Learn

How the plumbing works. Written from doing the thing, not from reading about it — every guide carries a screenshot, a dataset, a test result or a worked example.

Browsers strip your click IDs and leave your UTMs We ran one ad-click URL through Firefox's and Brave's published strip lists. Google, Meta and Microsoft click IDs go. LinkedIn, Reddit and Pinterest IDs stay.
Conversion modeling has a floor, and only Google publishes it Consent mode models lost conversions only above a traffic line. Google states its line; Microsoft and Pinterest model with no published threshold at all.
Server-side tags drop consent unless your code sends it Six ad platforms' server-side conversion APIs, read field by field: three can carry a visitor's no, and Microsoft reads a missing one as granted.
MMM vs last-click attribution: 104 weeks against one click Last click credits one order the day after it happens. An MMM needs two years of weekly totals and grades channels, not campaigns. Three MMM docs compared.
Client-side vs server-side tracking: where the event starts Server-side GTM, tag gateway, Conversions APIs and order webhooks all get called server-side. Where each event starts decides which sales it can see.
GA4's AI Assistant channel only counts the untagged clicks GA4 files a visit under AI Assistant only when a recognised AI referrer arrives with no UTM tags. We ran five kinds of AI click through Google's own rules.
Same three orders, five conversion rates One visit log, three orders, six tools' session rules: the conversion rate comes out at 27%, 30%, 33%, 50% or 60%, set by what each tool divides by.
Ads that do nothing get a lift result half the time Google reports lift at 50% certainty, which is 1 minus the p-value. We simulated 2.4 million lift studies: half of the ads with no effect got a lift result.
Your postback URL returned 200. The conversion can still be lost We sent 22 test postbacks and 12 lost or changed data, often behind a 200 OK: unreplaced macros, raw + and # signs, redirects. How to test first.
A hashed phone number is a phone number with an extra step We hashed all 6.4 billion possible US and Canadian phone numbers on one cloud VM in under six minutes. Why a hashed contact list is still personal data.
Your 90-day click window rests on a cookie that may last 7 LinkedIn credits a click for 90 days but keeps its click cookie for 30. Safari deletes script-set cookies after 7 days. Five platforms' windows and cookies.
Switching ad trackers? Test what it sends back to each ad platform Eight ad platforms each want their own click ID field, dedup key and deadline, and they hash one email in different ways. What to test in a tracker trial.
Primary demand advertising also sells your competitors' product Primary demand ads grow a whole category, so every seller gains. In the US, 21 federal checkoff boards buy them and measure the return without clicks.
The most visited US websites, and what AI does to the count Four rankings, four rulers. One counts DNS lookups, one hides the adult sites, and none sees an answer read inside an AI assistant.
Attribution is two jobs: analytics and the ad platform An analytics tool attributes to report across channels. An ad platform attributes to credit its own ads. Sending a conversion server-side does neither job.
Three tools say "last click", each means a different click Google Ads, GA4 and LinkedIn all offer a last-click or last-touch model. Each credits a different touch, so one sale gets three different answers.
Two kinds of click ID, and why they never swap Ad platforms and tracking platforms each create their own click ID. Each one only matches in the system that made it, so the wrong one attributes nothing.
LinkedIn counts a view for 90 days. Google counts 1 day Default view-through windows run from 1 day to 90 days across six ad platforms. Add their conversion counts together and you add unlike things.
Why your ad platform rewrites past numbers Ad platforms write conversions backward onto the click date, so past days keep changing for as long as the attribution window stays open.