The uncomfortable number in this year’s WordPress VIP consumer study is not that seventy-four percent of Americans think the internet feels less human than it did ten years ago. It is the one right next to it: sixty-one percent of the same 1,200 respondents cannot name a single brand using AI well, and sixty percent read AI in a brand’s messaging as a turnoff rather than an asset. The market a board is being asked to invest in is one where a majority of the audience is either indifferent or actively cooling on it.
I have watched enough technology waves land to know how this one resolves. The brands that survive it are not the ones shipping the most AI features. They are the ones whose site, catalogue, structured data and editorial workflow quietly hold up when an assistant asks a question about them. That is a boardroom decision about where money goes over the next two budget cycles, not a technical footnote.
The catch-22, in Brian Solis’s own words
Brian Solis, Head of Global Innovation at ServiceNow and author of Mindshift, has covered every major digital shift of the last twenty years from close range. His framing, given to WordPress VIP for their 2026 research, is the cleanest statement of the catch-22 I have read this year:
“No customer or user wakes up and says, ‘I hope I get to talk to a chat bot or an AI agent today.’ Human-centered design is truer today with artificial intelligence. Ironically, the answer is using AI to be more human.”
Brian Solis, Head of Global Innovation, ServiceNow
The entire problem sits in those three sentences. Buyers are not asking for AI. They are asking for what AI, done well, might make possible. Organisations that mistake the first for the second spend real money on features their customers register as friction. On his own site, Solis puts it more directly: “We’re using AI in the way that we’ve used every technology revolution — the use of technology to get further away from people.”
The exit is not a slogan. It is an operating stance and a set of quite specific things you build.
Size the problem before you spend anything on the solution
Two industry numbers, both from primary sources, define the shape of the decision.
The first is that a majority of marketing leaders cannot measure whether their brand is visible in AI-generated answers. In the Semrush 2026 AI Visibility Index, published on 26 June 2026 after analysing 126 million U.S. AI search prompts, forty-five percent of marketing leaders said they could not accurately measure visibility in AI-generated answers, and only nine percent had tools that track the relevant metrics across platforms. Whatever conversation your team is having about “our AI strategy”, the baseline is likely not yet measured.
The second is the traffic shift underneath it. The same index reports a 1,324% rise in AI-driven traffic to U.S. retail sites and 2,215% to travel sites between October 2024 and May 2026. The absolute volume is still a fraction of organic search. The direction is not in dispute.
Between those numbers sits Solis’s point. The audience is arriving through a new channel, most brands cannot see whether that channel is picking them up, and the majority of consumers who do see AI in the messaging read it as a warning sign. Three levers respond to that shape. One lever gets pitched constantly and does not yet work. Take them in order.
Lever one: structure the site so an assistant can quote it accurately
Solis has a formulation that reads well in a boardroom: “Experience is actually an emotion. It’s not a series of transactions. It’s not technology.” The consequence is uncomfortable, because the technology part is where the money goes. An assistant that misquotes your product, price, policy or position is delivering a small brand injury, on repeat, to people you never see arrive.
The concrete work is plumbing. Structured data on every product, every article, every organisational page. A machine-readable inventory of what the brand sells, believes and has said. This is not a content-marketing project or an SEO project. It is a data-quality project with a governance layer.
The clearest published example is Patagonia. In the Semrush 2026 index the brand held an AI visibility score of roughly 79–80 across the study period, and the reason Semrush’s analysts give is worth reading twice: the consistency of how the brand is described in third-party sources including OutdoorGearLab, REI, Switchback Travel and GearJunkie. The lesson is not that Patagonia is good at prompt engineering. A decade of coherent language about who the brand is, in places the brand does not own, is what feeds an assistant a confident answer. Retrofitting that is a two- to three-year job. Losing it takes one bad campaign.
Lever two: put a human governance step in front of every AI-touched surface
Solis frames this one plainly. “To be customer led is to be empathetic. And to be empathetic is to see the world as your customer sees it and let it inspire you.” That is not a soft observation. It is the argument for why the last hand on the content is a named person with an editorial title, not a pipeline. WordPress VIP and Human Made frame the same idea as a five-pillar readiness model — content architecture, platform flexibility, governance and control, operational control, strategic alignment — and pillar three sits under a business risk, not a technology risk.
The operational form of this is undramatic. A named editor signs off on any AI-drafted piece before it publishes. Every product change that flows through an AI-assisted process leaves a log entry with a human on it. The organisation can answer, on any given piece of published content, who approved it, what the model contribution was, and when. When a regulator or a customer asks, the answer is a system output rather than a defence.
Third-party evidence that this pays for itself is easier to find than the vendor decks admit. In its June 2026 AI Visibility Index, Semrush reports that eighty-one percent of organisations with an integrated SEO and AI-visibility workflow saw traffic or leads increase, against thirty-six percent of organisations that ran the two separately. The integration is not a piece of software; it is a governance stance about who owns the answer that leaves the building.
Lever three: measure whether an assistant can find you, then measure again next quarter
Solis’s line here is the one I would put in a memo: “If you’re waiting for someone to tell us what to do, we might be on the wrong side of innovation.” Nine percent of marketing leaders in the Semrush study had tools that track AI visibility across platforms. The rest are guessing whether last quarter’s content programme moved the needle, or being sold a fix for a symptom nobody has measured.
The measurement stack that holds up is unglamorous. A monthly panel of the prompts your customers really use, tracked across ChatGPT, Google’s AI Overviews, Perplexity and whatever the emerging fourth is next quarter. A baseline of citation rate per prompt and per brand. A view of referral traffic in GA4 for the assistants that pass a referrer, and honest acknowledgement that some do not. A share-of-voice rate against the two or three brands the executive genuinely competes with. Everything else is texture.
Run it quarterly, not continuously. The engines change. The indexes change. A daily read is noise; a quarterly read is a trend. WordPress VIP’s research puts the average weekly time an enterprise team invests in improving AI visibility at 16.6 hours. If that number is being spent in your organisation and no one can produce a chart of what it moved, the problem is not effort. It is that the effort has no scoreboard.
The lever that does not work yet, and why
Agentic checkout — the idea that an assistant does the shopping and hands the transaction back clean — is a pitch we hear on every discovery call this year. Outside a small set of provider-issued demos, it is not yet a channel a mid-market or enterprise brand should build the store around. API surfaces are unstable, attribution is muddy, and the customer-side experience for anything past “reorder my usual” ranges from awkward to broken.
Solis reads this pattern the same way. He writes that most organisations use every technology revolution “to get further away from people”. Agentic commerce is exactly that shape of temptation. Some of it will land quickly in narrow categories. Building the whole customer relationship around it in 2026 is the mistake I would most like to talk clients out of.
The pragmatic move is to keep product data and structured metadata clean enough that when agentic checkout does mature — probably in one vertical before the others — the brand can be present without a re-platform. The same feed that keeps the site readable to assistants makes it readable to agents when the standard settles.
What we would build, inside a real WordPress VIP install
The build section is the piece most operators want and least often get from a strategy piece. Here is what an AI-visibility programme actually looks like inside a real enterprise WordPress VIP environment, taken from the shape of work my team runs. None of it is exotic; the exotic version is what fails.
One: a canonical brand entity, expressed once and reused everywhere. Organisation schema, product schema and article schema, all pointing to the same set of core facts about the brand and its offerings. On WordPress this lives in a small block of theme JSON plus a handful of schema plugins wired to real post-type fields, not to guesswork. On WordPress VIP the deploy discipline keeps that block from drifting between environments, which is the failure mode we see on managed hosting: a good schema on staging, a stale one in production, and no way to prove which one Googlebot actually saw last week.
Two: a human-in-the-loop editorial workflow that survives audit. AI-drafted content routes through a named editor, who approves against a checklist the legal team has signed off on. The approval, the model contribution and the editor’s identity land in an audit log the CMS retains for the retention period the sector requires. On VIP this is a governance conversation as much as a code one; the platform’s audit trail is the argument for why regulated sectors put content operations on it in the first place.
Three: a machine-readable inventory of what the brand sells and says. A products feed and a content feed, both structured, both refreshed on a schedule the ops team owns, both linkable at stable URLs. For a WooCommerce catalogue behind VIP this is a two-week job to specify and a rolling operational commitment thereafter. It is the piece that most decides whether an assistant quotes your product accurately or a competitor’s approximation of it.
Four: a measurement panel your board can read. A monthly report that shows, for the prompts your customers really search, whether the brand was cited and by which engine, and how that moved against a locked baseline. On VIP this sits on top of Parse.ly and the platform’s analytics; on a self-managed enterprise setup it is a bespoke pipeline. Either way the deliverable is one chart the CFO recognises. Sixteen-and-a-half hours a week is a lot of hours to spend against no chart.
Five: honest coverage of what is not yet measurable, in writing. Some assistants do not pass a referrer. Some cite without linking. The measurement panel names those gaps rather than papering over them. A programme that admits what it cannot see is far more useful than one that pretends the picture is complete, and it is what a mature buyer expects from an agency in year one of a technology shift.
The question worth putting to your team on Monday
If sixty-one percent of your customers cannot yet name a brand using AI well, the opportunity is that the category has no incumbent. WordPress VIP’s own 2026 research puts this plainly: there is no established market leader. The organisations that are going to matter in two years are the ones setting up the plumbing this year, quietly, while the pitch decks fight about vocabulary.
The question I would put to a marketing lead or a CTO is not “what is your AI strategy”. It is narrower. What is the last piece of AI-touched content your organisation published, who was the named editor who approved it, and can you produce the audit entry in the next ten minutes? If the answer is a shrug, the plumbing is missing and no tool spend fixes that. If the answer is instant, the rest is a matter of scale, not principle.
We put this stack in on enterprise WordPress VIP engagements and, at the mid-market end, inside managed WooCommerce environments where the compliance floor is lower. If your team is working out where AI visibility earns its keep this year, that is the conversation we would rather have than a slide about the category.
Last modified: September 18, 2026
United States / English 
Slovensko / Slovenčina 
Canada / Français 
Türkiye / Türkçe