Intro
Most AI visibility debates collapse the moment someone asks: “Compared to whom?”
A score against the wrong competitor set tells the brand the wrong story. Pick competitors that are too small and the brand looks dominant on paper. Pick competitors that are too large and the brand looks invisible against names it never actually competes with. Pick a different set every quarter and there is no measurement left — only mood.
The competitor set is the second locked artefact of a Capston Core baseline, right after the prompt library. This page explains how the panel is built, why it must stay stable across a cycle, and when it is allowed to move.
Why a competitor set is mandatory
An AI visibility score is always relative. The question is never “Is the brand visible?” but “Is the brand visible against the right reference?”
Without a named panel, three things break:
- The score cannot move. “Visibility went up” is meaningless if the comparison drifts.
- The framing of the answer is invisible. AI engines rank brands inside an answer. Without a panel, there is no way to know whether the brand is first, second, or buried among intermediaries.
- Commercial risk hides. If OTAs and travel media are not in the panel, the brand cannot see when it is being substituted out of the answer entirely.
The competitor set is therefore not a slide for the report. It is the reference frame against which every dimension of the AI visibility scoring system is measured.
Three categories of competitor
Capston Core panels are built from three named buckets. Each one answers a different question.
Direct competitors
These are the brands the company actually competes with on a booking decision: same vertical, same market, same price tier, comparable promise. A five-star lifestyle property in Lisbon competes with other five-star lifestyle properties in Lisbon, not with chain hotels in Munich.
Direct competitors are the largest slice of the panel. They drive the “Comparison” prompts in the library and they carry the weight of the share-of-voice calculation inside the Capston Hospitality Scorecard.
Aspirational competitors
One tier above. Brands the company would like to be compared with — and that AI engines might already be substituting in their place. A premium independent property might list two flagship lifestyle groups as aspirational; a family-run vineyard might list two cult names.
Aspirational competitors are usually two or three names. They are not used to score share of voice on a like-for-like basis. They are used to see whether AI answers are routing demand upward, and whether the brand surfaces in answers that mention the tier above.
Intermediary or OTA
The “competitors for the answer”. Not other brands but the platforms and media that occupy the named slots in an AI response: Booking, Expedia, Tablet, Mr & Mrs Smith, travel desks of major newspapers, vertical aggregators.
This bucket exists because AI engines often resolve a buying question by naming a marketplace, not a property. If intermediaries are not tracked, the brand sees a clean score and misses the fact that the answer is being captured upstream. The intermediary class is tracked as one named group inside the panel.
How many competitors to include
The working range is five to eight named competitors, plus the intermediary class as a single tracked group.
- Fewer than five and the panel is too narrow to reflect the buying decision; one outlier moves the score.
- More than eight and the panel dilutes. Prompts are not long enough to surface ten brands; weak entrants pull down the share-of-voice math without ever being a real reference.
Inside that range, Capston Core typically lands on:
- 4 to 6 direct competitors
- 1 to 2 aspirational competitors
- 1 intermediary group (named operators tracked inside it)
The panel is signed off in week 1 of the baseline, alongside the prompt library. Sign-off is explicit: name, URL, category, reason for inclusion. No verbal panels.
Why the panel must be stable
The panel is locked for the full measurement cycle.
A score in month 0 and a score in month 3 are only comparable if the reference frame is identical. Adding a competitor mid-cycle changes the denominator of every share-of-voice calculation. Removing one moves the floor. Either change invalidates the comparison.
This is also why Capston Core treats competitor changes as a documented event, not a quiet edit:
- The change is logged with date, reason, and requested by.
- The cycle is either restarted (rare) or annotated (standard): the report carries a note that the panel changed at month X and the trend line is segmented.
- The previous score is preserved; it is not retroactively recomputed against the new panel.
The discipline is the same one the Capston Core methodology applies to the prompt library and to the engine set. Stable inputs are what makes the score a measurement instead of an opinion.
When to revisit the panel
Stable does not mean frozen forever. The panel is revisited on a defined cadence and on defined triggers.
Defined cadence. Once per cycle, at the end-of-quarter review. The question is: did the buying landscape move? New entrants, repositioning, acquisitions, market exits.
Defined triggers.
- A direct competitor exits the market or is acquired and renamed.
- A new entrant takes a real share of the prompts the brand cares about, repeatedly, for two consecutive measurement points.
- AI engines start naming a brand that was not in the panel as the answer to comparison prompts.
- The brand itself repositions and the price tier or market changes.
A panel revision is a deliberate act with a written rationale, the same way prompt edits are handled in the Capston QA standards. The goal is to avoid two failure modes: a panel that drifts every quarter and means nothing, and a panel that has gone stale and no longer reflects the buying decision.
How this fits into Capston Core
The competitor set is one of the foundation artefacts of a Capston Core engagement. It sits next to the prompt library and the engine set as the three locked inputs of the Capston Core methodology. Every dimension of the AI visibility scoring system is computed against this panel; the Capston Hospitality Scorecard inherits it directly.
→ Back to Capston Core
FAQ
Can we include a competitor we admire but do not really compete with?
Yes — in the aspirational bucket, capped at one or two names. Not in the direct bucket, because that distorts share-of-voice math.
Should OTAs really be counted as competitors?
For the AI answer, yes. They occupy named slots in responses and capture demand. They are tracked as the intermediary class, separate from direct competitors.
What happens if a major competitor launches mid-cycle?
The launch is logged. The panel is not edited mid-cycle unless the new entrant takes meaningful share for two consecutive measurement points; otherwise it is added at the next cycle boundary with a documented note.
Is the competitor set the same as the keyword competitor list in SEO tools?
No. SEO competitor lists are derived from keyword overlap. The Capston Core panel is a commercial decision, signed off by the brand, that names who the company actually competes with on the buying decision.
Final CTA block
Define the panel your AI visibility score is measured against.
Define your competitor panel
Read the methodology