Affiliate Tracking, Attribution Windows, and Postback Hygiene
Affiliate disclosure: We may earn from partner links. We follow local rules. Readers should see offers and risks with clear terms.
Measurement, attribution and the unglamorous plumbing of acquisition — written for people who have to defend the numbers.
This desk is written for the person who has to stand up in a meeting and defend a number. That makes measurement the subject and tactics the consequence, rather than the other way round: how a conversion is defined, what the attribution window does to the credit, which events are modelled rather than observed, and what happens to a reported figure when consent rates move underneath it. A tactic that works only until somebody checks the instrumentation is not a growth story, it is a reporting error with a budget.
The desk is sceptical of two things in particular. The first is any single metric offered as the health of a business, because a number that cannot be decomposed cannot be acted on. The second is a benchmark without a population — an average that does not say whose average it is, measured how, over what period. Where a figure here comes from a platform’s own documentation, the piece says so and says when it was read, because those documents are revised without notice.
Affiliate disclosure: We may earn from partner links. We follow local rules. Readers should see offers and risks with clear terms.
Online casino promotions look simple. “Get a bonus.” “Free spins.” “Cashback.” But behind the words, there is psychology. Casinos use ideas from…
Published: 2026-07-04 • Last reviewed: 2026-07-04 • Author: Alex Petrov, iGaming SEO Lead
Spam complaints, not subject-line cleverness, decide whether your bulk mail reaches an inbox. Google's published guidance advises bulk senders to keep complaint rates below 0.10% and avoid reaching 0.30%, measured against delivered mail rather than messages sent. Cross that second line and your mail starts routing to spam folders—or disappears entirely.
Branded search only works as a metric once you fix what "branded" means. Google Search Console now offers a branded queries filter that splits your Search results Performance report into two buckets: branded and non-branded. The catch?
A single year-over-year traffic drop proves nothing. The same dip could mean summer slowness, permanent demand collapse, or Google simply keeping the answer on its own page. The only way to know is to compare whole cycles—several years of the same weeks, mapped against market-wide data and whatever the results page looked like at the time.
When your affiliate dashboard shows a different number of conversions and your internal analytics counts a different number, the gap is rarely a mystery. The two systems are often measuring different things: browser-side pixels fire from the user's device when a page loads, while server-to-server postbacks transmit from your backend after a transaction completes.
A user who returns to a site after eight days can count as two visitors instead of one. Adobe's documentation states that under a seven-day expiry, a visitor returning on day eight is treated as new. The metric no longer measures loyalty. It measures identifier survival.
Independent trade desk. We take no commission on anything we describe and run no affiliate programme of our own. Every figure on this page names the standard, filing or organisation it comes from; where a number could not be verified the page says so. How we work and how we correct. Reviewed:
Cookies, and what this site stores. The Dispatch sets no advertising or analytics cookies and loads no third-party tracker. Closing this notice writes one key — icd-notice — into your browser’s local storage, so that the notice does not return. Nothing else is kept. The one thing a page here sends onward is what a reader types into the form on the contact page, and that is described before the form is used. What the policy says.