Protecting the Customers You've Already Earned: Why Past Clients Ask AI Too

By David Quenneville, MBA, Founder, Oscker — Published 2026-08-24T00:00:00+00:00 · Updated 2026-08-24T04:46:29.156883+00:00

45% of consumers now ask AI for local recommendations. Here's what that means for the customers who already hired you once.

Why Are Past Customers Turning to AI to Find Businesses They Already Use?

Because AI became a mainstream way to get a local recommendation almost overnight: 45% of US consumers used a tool like ChatGPT, Gemini, or Perplexity for a local business recommendation in the past year, up from just 6% the year before, and 63% of those users trust what the assistant tells them (BrightLocal, Local Consumer Review Survey 2026). A customer who already hired you once is now just as likely to re-ask an assistant as to scroll back through old texts to find your number.

I've had this happen. A customer I'd already worked for called me months later and led with an apology — she couldn't remember my company name, so she asked an assistant who did the kind of work she needed, and I wasn't in the answer. She found me by scrolling back through old messages. She still called. But she almost didn't, and I'd already earned her once. That's the part worth sitting with: the customer who liked your work isn't automatically the customer who can name you a year later. The encouraging part is that this is the cheapest kind of problem an owner has — you aren't buying a new customer, you're just making sure the places AI looks already say the right thing about you.

The Channel Shift Nobody Budgeted For

This isn't a slow drift. ChatGPT alone was used by 31% of US consumers for local recommendations, and Google AI Mode by 23%, in a single 12-month window. AI is now the third most-used tool for local business discovery, trailing only Google and Facebook. Trust is moving just as fast: 42% of consumers say they trust AI platforms as much as online reviews, and 23% are willing to rely on an AI-generated review summary alone when deciding who to call. Most still check, though — 88% of AI users verify sources before acting, which matters for what comes next.

The Risk Is Asymmetric, and That's the Part Most Owners Miss

Ranking in Google's local results and getting named by an AI assistant are not the same contest, and the gap between them is enormous. Across nearly 350,000 business locations and 2,751 multi-location brands, ChatGPT recommended just 1.2% of locations, Perplexity 7.4%, and Gemini 11% — against 35.9% average appearance in Google's local three-pack (Search Engine Land, reporting SOCi's 2026 Local Visibility Index). AI visibility, by that measure, runs three to thirty times harder to achieve than traditional local ranking.

Meanwhile the click that used to bail out a forgetful customer is disappearing. Google searches in the US ended without any click 68.01% of the time in the first four months of 2026, up from 60.45% in 2024 — a search that used to land on your website or Google Business Profile increasingly just ends on the results page itself (Search Engine Land, reporting SparkToro/Rand Fishkin's analysis of Similarweb clickstream data). AI Overviews now appear on more than 20% of Google searches, and click-through rates fall by nearly 60% when they do. Worth stating plainly, because the same research says it directly: branded searches and local queries still tend to benefit from ordinary SEO — a customer who remembers your actual name mostly still finds you. The exposure concentrates in the moment they don't remember, and reach for an assistant instead.

Why a Middling Reputation Now Costs You the Whole Answer, Not Just a Rank

In traditional local search, an average rating could still surface you on proximity alone. AI answers don't work that way. Recommended locations in the SOCi study carried average ratings of 4.3 stars on ChatGPT, 4.1 on Perplexity, and 3.9 on Gemini — reviews function as a filter, not a ranking signal. Underperforming brands with ratings near 3.4 stars and review response rates below 5% were, in the report's own words, effectively invisible in AI recommendations.

There's a second, quieter failure mode: getting named but getting named wrong. Business profile information was only about 68% accurate on both ChatGPT and Perplexity, compared with 100% accuracy on Gemini, which draws its data directly from Google Maps (SOCi). A past customer who does get your name back from an assistant might still be handed the wrong hours, the wrong service area, or an outdated phone number.

The Complaint You Never Hear Is the Customer You Already Lost

Operators used to get some warning before a customer walked. That warning is fading. One in two bad experiences now leads a customer to cut spending, and 30% of customers who have a bad experience say nothing at all — they simply switch, silently, an all-time high since Qualtrics began tracking the figure in 2021 (Qualtrics XM Institute, 2026 Global Consumer Experience Trends, 20,000 consumers across 14 countries). Only three in ten unhappy customers now explain why they're leaving. "Response rates have been falling for years, and they're not coming back," says Isabelle Zdatny, Qualtrics XM Institute's head of thought leadership.

Put the two findings together and the operator loses both signals at once: the click that used to confirm a customer found you, and the complaint that used to explain why one didn't come back.

What the Evidence Actually Shows About AI Moving a Past Customer

Here's where it's worth being precise instead of alarmist, because the research itself is precise. A recent preprint from Scrunch AI researchers Michael Iannelli and Alan Ai — not yet peer reviewed, and vendor-authored, both worth disclosing — found that when an AI assistant recommends a brand to someone with no recent engagement, that person's brand-name Google search rises 4.3 percentage points, their visits to the brand's own site rise 2.4 points, and retailer-page visits rise 1.0 point, measured against matched behavioral placebos. That's a real, measurable acquisition effect.

For someone already engaged with a brand, the picture is different and smaller: the same study found existing customers' activity after an AI mention mostly sits inside an episode already underway, a three-day increment of just 0.4 percentage points in discovery activity and roughly zero in retail visits — and the authors explicitly decline to rest a claim on that subset because the sample is small and the interval wide. The honest reading: no study located for this piece has directly measured a past customer switching providers because their prior business was left out of an AI answer. What the evidence supports is exposure and mechanism — AI is a powerful channel for moving people who are not currently engaged with a brand, which is exactly the state a past customer sits in three years after their last job. It does not yet support a specific measured loss rate, and claiming otherwise would be overreach.

Two Free Tools Worth Using Before You Spend Anything

Two resources are worth running before spending a dollar on this. Ahrefs' free AI Visibility Checker (no signup required) lets you enter your business name the way a real customer would say it — including a specific past job, like "the plumber in [town] who replaced my water heater" — and see whether ChatGPT, Gemini, Perplexity, Copilot, or Google's AI Overviews return your name or a competitor's. And TextMagic's free Churn Rate Calculator turns three simple inputs — paying customers, customers lost last month, and average monthly value per account — into a dollar figure for what a lapsed customer base is actually worth, useful for putting real numbers behind a decision before investing in AI-visibility work.

Where This Leaves an Owner-Operator

The two halves of this problem are each well measured on their own: past customers are turning to AI in large and fast-growing numbers, and AI names only a small, ratings-filtered slice of local businesses when they do. What connects them is still a mechanism, not a headline loss figure — which makes the response straightforward rather than urgent-sounding: keep the record AI reads about you accurate, keep the reputation signals strong enough to clear the filter, and don't assume a customer who hired you once will find their way back without help.

Free GEO Sample Report

Read on oscker.com