Practice 05 of 05
Signals
Measurement has a way of expanding until it becomes its own project. This practice exists to stop that. We put in enough tracking to answer the handful of questions that actually change a decision, write down what each number means, and resist building a dashboard that nobody opens twice.
Asset: knowing which of the above is working
- GA4
- Server-side tagging
- Consent
- Reporting
How we run it
The working approach.
- 01
Start from the decision
We ask what you would do differently depending on the answer. Anything that fails that test does not get instrumented, which is how a measurement project stays finishable.
- 02
Repair before extending
Duplicate events, consent firing after the page view, cross-domain journeys losing their thread. Most properties we inherit are wrong in ways nobody has noticed, and fixing that is cheaper than adding to it.
- 03
Write the definitions down
Every report ships with plain-English notes on what each number counts, where it comes from and when it will disagree with another system. Arguments about whose figure is right end there.
- 04
Leave it maintainable
Containers, properties and cloud resources sit in your own accounts, documented well enough that a new hire could pick them up. Measurement nobody in-house understands is a liability, not an asset.
What you are left holding
The things that stay after we go.
- 01
An agreed shortlist of questions the measurement has to answer
- 02
GA4 repaired or rebuilt against that shortlist, not against a checklist
- 03
Server-side collection where the spend justifies the maintenance
- 04
Consent handling that is honest and does not quietly break delivery
- 05
One report, with a written appendix stating what every metric counts
- 06
A note of what we deliberately chose not to measure, and why
On the ladder
How durable is this layer?
Every practice sits somewhere on the ladder of things a business can own. Knowing which rung you are buying is the difference between an investment and a subscription.
Paid reach
Last, deliberately. Advertising is excellent at accelerating something that already works and poor at substituting for something that does not. Placed at the top of the ladder it does its best work and costs the least.
- Brand and non-brand separated and reported apart
- Creative testing with stated arguments
- A written condition for reducing or stopping
Questions
About this practice.
Partly, and that is the honest answer. Some of the gap is genuine breakage and repairable. The rest is structural, because the systems count different things over different windows. We repair what is repairable and document the remainder so it stops being a monthly argument.
Where to start
Find out how much of it you actually own.
Tell us what you have and we will send back an independent read: which parts would still be working if the advertising stopped, which would not, and what we would build first. No obligation attached to any of it.