Deeplitic journal
Privacy-first analytics, explained clearly.
Notes on seeing the web clearly, respecting the people in the data, and building tools that make both possible.
Short field notes for teams who want useful measurement without the surveillance layer.
Analytics without the personal trail
A useful analytics system can show the shape of an audience without trying to recognize anyone in it.
Browse the journal
Ideas worth keeping.
The traffic your dashboard cannot see is still signal
Consent choices and blockers change the shape of reports. They should not change the questions your team is able to ask.
What to measure on a privacy-first site
Start with a small set of signals that map directly to decisions. More fields rarely make a report more useful.
Why we batch events before they reach PostgreSQL
The ingestion path should be fast for a visitor and boring for an operator. Batching helps with both.
You do not have to replace every analytics tool
Keep the workflows that already work. Add a clearer baseline where consent-dependent reports become hard to read.
Consent is a choice, not a measurement strategy
Consent matters, but it should not be the hidden filter that defines what your team thinks happened on the site.
A better way to read referrers
Referrer data is most useful when it explains an entry point without pretending to explain a person.
The smallest useful event model
A compact event model keeps analytics understandable, portable, and easier to protect over time.
Designing a dashboard for decisions
A dashboard earns its place when it shortens the distance between a question and a next action.
Why page duration needs context
Time on page becomes useful when it is read alongside the page, the journey, and the action that followed.
What a good analytics export looks like
Exports should be portable records that another person can understand without needing your dashboard open beside them.
Privacy by default starts at event collection
Privacy is easier to maintain when the collection path removes unnecessary detail before it reaches storage.
Operating a quiet analytics stack
The best analytics infrastructure is observable to its operators and invisible to the visitor using the site.
The case for short-lived visitor context
Many product questions need a visit context, not a permanent visitor profile.
How to audit a tracking script
A useful audit follows data from the first browser call to the final report, not just the snippet installed on a page.
What PostgreSQL gives an analytics team
For many product teams, a well-modeled PostgreSQL system is enough to make analytics reliable, queryable, and portable.
Campaign measurement without personal profiles
Campaigns need attribution context, not a record of every person who clicked a link.
When a chart should say less
A chart is clearer when its visual weight matches the certainty and importance of the question behind it.
Building trust into analytics metrics
Trust grows when the product makes collection, coverage, and uncertainty easy to see.
The analytics implementation checklist
A short implementation checklist can prevent a small analytics script from becoming a large, undocumented data system.