English
Deutsch
Polski
Comparing AI Citations and Website Visits: A Practical Bing and Analytics Workflow
A local CSV workflow compares cited pages and measured sessions without inventing a click-through rate or treating missing rows as zero.
Digital Analytics
59 articles in this category
Follow this category by RSSAbout this category
A number in a dashboard often raises a question: Where does it come from, what is missing, and can it be compared with another number? Digital analytics is my professional focus. Here I share explanations of the technical details behind it.
You will find posts on GA4, Google Tag Manager, BigQuery, events, attribution and data quality. Configurations, examples and diagrams help you follow the measurement process and narrow down possible sources of error.
Tutorial: Fixing “Unassigned” Traffic in GA4 Acquisition Reports
Diagnosing the Unassigned bucket in GA4 traffic acquisition by splitting it on session source and medium first, then fixing the actual cause: non-standard UTM tagging, campaign parameters lost across domains or redirects, or a consent gate that fires the tag too late.
Retroactive and Forward-Only Analytics Changes: Keeping a Year-over-Year Comparison Valid
Five kinds of reach in sixteen months: one change rewrote the whole period, one acts forwards and cannot be undone, one left a break with nothing restated, one arrived device by device, and one switches itself on and off. A year-over-year comparison crosses all of them silently.
GA4 Thresholding: How It Differs From Sampling and How to Get the Rows Back
Thresholds, sampling, (other) and truncated date ranges need different interpretations. GA4 response metadata supports a more precise investigation.
Four Kinds of Number in a Conversion Report: Counted, Modelled, Projected and Estimated With a Range
A conversion report now holds counted events, modelled figures from consent mode, projected conversions from a future window and model estimates with a range. Three of them look identical in a column, and only one survives a re-count.
GA4 Behavioural Modelling: The Two Eligibility Thresholds and Where Modelled Figures Appear
Behavioural modelling needs 1,000 events a day with storage denied and 1,000 daily users with storage granted, at the same time. The second condition counts users rather than events, which is why a high refusal rate makes modelling less likely rather than more.
Tutorial: What Arrives in page_location and Why the Fragment Never Reaches the Server
A URL has five parts, and each system on the way sees a different subset of them. The part after the hash is the surprising one: invisible in every server log, and stored in full by the analytics tag.
Tutorial: Finding the Container Version in Which a Tag Changed
The version list says who published and when, and nothing about what moved inside. One fingerprint per element, pulled across every version, turns a question about forty versions into a table with one changed row.
GA4 Hostname Filter: Removing Ghost Spam From the Measurement Protocol and the Order of Steps
GA4 gained a hostname filter on 11 June 2026. It removes events by the domain they claim to come from – but a data filter acts only forward and drops matched events permanently, which makes testing mode the part that matters.
Tutorial: Querying GA4 Analytics Advisor via API and Checking the Answer with runReport
A reproducible channel comparison with properties.chat, session context, quota information and a separate runReport check.
