Expertise
Measurement and attribution
If the work succeeded, how would you know, and how long would it take you to find out?
What it means
Measurement is deciding what to count and building the means to count it. Attribution is connecting what changed to what caused it. In search and AI visibility both are harder than in paid channels, because the systems doing the answering report almost nothing about why, and the interval between a decision and its effect runs to months.
Why it matters
Analytics gets installed once and then trusted. The tags sit on the pages that existed when somebody set it up and not on the ones added since. Goals go unverified for years. A working product can look like a failure because nobody instrumented the page it lives on, and a failing programme can look healthy because the only number anyone reads is total traffic. Deciding without measurement is expensive. Deciding on broken measurement is worse, because it feels informed.
What you control
- In your hands
- What you decide to count, whether the instrumentation is genuinely working, and the discipline to record a baseline before the work starts.
- Not in your hands
- What the platforms choose to report. AI systems disclose almost nothing about why they cite one source over another, and no amount of instrumentation changes that.
What we do
- Verify what is actually being recorded, which is regularly not what everybody believes is being recorded.
- Choose the small number of measures that would change a decision, and drop the rest.
- Record a baseline before anything starts, because a change nobody measured beforehand cannot be claimed afterwards.
- Separate brand from non-brand, which for most companies is the single number that says whether any of it is working.
A working product, nearly invisible
One engagement launched an assessment tool that reached the first page of results for its category within weeks of going live. Analytics recorded a single view of that page against more than a hundred search impressions, because the tracking container had never been added to it and the lead goal had never been verified. The product was working and the numbers said nothing was happening. Zero conversions was a measurement failure rather than evidence of no demand.
How we use the framework
The Digital Reality Gap is a distance, and a distance is a measurement. Without one, closing it is an assertion. This is the part of the framework that makes the rest of it falsifiable.
See the full frameworkCommon questions
What is the one number to watch?
For most companies, non-brand share of organic clicks. Total traffic hides everything: a brand doing the work, a market that is seasonal, a site nobody finds without already knowing the name.
How long before we can tell whether it worked?
Months rather than weeks for organic, and the baseline has to exist before you start. Paid answers some of the same questions in days, which is often reason enough to run it alongside.
Can AI visibility be measured at all?
Partly. You can measure what systems say about you, how often, and which sources they draw on. You cannot measure why they chose you, because they do not report it, and tools claiming otherwise are inferring.
Start with what the market can see.
A conversation is enough to know whether there is a gap worth closing.