Two questions. "Which business laptop brand is most reliable?" and "Most reliable business laptop."
Same buyer. Same intent. Same question, really, phrased two ways.
The first returns five brands. The second returns four. Acer is simply gone - and if it were your brand, nothing in your reporting would tell you.
I ran twelve prompts like these in July, benchmarking the UK business laptop market for a media partner across five tracked AI engines over fourteen days. 140 brand mentions, 205 cited URLs.
Ask which brands are best for business laptops and the whole field turns up: Lenovo, Dell, HP, Asus, Acer. Then the buyer gets specific, and the field quietly changes shape. Across the eight questions naming an attribute or a use case - battery life, screen quality, remote work, under £1,500 - Acer isn't detected once. Not a single appearance. Dell holds complete coverage on the general and value questions and half of it on display, battery, weight and price. HP disappears from remote work altogether.
Here's why that matters, and it's the reason I've stopped using the phrase "AI visibility". Every one of those five brands would pass a visibility check. All five get mentioned. All five would show green on a tracking dashboard. Only three are ever really in contention, and which three depends on what the buyer happens to type.
That's not randomness, either - different questions pull from different sources, and a brand that's missing from the sources a question reaches for is missing from its answer. Acer didn't drop out of eight answers by bad luck.
The second finding surprised me more: where the answers were coming from. It was almost never the brands' own websites. Across those 205 cited URLs, the four reported brand domains accounted for just eleven citations between them - everything else came from third parties: tech media like TechRadar, RTINGS and PCMag, and YouTube. One third-party article was cited 25 times, on its own more than all four brand domains combined.
It's a hybrid process, and that's the part people keep getting wrong
Laptops are a small, well-defined category, but what happened in those twelve prompts is happening across B2B. A buyer asks. A machine assembles a shortlist from sources the vendors mostly don't own. The vendor never sees the question, the answer, or the moment they fell out of it.
AI doesn't start the decision and it doesn't close it. It compresses everything in between.
Buyers arrive with a field already in mind - 85% buy from their 'Day One List', the vendors they had in their heads before any research began (Bain). They finish with a human: 69% take AI's output to a sales rep specifically to validate it (Gartner, 2026).
What's changed is the middle - where the research actually happens. It's moving out of the search box and into the chat window: Forrester's State of Business Buying 2026 puts genAI searches as the starting point for B2B buyers, and among software buyers specifically, 51% now begin their vendor research in an AI chatbot more often than in Google - up from 29% eleven months earlier (G2, The Answer Economy, April 2026).
Which is why the laptop pattern generalises. The same G2 survey found a third of software buyers bought from a vendor they'd never heard of before. The Day One List still dominates - but AI is now where that list gets reopened, and whether that helps or hurts you depends entirely on what the machine finds when it looks.
Most budgets are pointed at the wrong end of this
Across the clients I work with and the businesses I talk to, I see four patterns.
The first group has doubled down on bottom-of-funnel lead generation. More of the briefs I've seen over the last twelve to eighteen months ask for lower-funnel leads, which is a knee-jerk response to slowing pipelines and tactics that have stopped working. The problem is that a bottom-of-funnel lead is captured after the shortlist has been drawn. You're paying to harvest a decision made further upstream.
The second has embraced brand-to-demand but runs the two as separate disciplines - money into awareness on one side, money into lead gen on the other, and no bridge between them. That bridge is now where the decision happens.
The third is the most sophisticated, and often the most established. Full funnel, properly connected, activating across awareness, education, consideration and conversion. Executed well. Against a model of buying that has moved.
The fourth has noticed AI and is buying visibility tracking. Are we cited, yes or no, tick. Acer would pass that test. I'm also watching SEO budget migrate to agencies repositioned as AEO or GEO, on the assumption it's the same job with a new label.
Sales hasn't caught up either. Almost every sales organisation I speak to is still built around persuasion, and the penny hasn't dropped that the job has shifted to validation - confirming a decision the buyer has largely already made. That's a harder shift than it sounds, because it runs against everything a good salesperson has been trained and measured on. But the conversation they're walking into has already happened without them.
So here's the uncomfortable version. Almost every one of these teams is funding a stage that no longer decides anything - and the better run the operation, the more efficiently it does it.
The numbers say where it went. Buyers now reach the first vendor conversation earlier than they used to, at 61% of the way through the journey rather than 69% (6sense, 2025). That sounds like more room to influence. It isn't. The research that used to fill that gap hasn't gone away, it's been compressed and moved somewhere you can't see. Seven in ten software buyers say AI surfaced information that changed which vendor they chose (G2, 2026), and the vendor a buying group favours before first contact wins 77% of the time (6sense, 2025).
The favourite gets crowned during the comparison. That's the phase almost nobody is funding.
Demand isn't going away. It's just going elsewhere.
AI is redirecting it towards the brands it can see, understand and credibly stand behind.
Go back to the laptop market and look at where Lenovo's lead actually sits. Not on Lenovo's own site - seven citations there, 2% of the total. It's held in tech media Lenovo doesn't own and can't control. That cuts both ways. It's a strong position built on ground someone else owns, and it's a position nobody is seriously contesting: eleven citations across four brand domains says nobody is really trying.
Market share takes years to shift. Which sources a machine trusts when it answers about your category is a faster problem, and in UK business laptops it's a wide-open one. That won't stay true in every category for long.
The cost of doing nothing isn't losing the deal. It's never being in the set the deal gets decided from.
That's the work I've been mapping as The Selection Layer: whether B2B brands get included in AI answers at all, how they're described when they are, and whether they get preferred over the alternatives. This newsletter runs fortnightly - benchmark findings from real markets, and what they mean for how B2B teams go to market.
If you want to know where you stand today, the AI Shortlist Scan is the place to start: how AI represents you, how you compare, and how it arrives at the preference it hands your buyers. £1,995, delivered in seven days, money-back guarantee.
The full evidence behind the buying journey, every number sourced, is here: How B2B Buyers Actually Buy in 2026
John