Methodology
Depth beats frequency.
One isolated answer is not enough to measure a brand. hitak repeats observations, freezes their conditions and documents their limitations.
One answer · AI 1 Present True on that day, for that wording. 12 observations · same question Top 3 in 9 of 12 answers Corpus v3 · frozen IGPI
Global AI presence index.
A summary measurement based on a frozen corpus and repeated observations. Corpus v3 · 8 questions × 12 observations
IGPI · Your brand 41 % 39 of 96 observations One answer would say “present” or “absent”. Repeated measurement shows your brand in the top 3 of 9 out of 12 answers for “back mattress”, but only 2 for “€500 budget”.
Three principles
What makes a measurement defensible.
same question · 12 observations Repeat without moving the target
The corpus and its variations are frozen so repetitions address the same measurement question.
Separate search regimes
General-purpose assistants and web-native engines are not added together as if they produced the same observation.
State the limitations
An incompatible corpus or scope stops the comparison. A recommendation remains high, medium or marginal: never a predicted uplift.
Honesty
What hitak does not do.
No quantified uplift promised
A recommendation remains high, medium or marginal. It never comes with a predicted uplift.
No trend across a break
When the corpus or scope changes, comparison stops rather than being extrapolated.
No average across regimes
Pure AI and web search produce different observations; they remain read separately.
No invented causality
A cited source is not a demonstrated cause. Analysis describes what is observable.