AI visibility is not being remembered

What the brand recall survey found#

A survey is only as sharp as the question behind it, and the brand recall survey that Skyword fielded in August 2026 asked a good one: when a machine does the comparing, does the buyer remember who was compared? Across 1,000 online responses from adults aged 18 and older, the answer is mostly no. A fox knows the difference between crossing a trail and leaving one, and this survey measures the second.

The reach is enormous. PR Newswire carried the release with a Boston dateline of 6 October 2026: 70% of respondents have used AI tools such as ChatGPT, or encountered AI-generated summaries, when researching a company or product, and 61% have relied on those tools or summaries at least sometimes when deciding whether to buy from one company over another. Those are the visibility numbers, and they are vast. The recall numbers that follow them are the reason this piece exists.

Less than half, 42%, remembered specific details about the company or product, and a further 32% remembered general information without remembering which company it came from. Only 26% remembered the company or product name. Put plainly: a chatbot played matchmaker, and three in four buyers could not name the host it introduced.

The split is not spread evenly across the pack. CustomerThink reports that nearly 70% of Gen Z consumers said AI-generated sameness made it difficult to choose between companies, against 17% of Baby Boomers. Ask whether a distinctive voice helps a company stand out and the gap repeats at 58% to 21%. Younger buyers are both the most exposed to sameness and the hungriest for a signature.

Being cited is not the same as being remembered#

There is a comfortable story about AI visibility, and it runs like this: get cited by the answer engine, climb the mention charts, and the brand takes care of itself. The brand recall survey says otherwise. Skyword's own report page leans on the same 26% and the same generation gap, and its framing is blunter than the wire copy. A citation is a mention a machine repeats. Recall is a memory a person keeps, and only one of those comes back when the buyer is ready to spend. A fox follows a scent, not a template.

The two jobs pull in opposite directions. Visibility rewards volume: more pages, more formats, more surface area for a language model to lift a sentence from. Recall rewards difference. A page that says exactly what every other page says is easy to summarise and almost impossible to remember, and the waffle below is the share of buyers who proved the point.

That is the pivot. Andrew Wheeler, Skyword's chief executive, argues in the AI Journal that AI-enabled volume does not build category authority on its own, though it can scale the human insight that does. Read that as a warning to the content mill and a brief for the writer: the machine can carry a distinctive idea further, but it cannot supply the distinctiveness in the first place.

The prior evidence points the same way. An earlier Skyword study, covered by ContentGrip, found that when AI gets brand information wrong consumers look beyond the brand. Trust that depends on a summary is trust the brand does not own.

The sameness trap behind AI visibility#

Why does recall collapse so neatly? Because the inputs are the same. When every team briefs the same models on the same subjects and asks for the same tone, the outputs converge, and buyers notice. Nearly half, 49%, told the survey that different companies' websites, social posts or emails frequently sound similar to one another, per MediaPost and The Wise Marketer. Strip the logos and the whole warren looks identical, which is precisely the sameness the survey set out to measure.

The same research hints at what cuts through. Asked what made a website chat assistant feel like the company rather than a generic bot, the most-cited answer was company-specific knowledge, at 36%. Only 25% cited the logo, colours and design, and just 13% pointed to voice and tone. The lesson is unfashionable. A buyer recognises a brand by what it knows, not by how it looks.

There is a visual version of the same trap. A clean design is often just a safe one, and safe is the screenshot of a sentence any competitor could have written. Polish without a point of view is how a brand becomes beautifully forgettable.

• 26%

  remembered the company or product name

• 42%

  remembered specific details, brand aside

• 32%

  remembered general points, not the company

• 45%

  struggled to choose because answers sounded alike

How to measure brand recall#

Measuring brand recall used to mean one annual survey and a budget line. Now the habit has to be constant, because the comparison happens in private and the answer arrives without a click. A fox does not count every pawprint; it reads the ones that lead somewhere. Google's guidance on people-first content makes the same point in search language: write for people first, and the systems follow.

The practical upside is that recall, unlike visibility, bends to tools you already own. A short panel survey, a blind test, a monthly sweep of the summaries your category produces: none of it needs a new platform. What it needs is the discipline to ask the uncomfortable question, and the honest answer is usually that the brand is louder than it is remembered.

Turning visibility into recall#

So the strategy is not to retreat from AI visibility. It is to feed that visibility something worth remembering. Differentiation is the input and recall is the output, and the brand equity you build is the sum of the two over time. That is the cunning argument underneath the whole brand recall survey.

In practice that means fewer pages with more of you in them. One proprietary dataset beats ten rewrites of the same listicle. One named point of view beats a hundred balanced summaries. The survey's 36% cell is the tell: buyers bonded to company-specific knowledge, and that is the one input a competitor's model cannot copy.

Where the standing problem is interchangeable output, the fix is older than any model: a position no one else can occupy, expressed in a voice no one else can mimic. folkfox brand strategy is the lever, and our content marketing service builds the programme around it. Where the machine itself has to be briefed carefully, our AI consultancy starts from the workflow rather than the tool, and our SEO and GEO work handles the citation layer. If you are weighing where the budget belongs, our pricing is the plain version.

Two adjacent stories from this edition press the same nerve. How a streaming royalty is calculated and Google's business-name policy change both show a category where the rules, not the volume, decide who is remembered. Sameness is the default now. A signature is the work.

Frequently asked questions#