝗼𝗻𝗲 𝘄𝗼𝘂𝗹𝗱 𝗹𝗲𝗮𝗿𝗻 𝘁𝗵𝗲 𝗴𝗮𝗺𝗲.

And it would be much harder for them to separate strong candidates from bad ones.

So the process stays intentionally vague.

From my experience in tech, interviews usually come down to just a few things:

1) How you think through ambiguous problems
2) How clearly you communicate your thinking
3) How well you show impact from your past work

And every stage of the process has an internal scorecard, even if no one shows it to you.

Once you start preparing against that scorecard, interviews feel very different.

Much less random.
And MUCH more controllable.

———

♻️ Repost this to your friends that need this
👋 Follow Amney for more Daily Tips

How much A/B testing should a Data Analyst know?

These are the 10 questions I usually get during interviews:

1. How would you set up your A/B test?
2. What parameters to calculate the sample size?
3. Which randomization unit ID would you use?
4. How long should we run this test? 
5. How do you account for network effects
6. What would you do if we cannot run an A/B test?
7. What is the purpose of running an A/A test?
8. Explain "p-value" to a non-technical audience?
9. Your success metric is positive (stat sig), but a guardrail metric was negative. What do you do?
10. What do you do if you can’t run an A/B test?

Banani Mohapatra, DS Lead at Walmart is hosting a free workshop to go through the A/B testing interview process in more details and common interviews questions. 

📕 Save your spot here: https://luma.com/t5xh23i3

Then thank me later.

———

♻️ Repost this if you found it useful!
👋 Follow Amney for more Daily Tips

Universities have a major problem teaching Data Science.

They are full of 𝘱𝘳𝘰𝘧𝘦𝘴𝘴𝘰𝘳𝘴 with 0 experience in tech.

They teach you statistical theory. But they have no clue how to actually run an experiment at scale.

I wanna learn from those who work at the top.

Banani Mohapatra is one of them. She leads Data Science at Walmart, and she is doing a live workshop covering the real ML/DS interview process.

She is going to walk through:
- The end-to-end DS interview loop
- Real A/B testing interview cases
- And answer your questions

You shouldn't miss out. Really.

And it's completely free.

📕 Join here: https://luma.com/t5xh23i3

Enjoy. And thank me later.

taplio

The Battle

Sundar Swaminathan

Amney Mounir

Followers

Posts

Engagements

Likes

Performance ScoreSundar Swaminathan's LinkedIn Performance

Performance ScoreAmney Mounir's LinkedIn Performance

Ready to grow your LinkedIn presence?

taplio

The Battle

Sundar Swaminathan

Amney Mounir

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);}Sundar Swaminathan's LinkedIn Performance

Performance ScoreAmney Mounir's LinkedIn Performance

Ready to grow your LinkedIn presence?

Performance ScoreSundar Swaminathan's LinkedIn Performance