Your Customers Aren't Leaving. New Ones Just Can't Find You.

What does the data actually say?

Stripe's benchmarking data covers about 181,000 subscription businesses, and an analysis of it published this month found a strange combination. In October, revenue at the median company shrank slightly, and the fastest-growing companies slowed sharply. But customers were not cancelling: churn fell, from 7.45% to 7.24%, and customer lifetime value was up over the year in 41 of 48 industry groups. The only size band still growing was the largest, companies over $5 million in annual revenue. In plain terms, the existing customers stayed and kept paying. The new signups are what went missing. One caution belongs with those numbers: this is one month of data, it covers subscription businesses only, and a single month can bounce back. The pattern is worth taking seriously. The month by itself is not proof.

Mack Grenfell: The SaaS apocalypse started in October (analysis of Stripe benchmarking data)

Why are new customers harder to reach?

Two forces squeeze the front door at the same time. The first is crowding. Software, websites, and marketing tools have all become cheap to produce, so every niche has more competitors in it than it did two years ago, all chasing roughly the same pool of new buyers. The second force is where the search itself happens. A customer who once typed a few words into Google and compared ten blue links increasingly asks an assistant a full question instead, and reads one synthesized answer. That answer does not list ten businesses. It recommends a few, sometimes one or two, and the rest are not ranked lower. They are absent. A business can be excellent, well reviewed, and fairly priced, and still never be mentioned, because being good and being named have become two separate problems.

Mack Grenfell: The SaaS apocalypse started in October (analysis of Stripe benchmarking data)

How many customers now ask AI first?

For local businesses, the best measurement is BrightLocal's Local Consumer Review Survey 2026, which asked 1,002 U.S. adults how they find businesses like yours. Forty-five percent said they had used an AI tool for a local business recommendation in the past year, and ChatGPT was the most used, named by 31%. AI is now the third most-used source of local recommendations, behind Google Search and Google Maps. One more finding matters for what comes later in this article: 97% of the people who used AI for a recommendation checked it against real reviews at least some of the time. The assistant makes the shortlist. The reviews confirm it. A business needs to survive both steps.

BrightLocal: Local Consumer Review Survey 2026

BrightLocal: AI trust report (March 2026)

Why doesn't a good Google ranking settle it?

Because the assistant is far more selective than Google, and it does not simply copy Google's results. SOCi's 2026 Local Visibility Index studied more than 350,000 locations and found that ChatGPT recommended just 1.2% of them. The same businesses appeared in Google's local 3-pack 35.9% of the time. Fewer than half of the brands that lead in traditional local search also appear among the top AI recommendations. Two honest limits: SOCi studied large chains, not single-location businesses, so treat the exact percentages as a picture of the cliff rather than a forecast for your street. And the two channels measure slightly different things. The direction is not in doubt, though. Getting found the old way and getting recommended the new way are different achievements, and the second one is much rarer.

SOCi: 2026 Local Visibility Index (figures as published by SOCi)

What is a customer who arrives through AI worth?

More than one who arrives the old way, at least in the one industry where the money has been measured carefully. Adobe tracks over a trillion visits to U.S. retail sites, and in September 2026 visitors who arrived from AI sources converted 90% better than other traffic and produced 83% more revenue per visit. They had done their research inside the assistant before they clicked, so the visit itself was a decision, not a browse. This is retail data, and it does not prove that any single change to a website causes those numbers. But it tells you what is at stake in being named: the customers who come through an AI answer are the most decided customers on the internet right now.

Adobe: AI Traffic Drives 83% Higher Revenue Per Visit (Adobe Digital Insights, October 2026)

What can a local business actually do about this?

Start from how the assistant builds its answer. It gathers what your website says, what your Google profile says, what directories and review sites say, and what other people have written about you, and it recommends the businesses those sources agree about. So the work is not a trick and it is not an ad. It is making one clear, consistent account of your business exist in all the places the assistant looks, with your own website carrying the fullest version of it: what you do, who it is for, where you serve, what it costs people to choose wrong, answered in plain language on pages that answer real questions. Then keep the confirmation layer healthy, because nearly every customer checks the recommendation against reviews before they call. And measure the thing itself: ask the assistants the questions your customers ask, repeatedly, and count how often your business is the answer. One screenshot is weather. A rate over time is climate, and it is the only version of this number worth making decisions on.

BrightLocal: Local Consumer Review Survey 2026

SOCi: 2026 Local Visibility Index (figures as published by SOCi)

The short version

Your customers are not leaving. Churn fell in the biggest dataset going. What changed is the front door: new customers increasingly meet your business first as an AI answer, that answer names almost nobody, and a good Google ranking no longer guarantees a place in it. The businesses that will keep growing are the ones a machine can understand, verify, and recommend with confidence. That is a buildable thing. It is also, increasingly, the whole game.