A year ago, AI in the buying journey was still a debate. Brands were rushing to deploy it, consumers were openly resistant to it, and the conversation in marketing and CX rooms centred on whether the trade-off was worth it.
Twelve months later, the debate has moved on, mostly because the AI got better. It's faster, more polished, more useful. So good, in fact, that seven in ten UK consumers can't reliably tell when they've been talking to it. And the share who say their last AI interaction made things worse has dropped sharply year-over-year.
This is the story of a technology that crossed a threshold quietly. The brands that deployed AI well in the last year have been rewarded with consumers who can't tell the difference between AI and human, and who often don't care. The brands that deployed it poorly have been punished, and not in the way you might expect. When AI interactions go wrong, UK consumers don't blame the AI vendor. They blame the brand that chose to deploy it. By more than three to one.
That's the headline of the 2026 B2C Buyer Experience Study, and it changes the calculus of AI investment for marketing and CX leaders. AI is no longer a futuristic add-on or a risky experiment. It's table stakes. The question is no longer whether to use AI in the customer journey. It's whether your AI is good enough to protect your brand, identify itself when it should, and hand off to a human when it can't. The brands that get this right are going to win the next chapter of the AI era. The brands that get it wrong are going to lose customers without ever knowing the AI was the reason.
Read on for what consumers told us, what's changed since 2025, and where the data points your CX investments next.
When consumers are making high-stakes purchase decisions, there are many touchpoints along the way where a brand's AI can step in. Last year, the most common feeling about that AI was somewhere between neutral and negative. This year, the most common feeling is no feeling at all. 38% of UK consumers say interacting with brand AI made no difference to their experience, the modal response and a touch above 2025. The share who said AI made things worse has dropped from 24% to 20%, a four-point improvement that suggests something meaningful has changed in how AI shows up in the customer journey.
42% of UK consumers say brand AI made their buying experience better. 38% say it made no difference. 20% say it made the experience worse, down from 24% in 2025.
This is the central thread of the 2026 data. The negative sentiment that defined the 2025 conversation about brand AI has softened across nearly every measure. Fewer UK consumers feel less valued. Fewer say they were forced to interact with AI most or all of the time. Fewer say their buying experience suffered. And yet, the share who say AI made their experience meaningfully better has only ticked up modestly. The bulk of the movement is from "worse" to "neutral." AI is no longer making a bad impression. It's making no impression at all, which, for an embedded technology, is often the goal.
For brands, the implication is that the bar for AI deployment has shifted. A year ago, the question was whether to deploy AI without alienating customers. Now, the question is how to deploy it well enough that customers don't notice, and how to make sure that when they do notice, they're glad they did.
The average masks dramatic differences across industries. UK Healthcare leads the AI experience, with 66% of consumers saying it made things better and only 9% saying it made things worse. UK Insurance is the laggard, where just 35% say AI made the experience better, 24% say it made things worse, and 41% say it made no difference at all. That's a 22-point spread on positive sentiment between two regulated industries operating in the same market. Financial Services is the second-best at 57% positive. Automotive, Telecommunications, and Travel cluster in the middle. The brands deploying AI well are creating a meaningfully better customer experience. The brands deploying it poorly are creating a meaningfully worse one.
The most striking finding in the 2026 data is also the simplest. When we asked UK consumers whether they had ever realised, after the fact, that an interaction they thought was with a human was actually with AI, only 30% said yes. The other 70% either said they have never had that realisation, or weren't sure. Fewer than one in three UK consumers can confidently say they've ever caught a brand's AI in the act.
70% of UK consumers either didn't realise they were talking to AI, or aren't certain they would have noticed if they were. Only 30% have definitively caught a brand's AI in the act. 57% say they've never been fooled. 13% say they aren't sure.
This is a leading indicator that the underlying technology has crossed a quality threshold. A year ago, brand AI was identifiable by its stiffness, its scripts, its refusal to deviate from preset paths. Today, voice and text AI is good enough that the majority of UK consumers can't reliably distinguish it from a human agent. That's a win for the technology and a win for the brands that deployed it carefully.
It's also a risk. Consumers who eventually learn they've been misled, even by a competent AI, lose trust in the brand. And our data is clear about which way that trust falls. When the experience is good, consumers often don't know they were talking to AI. When the experience is bad, they figure it out fast.
We asked consumers a question we hadn't asked before: when an AI interaction with a brand goes badly, who do you primarily blame? The result is a wake-up call for any CMO or CX leader who treats AI deployment as a vendor decision.
46% of UK consumers blame the brand alone when AI fails. 14% blame the AI technology itself. 23% blame both the brand and the AI equally. 17% don't blame anyone.
Blame the brand outpaces blame the AI by more than three to one. Add the consumers who hold the brand at least partially accountable, and roughly seven in ten UK consumers will tie a bad AI experience back to the company that chose to deploy it. The vendor takes none of the heat. The customer's relationship is with the brand, and so is their disappointment when the brand's AI lets them down.
This reframes AI deployment as a brand-equity decision, not a procurement decision. The question is no longer "what's the cheapest AI we can plug in." It's "what's the AI quality bar that protects our brand reputation." For high-stakes purchases, where the buying journey already carries emotional weight, the answer is "high."
UK consumers have made up their minds about whether AI should identify itself in customer interactions. They want it to.
82% of UK consumers say it matters that a brand's AI clearly identifies itself as AI. 54% say it matters a great deal. 28% say it matters somewhat.
Only 6% of UK consumers say it doesn't matter to them. This is one of the most directional findings in the 2026 dataset, and it lines up with the legislative direction across the EU and in UK consumer-protection guidance, where AI disclosure rules are advancing quickly. UK consumers are not waiting for regulators to force the issue. They already expect disclosure.
For brands, this is a low-cost, high-trust signal to deploy now. Telling a customer they're talking to an AI doesn't reduce satisfaction. The data shows that the share of consumers who feel positive about AI interactions actually rises when they know what they're interacting with up front. The risk is in the opposite direction: a consumer who realises after the fact that they were fooled becomes a consumer who blames the brand, and 10% of UK consumers told us that being told up front when they're talking to AI would be the single biggest improvement to their AI experience overall.
A year ago, 59% of UK consumers said they felt forced to interact with a brand's AI most or all of the time. This year, that figure has dropped sharply to 51%, and the share who said "all the time" specifically has fallen from a substantially higher baseline. The shift mirrors the broader sentiment improvement: AI is showing up just as often, but it's bothering UK consumers less.
51% of UK consumers say they feel forced to interact with AI most or all of the time, down from 59% in 2025. 96% say it happens at least occasionally.
The "at least occasionally" number is the one to focus on. AI is now functionally inescapable in the buying journey for 96% of UK consumers. The shift in sentiment isn't because brands have backed off. It's because the AI got good enough that being routed to it doesn't feel like a wall anymore. UK consumers are still being sent to AI. They've just stopped resisting because the experience improved.
The risk for brands is complacency. The data shows UK consumers feel less valued by AI than they did last year only by a small margin, and the slope hasn't reversed. 28% still say AI interactions make them feel somewhat less valued, 15% feel much less valued, and only 26% feel more valued. When AI is the front line, value perception is fragile, and 18% of UK consumers told us that more personalised interactions would be the single biggest improvement they could see.
We asked UK consumers when they actually prefer AI to a human, and the answers are remarkably consistent with last year. Simplicity and speed win.
46% of UK consumers prefer AI when the task is simple. 43% prefer AI when they need fast answers. 32% prefer AI to avoid waiting on hold. 17% prefer AI when they don't want to talk to someone. 21% never prefer AI over a human.
The "avoiding hold" finding is the one that should move investment decisions. Roughly a third of UK consumers will choose AI specifically to dodge a hold queue, which means brands without a competent AI front line are losing the moment a consumer chooses to call. AI is the consumer's preferred alternative to bad customer service, not the consumer's preferred alternative to a human.
On the other side of the question, UK consumers told us what AI is worst at. The top answers: understanding context or nuance (47%), solving complex issues (41%), and providing empathy (35%). These are also the moments in a high-stakes buying journey that determine whether a sale happens. AI handles the easy turns. Humans still close the deal.
If there's a single piece of data in this report that should change a marketing budget, it's the gap between what consumers expect and what they get.
53% of UK consumers expect a business to respond within one hour after they fill out a form. 33% actually receive a response in that window. 79% say they'll switch to a competitor that responds faster.
That's a 20-percentage-point gap between expectation and delivery, and a 79% willingness to walk if a faster option appears. This is the consumer-side proof point for the speed-to-lead investments that high-performing marketing teams are already making. Brands that respond in the first minute see meaningful conversion lifts in Invoca's customer data. The 2026 consumer data explains why: nearly four in five UK consumers are already prepared to leave for the brand that gets to them first.
What do UK consumers do when the response is too slow? 40% try to contact the brand again, but 29% move to a competitor and another 3% give up on the purchase entirely. 28% wait it out. The cost of a slow response isn't a delayed conversion. It's a lost one.
For all the talk of digital-first buying journeys, UK consumers still pick up the phone when they need help with a high-stakes purchase, though the margin is tighter than it has been. 36% prefer calling when they need to talk to the brand, down from 41% in 2025. Email holds an unusually strong second position in the UK at 20%, well above what we observed in 2025, and a notable cultural divergence in the way UK buyers prefer to escalate.
36% of UK consumers prefer calling when they have a problem and need help. 24% prefer in-person assistance. 20% prefer email; 10% prefer online channels. 5% prefer an AI assistant or chatbot.
Transactions tell a very different story, and the UK pattern diverges from the US. When it's time to actually complete the purchase, UK consumers lean digital first. 34% prefer online, the modal answer. In-person comes a close second at 30%. Calling drops to 17%, less than half its share for help. Email is unusually strong here too at 11%, well above what's typical in the US. UK consumers research online, call for help, and then complete the transaction online or in-person at roughly even rates.
The 30% in-person preference matters. It means that even though UK consumers favour the digital transaction, nearly a third still want a physical moment when they commit, and that share rises sharply with age. A UK consumer who has spent weeks researching online may still walk into a branch, dealership, or store to close. The implications are practical: the in-person experience is not a vestigial channel even in a digital-led market. It's where the second-largest share of UK transactions happens, and where older consumers concentrate.
The age gradient runs in the same direction as the US, but is shallower at the older end. 46% of UK Boomers prefer calling for help, and 35% prefer in-person for the transaction. Gen X tracks closely. Younger UK consumers split more evenly across channels, with online and email taking meaningful share where the US sees a clear phone-and-in-person duopoly. For UK brands, the takeaway is that the in-person moment matters most at the older end of the customer base, while younger UK consumers can be served well by digital-first transaction journeys that still preserve a phone option for help.
When UK consumers do call a business during a high-stakes purchase, they're calling for information. 47% called to gather more information about the product, service, or provider. 36% called to gather information about the purchase process. 24% called because the information they needed wasn't available online, a stat that has barely moved in three years despite continued investment in digital experiences.
Closing the online-information gap is one of the highest-leverage CX investments a brand can make. Every call driven by a missing fact on a website is a call that didn't need to happen. The information that does get a consumer to call is also the information they're most ready to act on, which means the call is the highest-intent moment in the journey.
We asked consumers whether they had used a generative AI tool, like ChatGPT, Gemini, or Claude, to help research a high-stakes purchase. The result is one of the biggest year-over-year shifts in the dataset.
The gap between brand AI and generative AI is closing fast on the consumer side. A year ago, generative AI was something younger consumers experimented with. This year, it's a default research step for most adults. The top use cases in the UK: getting a quick summary of options (26%), comparing different companies or brands (25%), and better understanding complex topics (20%). UK consumers are arriving at the brand more informed than they were a year ago. Some of that information will be wrong, hallucinated, or out of date. All of it shapes the buying decision before the brand gets a chance to engage.
This is a marketing problem as much as a sales one. Brands need to know what generative AI says about them, because their prospects already do.
The biggest generational shift in this year's data isn't among the digital natives. It's among the holdouts.
22% of UK Boomers have used generative AI to research a high-stakes purchase. A year ago, that figure was in the single digits. UK Boomers have caught the same wave that US Boomers did, and they were starting from a lower baseline.
UK Gen X moved nearly as much in directional terms. 39% of UK Gen X consumers used gen AI for research, a notable gain over 2025. Millennials and Gen Z continue to lead in absolute terms at 67% and 65% respectively, but they were already there a year ago. The story of 2026 is the rest of the audience catching up, including older UK consumers who had previously been the slowest to adopt.
Boomers haven't softened on every front. 80% of UK Boomers still prefer a human representative when both options are equally available. 87% say human connection is important during a high-stakes purchase. The picture that emerges is of an older audience that is now using generative AI as a research tool, while still preferring human-led conversations when they're ready to make the decision. The implication for brands targeting older consumers: AI matters at the discovery stage, humans matter at the close.
The human-preference gradient extends across every generation. Younger UK consumers are roughly half human, half open to AI. Gen X tips firmly human at 69%. By the time you reach Boomers, nearly four in five want a human representative when both options are equally available, and the share willing to take AI alone drops to 1%. The buying journey may be increasingly AI-mediated upstream, but the moment of decision is still a moment of human contact for the audiences brands depend on most.
For all the AI advances, UK consumers haven't changed their mind about what they want at the moment of a high-stakes purchase decision.
83% of UK consumers say human connection is important or very important during a high-stakes purchase. 55% say it's very important. 58% prefer a human representative when both AI and human options are equally available.
These numbers are essentially flat versus 2025, and the floor hasn't moved. Across every generation, every industry, and every channel preference cut, UK consumers continue to anchor the moments that matter on human contact. The 2026 data simply adds a layer of nuance: AI is welcome in the journey, especially upstream, but the human stays in the chair when the customer is ready to commit.
Two data points that look contradictory at first, and tell a connected story on a second look.
71% of UK consumers have hung up after being placed on hold, up from 43% in 2025. 45% say they're very likely or likely to stop doing business with a brand after one bad customer-service experience, down from 75% in 2025.
The hang-up figure overstates the year-over-year change because the question wording moved from "hung up on hold" to "hung up because you were placed on hold for too long." But even allowing for that methodology shift, hold tolerance is down sharply. UK consumers will not wait. They expect to be heard, and if they're not, they'll abandon the call rather than abandon the brand.
The stop-doing-business number is the more nuanced finding. UK consumers became substantially more forgiving year-over-year, a 30-point drop in the share saying they'd stop after one bad experience. They're willing to put up with one bad experience. They're not willing to put up with the journey to that experience. The brand that picks up the phone in two minutes earns the right to make a mistake. The brand that leaves the customer on hold for ten doesn't even get the chance.
Invoca's voice AI products solve directly for this. When consumers are about to hang up, an AI that can engage them in the first 30 seconds, take the intent, and either resolve the issue or route to a human is the difference between a saved sale and a lost one.
This year, for the first time, we asked UK consumers whether they had spoken to an AI voice agent on the phone during a recent purchase journey. The result confirms what brand-side data has been hinting at for months.
28% of UK consumers say they have spoken to an AI voice agent in the last 12 months. 18% think they did but weren't completely sure. 53% say they did not.
That's 46% of UK consumers who have either definitely or possibly spoken to an AI voice agent in the last year, meaningful adoption for a technology that barely existed twelve months ago. Among UK consumers who did interact with an AI voice agent, the modal answer when asked how the interaction compared to a human is "about the same," which is exactly the right place for a new technology to be sitting after twelve months of deployment.
The opportunity, and the risk, is the tail. The consumers who said the AI voice interaction was much worse than a human are the consumers most likely to blame the brand and to leave. The brands that win the AI voice agent moment will be the ones that deploy them carefully, disclose them clearly, and route to a human the second the conversation needs one.
Exposure to AI voice agents in the UK is uneven across industries, and the order is almost the inverse of what we see in the US. UK Healthcare leads at 59% of consumers reporting they spoke to one (definitely or possibly), with Home Services close behind at 58% and Financial Services at 57%. UK Insurance lags at 38%, the only industry below 50%. That's striking when set alongside the Q23 sentiment data, where UK Insurance also has the worst AI-experience scores. The two findings are consistent: UK Insurance has deployed AI less aggressively than its peers, and where it has deployed AI, consumers report a worse experience.
The 2026 data describes an inflection point. The brands that win the next chapter of the buying journey are the ones that treat AI as a brand-equity decision, deploy it well enough that consumers don't notice in good moments and trust the disclosure in bad ones, and connect every AI touchpoint to a human moment when the buyer is ready to commit.
The phone is still where the decisions get made. The data is still where the optimisation happens. The brand that connects the two, and that does it with AI that respects the consumer, wins.
Learn more at invoca.com →For this report, Invoca surveyed 1,356 consumers in the US and UK who researched and made a high-stakes purchase in the last 12 months across seven industries: automotive, healthcare, home services, insurance, financial services, telecommunications, and travel. Only UK data is used in this version of the report, representing 663 respondents. A high-stakes purchase is one where consumers take time to weigh options, research, and put more thought into the decision because of cost or complexity, generally above £500, or above £1,000 for travel. Results may not total 100% due to rounding and multi-select question formats. The field survey was performed via the Trycycle Gather conversational survey platform between 8 and 22 May 2026.
Results may not total 100% due to rounding and multi-select question formats. Multi-select questions are clearly flagged on each chart. Year-over-year comparisons reflect minor differences in question wording where noted in the body of the report. The "industry" cut is based on the high-stakes purchase the respondent made in the last 12 months and is multi-select, so a single respondent could appear in more than one industry if they purchased across categories. Generational definitions follow the Pew Research Center cutoffs. Powered by Gather (gatherhq.com).