Measurement
Can we trust these numbers?
Every platform counts the same sale
Meta says it made eight sales this week. Google Ads says eight too. The shop recorded six. Nobody is lying: each platform counts every sale it had a hand in, within its own window, and the same sale gets claimed twice.
The fix isn't picking the platform you like best. It's lining the claims up against what actually happened, so you know which channel is really doing the work.
Attribution windows. How long after a click or a view a platform still claims the sale. Each window is fair on its own. Added up, they count the same sale twice.
The visitors who aren't people
A spike in traffic looks like good news until you notice nobody bought anything. Some of it is bots, some is your own team, and some is the same visit counted twice.
I find the patterns that give them away and filter them out before they reach a report, so the number in the meeting is the number of people.
Bot filtering. Separating visits by people from everything else that loads a page: crawlers, scripts, monitors, and your own team.
What Measurement covers
Tracking and tag management
Client-side and server-side tagging. Deciding what to collect before collecting it.
- server-side
Analytics
Reports that answer a question someone actually asked.
Advertising measurement
Why every ad platform thinks it made the sale, and which one to believe.
Bots and bad data
Traffic that isn't people, found and filtered before it reaches a decision.
Pipelines and processing
Getting the raw data somewhere you can query it, clean it and join it.
- first-party data
Experiments
Tests that measure the thing you meant to measure. (It's rarer than you'd think.)
One click, all the way down
Between a click and the number in your report, the data passes through half a dozen hands. Each one can lose it, double it or bend it.
Most measurement problems aren't in the report. They're somewhere along the way, and finding where is most of the job.
- The click: Someone taps Buy. What goes wrong: Ad blockers and privacy browsers stop some tags before they load.
- The tag: A tag on the page notices, and packs up the details. What goes wrong: Fired twice, fired on the wrong page, or never fired at all.
- Consent: The consent state decides what may be sent. What goes wrong: Declined visits vanish from some tools and not others.
- Server-side: Your own server receives it, and passes it on. What goes wrong: Mapping mistakes: the right event with the wrong value.
- The platforms: GA4 and the ad platforms each record it. What goes wrong: Each applies its own attribution, sampling and thresholds.
- The report: Someone reads a number in a meeting. What goes wrong: It says 'sessions' and everyone hears 'customers'.
Writing on measurement
All of it
Not sure which number to believe?
Tell me what you can see going wrong, or what you'd like to be possible. You don't need the technical name for it.
Contact
Tell me what the problem is. You don't need the technical name for it: working that out is part of the job. If I am not the right person for it, I will say so. I reply myself, from Hobart, which is ahead of most of the world (by time zone, and in several other ways I am too modest to list).
- or call
- Book a call
- or code
- github.com/neobadger
Sent. Thank you. It goes to James, and nobody else reads it. I reply in person.