Practical writing on measuring a website honestly: consent and the law, cookieless measurement, migrating off Google Analytics, and getting numbers you can act on.
Consent law is triggered by storing something on a visitor's device, not by measuring traffic. Here is what that distinction actually means, and when analytics can run without a banner.
A pageview-by-pageview breakdown of Workers, D1 and Durable Object usage — why the write pattern decides everything, and where the free tier finally runs out.
Lab scores and field data measure different things, and only one of them counts for ranking. What each metric means, what causes bad ones, and which fixes are worth the effort.
Most funnel reports flatter you. Here is how step ordering, attribution windows and revenue rounding quietly inflate the numbers, and how to define goals that survive scrutiny.
Roughly half of all web traffic is automated. Most analytics tools catch only the bots that identify themselves. Here is what the rest looks like and how to exclude it.
Two tools on the same page will disagree by 10–40%, and neither is broken. Here is every reason why, and which number to trust for which decision.
Export the history, run both tools in parallel, reconcile the gap, then switch. The whole migration is an afternoon — provided you export before you delete anything.
The daily-rotating salted hash, explained properly: what it can measure, what it deliberately cannot, and why the limitation is the feature.
Data protection authorities in Austria, France, Italy and Denmark all ruled against Google Analytics. Then the Data Privacy Framework arrived. Here is where that leaves you.