Nobody can answer this cleanly, which is the honest starting point. There is no established cost per mention, no agreed conversion benchmark, and no way to isolate the revenue that followed from an assistant naming your company to someone who then arrived a week later through a branded search.
That does not make the question unanswerable, only imprecise. You can bound it from several directions, and doing so is more useful than either the confident claims or the blanket dismissals. Anyone allocating budget toward AI Brand Mentions needs at least a rough valuation, and here is how to construct one you can defend internally.
Why the Usual Comparison Understates It
The instinct is to treat a mention like an impression. That comparison is misleading in both directions and worth taking apart.
A search impression fires whenever your listing renders, including positions nobody reads. Ten thousand impressions might represent a few hundred people who actually registered your existence.
A mention in a generated answer sits in a set of two to five companies, presented as a considered response to a specific question, usually with a short justification attached. The person asked, received a short list, and read it. The engagement is closer to a shortlist placement than to a display impression.
But the comparison overstates in one respect too. That person may be researching casually, months from a decision, or not a buyer at all. Volume is far lower and intent is far more variable, so treating each mention as a qualified lead is equally wrong.
Three Ways to Put a Number on It
None is authoritative. Together they produce a range that survives scrutiny.
Method one: substitution cost. What would it cost you to reach the same person through paid channels at the same moment of intent? Take your cost per click on your highest intent commercial terms, then adjust for the fact that a mention arrives with an implicit recommendation attached rather than as an advertisement. Conservative adjustment is to treat the two as equivalent and no better.
Method two: shortlist economics. If your category typically produces a shortlist of three or four providers before a decision, and you know your close rate from shortlist inclusion, you have a rough value per shortlist appearance. A mention is not identical to being shortlisted, but it is the same functional event: you became a candidate.
Method three: branded search lift. The most defensible and the slowest. Track branded query volume and direct enquiries against mention growth with a two to four week lag. It is correlation rather than proof, but across two quarters it produces a pattern that finance departments accept more readily than a per mention estimate.
Run all three and take the lowest. A range you can defend beats a single number you cannot.
Why Not All Mentions Are Worth the Same
Averaging across every mention destroys the useful information, since the spread between the most and least valuable is enormous.
- Highest value: appearing in answers to buying questions, with a justification that matches how you actually differ from competitors
- High value: appearing on comparison questions between you and named alternatives, since the buyer is close to deciding
- Moderate: appearing on orientation questions where the buyer is still working out what they need, which shapes criteria but does not close
- Low: appearing on definitional or informational questions with no purchase context attached
- Effectively zero: appearing in answers to questions your customers never ask, which is what happens when the question set was assembled from search volume rather than sales conversations
That last category is where reported mention counts get inflated. Total volume rises, commercial value does not, and the report looks better than the pipeline.
The Multiplier Nobody Accounts For
One factor makes mentions worth substantially more than the arithmetic suggests, and it applies particularly in considered purchases.
An assistant naming three companies is not just exposing you to a buyer. It is defining the consideration set. Companies absent from that answer are frequently absent from the decision entirely, because the buyer never learns they exist and has no reason to search further.
So the value is partly what you gain and partly what competitors lose. In a category where four providers could plausibly serve a request and assistants consistently name three, the fourth is not slightly disadvantaged. It is structurally excluded from that research path.
That asymmetry does not show up in any per mention calculation, and it is the strongest argument for treating this as a competitive necessity rather than an incremental channel.
What the Growth Curve Tends to Look Like
Valuation is easier once you know the shape of the accumulation.
Ken Ganley Kia, an authorized dealership in Ohio, had strong offline recognition and minimal AI presence. The work focused on conversational query patterns and produced a large increase in ChatGPT presence. Low base, steep climb, and the value of each early mention was high because the alternative was total absence.
EC Council, the global certification body behind the CEH and CND programmes, sat at the other end. Already globally recognized with a large content estate, six months of technical restructuring produced substantial AI Overview presence. Higher base, larger absolute gains, and lower marginal value per additional mention because presence was already partially established.
The pattern worth internalizing: early mentions in a category where you are absent are worth disproportionately more than later mentions once you are established. Which argues for acting when you are invisible rather than waiting until the channel is proven, since the cheap gains sit at the start.
Where the Valuation Argument Legitimately Fails
Three situations where the numbers do not support the spend, stated plainly.
Your buyers are not researching this way. Adoption varies by sector, region, and buyer profile. If enquiries come predominantly through referral or repeat business, mention value is genuinely lower for you than for a company selling to buyers who research independently.
Your category resolves on price alone. Where the decision comes down to a comparable quote, being named earns you a place in a bidding process rather than a preference. Still worth something, worth less than in categories where trust drives selection.
You cannot convert the attention. A mention sends a buyer to a website. If that site does not convert well, you are paying to fill a leaky container, and the conversion work is the better investment first.
The first of those three is testable in an afternoon. Ask your last ten customers how they found you, and if none mention an assistant, treat the valuation as speculative for now and check again in six months.
What to Actually Track Instead of Total Mentions
Total count is the metric most likely to be reported and least likely to be useful.
Better inputs to a valuation:
- Mention rate on commercial intent questions specifically, ignoring informational ones entirely
- Share of relevant answers rather than absolute counts, since the denominator shifts as platforms change how often they generate answers
- Consistency across repeated tests of the same question, since fluctuating presence is worth far less than reliable presence
- Branded search and direct enquiry movement on a lagged basis
Independent measurement matters here more than in conventional search, because vendor composite scores blend inputs with undisclosed weighting and cannot be audited. Reviewing what the current chatgpt seo trackers actually count is worth an hour before you build a business case on top of any of them.
Build the range from all three methods, present it as a range, and state the assumptions openly. A defensible estimate with visible reasoning survives a finance review. A confident number that nobody can reproduce does not.
