deepskydata ApS

It took me over 7 years to find the approach to create a data agency. All the ones I saw before did not work for me. But I think, I found the setup I want to scale: The leading agency for tracking & data collection. Building the data infrastructure that enables data, data science, product, customer success, marketing & sales team to shine.


So we love tracking - isn’t that problematic

Let’s be specific: We love to track products, we love to track subscriptions, love to track purchases, love to track content. We do not love to track users or customers. We do all the things to improve products or marketing communication not to stalk anyone. Data protection is a core asset for all our projects and we don’t do projects where we are asked to work against it.

Founder, Event Data Architect & Consultant

Audible GmbH

Product and AWS Architect for internal services

Timo Dechau

I started 10 years ago in product management. My approach to improve products was always very data focused. So my first steps in all jobs was to get the tracking and analytics in shape and enrich the data with other data sources, so I have my perfect data setup for making product decisions.

In 2014 I decided to bring my data experiences to more companies as an independent consultant. In the last 4 years I have done about 20 analytics setups mainly with tag management systems. Most of the time for the Google stack (Analytics, Tag manager and Big Query) but also with Adobe Analytics & DTM, Tealium, Mixpanel, Amplitude and Segment.

My background is economics and computer science. So I feel comfortable to translate business requirements down to tracking requirement (javascript code).

Besides I can write Python fluently and use it to wrangle data, consume data apis and do visualizations.

Freelance Data Consultant 

Adspert.de - Bidmanagement GmbH

Lead the product development of adspert - an automatic service for optimizing your search engine marketing campaings. Product owner.

Head of Product

Manage the product requirements / features / ideas for the german marketplace. Collaborate with the product teams in US & UK.

Product manager

d-SIRE GmbH & Co KG

Manage different major customer accounts (E-Commerce, Fashion brands), work on online strategy project, plan and manage relaunch projects

Projektmanager

dhd24.com - der heisse draht Verlagsgesellschaft

stellv. Leiter Internet Services

Projekt-/Produktmanager

Data & Product 🕹🧙🏻‍♂️ - Host of the Better Together show (For data & product lovers) - Chief Content Maker at Deepskydata (we produce video content for data brands) - thoughts -> substack.com/@timodechau

Do you have Analytics & Tracking questions?
Write me a message right now.... I'm curious to learn more about your tracking scenario and your questions.

I can help you to set up, refine, update or extend your current tracking & analytics system. We can join data from your different data source to create powerful user profiles for high performance targeting campaigns.

8 years product development experience. Set up various tracking systems with Google Analytics, Mixpanel, Kissmetrics and some more. Create reports that help stakeholders to make better decisions based on data. 


Yesterday I did my first free & live workshop on growth intelligence focusing on some selected growth metrics like: activation rate, revenue cohorts, activation to monetization rate, PQA scoring and more. All broken down by useful micro segments like LinkedInAds-WebinarA-Pro_plan-10_100p. 

To be honest it was quite a roller coaster ride for the 55 people attending (out of 200 signed up who will hopefully catch up with the recording). When I did the dry run before I had the feeling that I putting 3h into 1h. So next time, I will do a better refinement. But I got excited to share what I am thinking about and working on for the last 12 months.

As bonus I also recorded a short loom how weekly business review dashboards with these metrics look like and what AI can do with these enriched datasets (definitely more than just chat with your data).

Watch this space - there will be a next workshop in December with designing a growth intelligence roadmap for 2026.

Event Data Wizard 🕹🧙🏻‍♂️- Author of the Analytics Implementation Workbook. Thoughts: -> timodechau.com

Wednesday is the day.

I'm running my first Growth Intelligence workshop—the one where I share the lessons from my experiments trying to bring product, marketing, and revenue analytics together. 

Nov 19, 7 PM CET (1 PM EST for US East Coast, 10 AM PST for West Coast)

Here's what we're covering:

• Why your Amplitude/Mixpanel setup can't answer revenue questions (and what's missing).
• The 5 questions product teams often can't answer and 10 metrics that can help you with that
• How to actually build this. I'll show you the framework, the architecture, what data you need, and where most people get stuck.

  ---

  You'll walk away with:
  - The full framework as a PDF
  - Recording of the workshop
  - Clear next steps for your specific setup
  - Option to book a free assessment call if you want help mapping your metrics

  75 minutes. Live. Free.

If you've been tracking DAU/MAU and feature adoption but still can't answer "what's actually driving our growth?", this is for you.

Click on "Visit my website" under my name to join the workshop.

I am working on a decision tree for extending your product analytics setup to make it better in telling where you loose new accounts on the way from new -> activated -> active (stay active) -> revenue -> at risk -> dormant -> churned.

Tonight, I will use this tree to explain in which parts you should invest in your analytics setup in my last free workshop this year. 

Sign up if you like to watch it live or get the recording or get the decision tree later. Click on visit my website under my name to get to the Luma page.

Activated accounts become more useful when you break them into levels instead of treating activation as a single event.

Activation is about defining what “first value” looks like within a short period, usually the first 7 days.

If someone reaches this point, the likelihood of converting later increases significantly.

But defining activation requires experimentation.

It’s rarely obvious which behaviours correlate with long-term value.

Introducing levels makes this more practical.

Take a community example, like Alt+.

Different behaviours signal different levels of engagement.

For instance:

• Level 1: liking a post
• Level 2: commenting on a post
• Level 3: writing a post
• Level 4: attending an office hour
  
These behaviours are easy to track and reflect increasing commitment. 

You can analyse how people at each level convert into paid membership.

Ideally:

• Level 4 converts at the highest rate
• Level 1 converts at the lowest
  
If this pattern doesn’t hold, you refine the levels.

The advantage is that within the first 7 days you already have strong signals about which new accounts are likely to convert.

You can extend this level system into ongoing engagement to identify churn risk, track active members, and improve onboarding flows.

Levels turn activation from a binary metric into a structured view of user progress.

It helps you understand quality earlier and make better decisions across acquisition, retention, and product.

---
I am running a new free workshop on 10.12. -> Build Your 2026 Growth Intelligence Roadmap. Click on "Visit my Website" under my name to claim the seats (right now, only 50 are available).

Sure, we are either talking about text to sql or agentic analytics or on the other side how ridiculous this is when we can't even define customer or ensure their data quality.

But there is something happening deep under it, which is more interesting.

10y ago Snowflake, Bigquery and others changed the analytical data storage and compute game by splitting it up and make it scale. Unlocking data stacks for massively more projects (potentially we could give Redshift here some credit as well to get it all started). What followed, we all know - MDS and stuff (which we can now happily declare to be over).

I have the feeling we are approaching a similar point in time. Open table formats and things build on top is not my area of expertise, but I follow Julien Hurault and Mehdi Ouazza very closely writing about it and have my regular nerd talks with Tim Frazer 🚀 and Pete Fein. So I follow closely how things are moving.

Yesterday I came across Boyan Balev's beautiful designed post about cost implications comparing Snowflake and others to OSS stacks. He is not the first to write about it but he does it in a very compelling way (also triggered me to go back to the drawing board with my current blog redesign). Check it out - link is in the comments. His post is about the economics not the tech. But economics kicked off the thing in 2015. Well, it might do it again in 2026. 

It's a natural cycle - new technology unlocks new use cases that need new technology to unlock new use cases.


----
I am doing my next free growth intelligence workshop in just 8 days - and it is not about data warehouse economics. But data warehouses will play a role in it - mostly that it becomes more a requirement in product and growth analytics. 
Click on "visit my website" to join.

Maybe 2026 is the year where product analytics can tell you something about if your product really works and where you are loosing too many users?

One of the major struggles with product analytics setups is that they focus on user interaction. Like, if someone has clicked something. They also rely on custom funnel and cohort setups to get some more generic performance insights about user retention and success. But these usually require advanced analysts to create them and these are rare and only have 8h a day (if the coffee is good).

My recommendation for these setups is to handle product analytics in a different way. Moving away from interaction measurement towards use case / jbtd, user states and combine it with revenue. This is a substantial investment since these concepts are ready easily but need work to get them implemented (interestingly not so much by the developers).

On Wednesday I will run my final free workshop on exactly this. What could be a roadmap for 2026 to get product analytics in a shape that it actually can help you with your backlog planning and growth goals.

Click on "view my website" under my name to sign up for the workshop. As always the recording will be shared, so make sure to also sign up when you can't make the live time.

Time can distort how we read metrics.

In growth, this happens often: the metric looks right, but the timing behind it tells a different story.

Take a simple example.

Imagine an AI tool with a free plan where users typically upgrade after roughly 75 days.

That delay alone creates issues in classic BI reporting because monthly numbers mix users from completely different periods.

That’s the first problem I want to unpack.

When you look at a typical BI report, you usually see monthly numbers: new accounts, new subscriptions, new revenue. 

This is useful for understanding the overall shape of the business, but it becomes misleading when you try to create relational metrics from it.

Here’s a simple number example.

In July 2025, you might see:

• New Accounts: 4,200
• Free → Paid Conversion Rate: 8%
• New MRR: $12,600
  
On the surface, it looks like the month performed well and that 8% of July’s accounts converted into revenue. 

But that’s not what happened.

Those revenue numbers mostly come from users who signed up in April or May. 

With a 75-day upgrade cycle, July’s accounts won’t show their revenue impact until much later.

This is why classic BI reporting looks clean but leads to wrong conclusions when you try to evaluate performance.

To avoid this, you need two separate reporting layers:

1. Classic BI reporting Monthly new accounts, subscriptions, MRR. 

Good for understanding outcomes.

2. Cohort-based performance reporting 

For example:

• “Of the users who signed up in July, how many have converted by today?”
• “How does the August cohort compare?”
• “What is the conversion curve for each cohort?”
  
One layer describes the business.
The other describes performance over time.

Most teams have the first layer. 
Far fewer have the second one consistently in place.

---
I am running a new free workshop on 10.12. -> Build Your 2026 Growth Intelligence Roadmap. Click on "Visit my Website" under my name to claim the seats (right now, only 50 are available).

What's your growth metrics roadmap for 2026...

I know, I know - this is a typical start for a generated LinkedIn Post. So let's do it a bit more casual.

Measuring growth is an essential thing, but it sounds easy on paper but is much harder to do. Just measuring conversions in GA and user clicks in Amplitude does not really tell you something about how your product and business is growing. And just looking at the Stripe dashboard does not tell you where this MRR is coming from and if it still will be around next month. 

And yes, we can pop numbers now into our favorite models and get some good ideas back what to do next. But as I wrote in a post a while ago, we still need a foundations layer that shows us our core growth metrics to see if all the buzz is actually changing something.

Therefore, I am doing my next free workshop on the 10th of December about how to build up this foundation in 2026. It will be slower than my last workshop and put some more nuances on the basics and not the full potential of growth metrics (but everyone signing up will get the recording of November's workshop as well). 

Click the "Visit my website" under my name and sign up. Seats are initially limited to 50 (I might extend).

A few days ago, I wrote about how time distorts performance metrics when users take around 75 days to upgrade. 

There is a second challenge this delay creates: slow feedback.

If the free-to-paid cycle is this long, marketing needs to wait around 70–75 days to see the actual revenue impact of a campaign. 

This makes optimisation extremely difficult.

Using the same AI tool example, this delay means that marketing investments made today only show meaningful revenue signals months later.

That’s the second problem I want to walk through.

Imagine marketing invests $20,000 in July to acquire new users.

With a 75-day delay between sign-up and conversion, you only know in late September whether this investment actually produced paying customers. 

Early signals like new accounts can show something about targeting and landing pages, but they don’t tell you whether those users will eventually pay.

This becomes more complicated when the investment continues over multiple weeks. 

Let’s say you plan to spend $80,000 over 10 weeks, roughly $1,000 per day.

If you wait 75 days to understand whether this investment is working, you’ve already spent $75,000 before receiving meaningful feedback. 

That’s not a manageable loop.

You need proxy metrics that provide earlier signals.

One of the most effective ones is the activated account metric. 

An activated account is a user who completes specific behaviours in the first 7 days that historically correlate with becoming a paying customer.

In a presentation AI tool, this could include:

• Generating multiple slides with AI 
• Sharing a presentation 
• Updating slides after creation

The exact activation definition requires experimentation, but once you have it, your feedback window shifts from 75 days to 7 days. 

You can quickly identify which channels and campaigns bring in high-quality users and adjust your acquisition efforts much earlier.

Proxy metrics don’t replace revenue metrics. 
They help you see revenue potential sooner.

---
I am running a new free workshop on 10.12. -> Build Your 2026 Growth Intelligence Roadmap. Click on "Visit my Website" under my name to claim the seats (right now, only 50 are available).

taplio

Timo Dechau's Linkedin Analytics

Timo Dechau

Performance ScoreTimo Dechau's LinkedIn Performance

Timo Dechau's Best Posts (last 30 days)

Ready to grow your LinkedIn presence?

taplio

.css-6ytdur{background-image:linear-gradient(284deg, rgba(39, 43, 72, 1) 0%, rgba(33, 47, 183, 1) 100%);color:transparent;-webkit-background-clip:text;background-clip:text;}Timo Dechau's Linkedin Analytics

Timo Dechau

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);}Timo Dechau's LinkedIn Performance

Timo Dechau's Best Posts .css-1emoo84{font-size:var(--chakra-fontSizes-md);font-weight:400;color:var(--chakra-colors-gray-600);}(last 30 days)

Ready to grow your LinkedIn presence?

Timo Dechau's Linkedin Analytics

Performance ScoreTimo Dechau's LinkedIn Performance

Timo Dechau's Best Posts (last 30 days)