This is the most uncomfortable pattern in our measurements — and at the same time one of the most practically relevant. In open category questions, a small number of brands keeps returning, while other brands barely break into them — even when their content is strong and their classic visibility is in order. Inclusion threshold is the Groundbase name for that observable point: the boundary above which a brand appears consistently enough in open category questions to be in the game.
One thing this concept emphatically is not: a look inside the model. Whether a threshold or a shortlist literally exists in a model, nobody can establish from the outside — and the cause of the pattern can differ per AI channel.
What we measure is the pattern: who is structurally present in open category questions, and who is not. Why — that is a hypothesis per channel, not a mechanism we can read off.
First rule out what else it could be
Before concluding that your brand sits “below the threshold”, other explanations belong off the table. A brand can also be absent through language or localisation (the question is answered for another market), because the question triggers no web search and the answer comes from memorised model knowledge, through indexing or access problems, through the scope of the question — or through plain chance in a measurement that is too small. That is exactly why measurement comes first: repeated, across several channels, with clean variants. One absence is a data point; a pattern across channels and phrasings is a finding.
What the pattern points to
The most informative contrast: branded versus open questions. If a channel answers branded questions about you correctly and completely — so it knows you — while open category questions structurally pass you over, that points to the problem not sitting at page level. Our hypothesis, consistent with what external research into brand preference in language models finds: broad, consistent presence weighs in — how often, in how many places, how consistently and by what kind of sources a brand is mentioned in its categorical context. That is a direction, not a published formula; nobody knows the weighting, and it presumably differs per channel.
What it means for your priorities
If the pattern holds up after measurement, priorities shift — not because a model threshold dictates it, but because the measurement says so: yet more page optimisation is then not the bottleneck.
- External presence and PR become relatively more important than further on-site refinement: every publication that places you in the right categorical context builds the breadth the pattern points to
- Brand penetration across multiple channels and sources comes before polishing pages that are already good
- Wikipedia, Wikidata and authoritative registers are valuable anchors where they are attainable — the editorial standards are strict, and it is not a precondition: plenty of brands appear just fine without
- Consistent categorical terminology next to your brand name — ‘certified installer’, ‘savings bank’ — strengthens the association between brand and category in everything written about you
And the rebrand footnote, because that question always comes: a new name starts with less history in the sources, and that can temporarily depress your presence in open questions. Systems can link old and new names to each other — help them by making the link explicit everywhere, from your own site to registers and press. How quickly that recovers differs per channel; measure it, rather than counting on it.