03-18, 14:00–14:30 (Europe/Madrid), Room 1
Tracking plans rot in spreadsheets while tags drift in production. Treat yours as code: three YAML files in Git, CI checks on every push, pull requests as governance records, and one spec generating implementations for Matomo, GA4 and more.
Where does your analytics strategy actually live? Scattered across chat channels, in a "Tracking Plan v3 - FINAL" spreadsheet last edited months ago, and in the memory of people who have since left. Audits help, but an audit is a snapshot, and drift is continuous.
This talk presents an "analytics as code" framework built on just three hand-written files in a Git repository:
- product-model.yaml: what the product is (entities, screens, funnels, squads);
- tracking-plan.yaml: what we measure (events, properties, owners, governance tags);
- validation-rules.yaml: what must hold true, checked on every commit.
Through a not-so-fictional SaaS case study, we follow a request end to end: a product manager files a ticket, an analytics engineer updates the YAML and opens a pull request, CI validates schema, naming, owners and PII rules, a lead reviews and merges, and the implementation guide and tag configuration are generated and deployed. One week later, a reconciliation audit catches what nobody announced: deprecated events still firing, events that were never in the plan, and git blame tells who, when and why.
We will also cover:
- why executives never need to open Git (Git is infrastructure, not interface);
- portability: one neutral spec, platform adapters for Matomo, GA4, Piano, Piwik Pro, PostHog and Snowplow, so changing platform stops being a rewrite;
- automated tests that assert every tracked hit matches the plan, including consent-granted and consent-denied journeys.
Takeaways: your tracking plan is infrastructure, pull requests enable governance (review, not restriction), and Git is your audit trail. Start small: one repository, three YAML specs, a handful of validation rules.
Julien Coquet is a digital analytics consultant and evangelist at N7 Data, based in Montélimar, France, with more than 25 years of experience in analytics. Julien is also an author, trainer, blogger and translator, and a regular speaker on the conference circuit, including Superweek, MeasureCamp and MeasureSummit. Away from data, Julien cooks and practises martial arts. At MatomoCamp 2027, Julien presents "Analytics as Code": how to keep a tracking plan in Git, validate it on every commit and generate implementations for Matomo and other analytics platforms from a single spec.