bibb Analytics: Build in Public

If you work in BI, you have likely built data models for other people’s businesses. The odds that you have applied the same rigor to your own side project are low. I know because I did not, for a long time.

Marketing analytics is also one of the most in-demand adjacent skills for BI professionals. Companies want people who can connect commercial questions to the data that answers them. That combination is closer to what BI people already know than most realize.

This series documents doing it properly, on bibb.pro, in public.

Pascal and I connected back in October 2025. He is co-founder of Dashflow, a Swiss marketing analytics startup working with some of the largest companies in Switzerland. He brings marketing expertise with a strong BI background; I bring the analytics depth with a solid grasp of marketing. We found common ground early and the collaboration felt natural.

It is important to note that this collaboration between bibb and Dashflow does not carry any financial agreement. We are very thankful to Pascal and the Dashflow team for allowing us to use their platform and for the consulting we have received from them.

Marketing analytics beyond the Google Analytics (GA4) basics has always been something I wanted to get right for bibb. The blocker was never interest, it was everything that comes with it: bibb has grown to a size where the native connector’s shortcomings are painfully obvious. The complexity of marketing analytics is high, and doing it right is easier said than done. And running this solo, on top of everything else bibb already demands, is a real constraint. This one stayed in the to-do bucket for a long time.

Happily, two things changed. Sajjad joined the project, which brought the capacity to actually do this properly instead of just plan it. And my book was finally published, which freed up the focus to commit.

bibb’s setup looks simple on paper: GA4, Dashflow, Mailjet and Gumroad. In practice, marketing analytics is one of the more complex integration problems in the space. This series documents that honestly.

To be completely honest, we are not starting from zero. GA4 has been running since day one, the client-side configuration is solid, and the consent architecture is fully GDPR compliant.

The goal of this project is collecting actionable data that supports bibb’s growth.

The series will be broken into several posts, each covering a different aspect of the marketing analytics implementation:

Planned posts in the series

This series is published as it happens. Posts get updated as decisions are made and results come in. The log below tracks what changed and why.

Everything documented here reflects real decisions on a real platform.

Diary and log change

• 2026-07-23: Pascal proposed the build in public structure: tracking setup on bibb.pro, Google Ads as an additional channel, and a public dashboard built by Sajjad.
• 2026-08-19: Decided to stay with direct GA4 over GTM. Pascal recommended GTM for easier future deployment, but switching would require reworking the existing consent-first architecture, which already gates all tracking scripts behind user consent. The risk was not worth the benefit. The direct implementation uses a ga_client_id user property to give Dashflow the session identifier it needs for matching. Pascal confirmed this works on their end.
• 2026-08-19: Gumroad receipt snippet approach confirmed. Gumroad supports custom JS on the receipt page and exposes the order ID via $ORDER. Plan: capture ga_client_id and order ID together on the receipt page. Pascal confirmed this is the best-case scenario for matching transactions to sessions.
• 2026-08-19: Dashflow confirmed Gumroad data ingestion via API pull on a schedule. Schema and authentication to be defined in a follow-up meeting with the Dashflow team.
• 2026-08-10: Broke the original post into three standalone posts (strategy, GA4, implementation) for a better reading experience. Added a Dash-Flow post to the series. Added this diary section.

Start with goals, not tools

I want to thank Pascal. Through our conversations, he helped me clarify the business goals behind marketing analytics measurement. His deep understanding of how site analytics, attribution, and business value connect helped bibb get these goals right.

The first move in any BI engagement is not to open Power BI. It is to understand what questions the business needs to answer and what decisions depend on the data.

Same principle applies here.

At a high level, bibb has two main goals: user experience and commercial outcomes. You cannot achieve the latter without getting the former right. The commercial priorities are clear: sell templates, grow the newsletter, deliver on sponsor commitments. Everything else is either a leading indicator or a diagnostic signal. We organized this into three layers.

This is a balance game. Early on, we decided that UX comes first and that no commercial decision should compromise it. For example, bibb reached 48K users in the year to August 2026. We decided against ads anyway. Sponsor content is clearly labeled and only accepted when it brings real value to our readers.

The same applies to the free tools. They should be fully usable with no purchase required. Users should come back because the content and tools are genuinely useful. That is the bar. Our 28.3% return rate tells us we are on the right track.

With our high-level goals defined, we can now move forward to our layer measurement strategy: conversions, engagement, and diagnostics.

The three layers of measurement

The layers are not a hierarchy of importance. They are a way to group and organize the measurement model.

Conversions Layer. These are the outcomes the business exists to produce. A reader who buys a template is telling us the content and tools delivered enough value to pay for. A visitor who subscribes to the newsletter is making a longer bet, choosing to keep hearing from bibb without an algorithm deciding whether they do. A reader who clicks through a sponsor placement is completing a deliverable we promised. All three are things we can point to and say: this happened.

Engagement Layer. Not everything worth measuring converts immediately. Theme downloads, tool usage, blog visits, carousel and lightbox interactions do not generate revenue directly. Users who engage at this level accomplish our foundational User Experience goal and convert at higher rates. That makes them useful signals for diagnosing the funnel. Not proof that something worked, but evidence that something is moving in the right direction.

Diagnostics Layer. Consent acceptance rate, client-side errors, AI magic failure rate. These are not KPIs. They are checks. But they also serve the User Experience goal directly. A spike in client-side errors means something in the product is broken. That is a UX problem before it is an analytics problem. The kind of signals you look at when something feels off, not when you are building a report.

This separation matters for the same reason it matters in any data model. You do not want to aggregate across layers. A lightbox impression and a template purchase are different facts. Keep them in different buckets, measured with different expectations.

What is next?

Strategy without implementation is just a document. The next post will cover Google Analytics: how sessions work, how UTMs flow into attribution, and how we keep session context intact when a user crosses from bibb.pro to Gumroad to complete a purchase. That last part is harder than it sounds.