Principles
Read before claiming. Every tactic names its nearest prior work and the exact difference. A tactic is not new because its acronym is.
Visibility claims are not business claims. Mentions, citations, referrals, leads and revenue are reported separately. More mentions is not more traffic; more traffic is not more revenue.
One run is an anecdote. AI answers vary. Every score states its engine, date, prompt set and number of runs, and comes with its margin of error.
Measured, or not said. A number appears here only with its method beside it. Negative results are published as findings.
Interests are declared. I work in marketing. Where an experiment touches a product I work on, it says so.
What this lab never promises
No guaranteed AI visibility. No reverse-engineered algorithms. No secret formula. In a field full of confident promises, this lab holds itself to evidence, and every claim it makes has to earn its place.
What would falsify the vision
- No page-level change lifts citation rate against a matched control on more than one engine.
- Answers vary so much between runs that no visibility score can be told apart from noise.
- False statements about a brand persist whatever the brand publishes.
- An SEO-only control does as well as every AIO programme on qualified leads.
Each of these outcomes is a finding, and each one gets published.
How a tactic earns its place
A tactic moves from Phases I to III into the Phase IV trial only if all three hold:
- Its primary metric has a 95% confidence interval that clears the threshold declared in advance, against a matched control.
- The effect holds on at least three AI engines, each reported on its own.
- It repeats on a second website.
The rules are written into each Phase IV trial before it starts, so the winning tactic cannot be chosen first and justified afterwards.