What cannot be measured.

What can this approach not do?

Every measurement system has an edge, and the ones that do not publish theirs are usually the ones scoring past it. This page is the edge of the Digilu Observatory, stated deliberately, because a client who discovers a limit on their own stops believing the parts that were accurate.

Relevance is judgment, not instrumentation

Trust signals can be scored from a page’s own HTML. Visibility can be sampled repeatedly across AI engines and returned as a number with a version attached. Both are genuinely deterministic, which is why they are published as scores.

Relevance is not. Whether a business still matters to the people it serves, in the language those people now use, cannot be scored honestly by a crawler. It requires knowing what customers are actually deciding between, which changes, and which no engine has access to. That part is done by a person, and it is the clearest difference between a membership and a dashboard subscription.

What the Observatory does not see

• Review sentiment. What is measured is the machine-readable layer: whether ratings and profiles are marked up and linked. What customers actually said, and whether the complaint is fair, is reading, not scoring.
• Sales conversations. Everything after someone picks up the phone is invisible to it. A business can be highly visible, well structured, deeply trusted online, and lose every deal at the quote.
• Product quality. No amount of visibility work survives a product people regret buying, and no engine here can tell you that is what is happening.
• Why an AI engine answered the way it did. The answer can be sampled and the pattern reported. The mechanism is not disclosed by the providers, and any confident explanation of it is a guess wearing a lab coat.

Where continuous observation is the wrong purchase

Three cases, stated plainly because the alternative is selling something that will not work:

• The problem is already known and specific. A business that knows exactly what is broken and needs it fixed once should buy the fix. Continuous observation is for businesses whose problem is that they will not find out in time.
• Nobody can act on the findings. Observation at a membership level without implementation produces recommendations for someone to carry out. If there is no such person, the briefing becomes a monthly reminder of work not being done.
• The timescale is short. The value here compounds through history and baselines. Three months produces a picture. It does not produce the thing that makes this worth paying for.

Why publishing this is the strategy

A measurement business is bought on whether its numbers can be believed, and the cheapest way to make a number unbelievable is to score something that cannot be scored. Every claim in the Digilu ecosystem is either evidenced or labelled as still developing, unknown is recorded as unknown rather than as zero, and no engine threshold is ever relaxed to produce more output.

That discipline costs output. It is what makes the remaining output worth reading, and it is the same standard Adaptive Brand Management is sold on: taking responsibility for trust means being the first to say where the evidence stops.