Updated September 2026
Know which channel actually produced the customer.
Most startups have five dashboards and no answer. Google Ads says it drove 400 conversions, GA4 says 180, the product database says signups were flat, and the CRM shows a different number again. I build the layer that makes them agree, then the weekly view that tells you where the next dollar goes.
Section
Why the numbers don't match.
Ad platforms count what's good for the ad platform. Analytics counts sessions. Product data counts users. Nobody's lying; they're measuring different things at different points. Until you join them at the user level, "what's working" is a matter of opinion, usually the loudest one in the room.
Section
Step one: make the tracking true.
- GTM and GA4 with a clean event model, not forty auto-events nobody uses
- Ad-platform conversions (Google, Meta, LinkedIn, TikTok) fed from the same events, with conversion values that reflect what a signup or a booking is actually worth
- Events for the things that matter and are easy to miss: embedded booking widgets, single-page-app forms, in-product actions
- Offline and server-side conversion imports where the browser can't be trusted
- Consent handled properly, because Europe is a customer too
Section
Step two: one view.
Product events (Mixpanel, Postgres) joined with GA4, Search Console, CRM, and ad spend into a single weekly report: visitors to signups to activated to paid, by channel, with cost. At Quadratic this replaced a Monday morning of screenshots with a view the whole company used, and it's what made a 60% reduction in blended CAC visible enough to act on.
I build these in whatever your team already has: a warehouse and BI tool, a Quadratic sheet, or a well-structured spreadsheet if that's where you are today.
Section
Step three: attribution you can defend.
Last-click flatters search and punishes everything upstream of it. I use methods you can explain to a board:
- Branded versus non-branded search lift for creator and brand campaigns
- Promo codes and dedicated links where they work; geo or time-based holdouts where they don't
- Incrementality reads for paid social instead of platform-reported ROAS
- A written attribution policy, so the numbers mean the same thing every week
Section
Step four: see how you show up in AI answers.
Nobody publishes query volume for ChatGPT, and every AI-visibility tool runs on synthetic prompts, so I treat this as directional and say so. What's measurable: sessions and conversions from AI sources in GA4 by landing page, AI Overview presence by keyword, and share of mentions and citations against named competitors from prompt monitoring. Useful for spotting which topics you're absent from and which third-party sites get cited instead of you. Not useful for forecasting.
Section
And the data that isn't in a dashboard.
Automated social listening (n8n, Reddit, X) that surfaces the threads where your buyers are asking questions, so the team can show up while it matters and not three weeks later.
Results
What this has produced.
- Built the growth reporting system at Quadratic across Mixpanel, Postgres, GA4, Search Console, CRM, and ad platforms
- Made a 60% quarter-over-quarter reduction in blended CAC measurable, and therefore repeatable
- Conversion tracking implementations across GTM, GA4, Google Ads, Meta, and LinkedIn, including custom listeners for embedded booking tools and single-page-app forms
- Full-funnel benchmark view at Quadratic: roughly a quarter of visitors converting to signup, tracked through to paid by channel
Formats
Usually a project sprint (tracking audit and rebuild, reporting view, attribution policy in two to six weeks), then ongoing as part of a fractional engagement. How we work →