UTM Chain Check

Which parameter changed, and at which redirect.

A working campaign link experiment, with tested instrumentation. Follow a link through every hop and read what happened to the campaign parameters, value by value.

Functional checks completed. Growth impact not evaluated.

Control chains:

Why this experiment

The RedirHub guide on preserving UTM parameters through redirects asks the reader to follow a link through every hop, confirm the final address still carries the expected values, and keep a small log of expected against observed. This prototype is that interaction, already built.

On the public interfaces reviewed on 17 September 2026, a chain check reports status codes, the hops and the final destination. Naming which expected parameter disappeared or changed, at which transition, with the values before and after, was not observed in those interfaces. That single answer is what this prototype returns.

Whether it changes anything for a reader of the guide has not been tested, and nothing here claims it does.

The pages that were reviewed, the rules used to read them, and each observation with its status are listed in the comparison manifest. The guide itself is here.

Ready to measure

The instrumentation runs end to end already: an action in the browser, an event received, stored and deduplicated, and a query executed over what was recorded. Carrying it would take one entry point from the guide, subject to your agreement, since a page that sits outside your site builds no audience on its own. Nothing is measured on your side, and no traction is shown.

Hypothesis
A reader who is told which parameter changed, and where, is more likely to act on their own links than a reader told only that the chain has three hops.
Primary measure
Sessions that ran a scan on a link of their own and then clicked through to the guide, over sessions exposed. Computed by a query that runs in the checks, against the event contract, on the recorded test dataset.
Guardrails
Scans that end in a refusal or a timeout, events rejected by the collector, and outbound requests per minute. Any security defect stops the run.
Decision
Review at fourteen days as a management choice, not a statistical claim. Insufficient exposure is reported as insufficient exposure, and no conversion target is invented.

Event contract, recipe cases, the executed query and its ClickHouse wording are in the experiment card. Mapping it onto an analytics store of yours remains to be agreed.