The map re-sorted under every firm we manage. We measured what actually happened.

Nine plaintiff firms. Twenty-nine offices, from single-location practices to multi-state brands. All of them felt the phone getting quieter this year — and the reporting they had showed nothing wrong. So we built instruments and put the whole book under the microscope.

The situation — everywhere at once

Through the first five months of 2026, calls from Google Business Profiles across our entire book ran 26% below the prior year — with individual firms down anywhere from 20% to 51%. Big brands and single-office practices. Six states. Every intake team asking the same question: where did the calls go?

The easy answer would have been a service menu. We built instruments instead: a 49-point ranking grid over every office, scanned weekly with verified-location matching; competitor tracking on every firm holding pack positions; 27,000 search terms classified branded versus discovery; and eighteen months of performance data reconciled day-by-day against a second independent provider — 160,974 datapoints, zero disagreements.

Finding one: only the action Google can lose track of declined

Here is the anatomy across all twenty-nine offices, same months, same profiles:

• Calls: down 26%
• Website clicks: up 5%
• Direction requests: up 6%

A demand collapse drags all three down together. This dragged down exactly one — the one with a counting problem. AI answer surfaces now display firm listings, and often the phone number as plain text, directly inside answers. The caller dials natively; Google's tap-counter never fires. We even found hundreds of searches where people typed a firm's phone number itself into Google — behavior that only exists when numbers are being served as answers.

So the scoreboard changed: every client's intake system call log, reconciled monthly against Google's curve. A dialed call Google never saw is still a signed case, and it deserves to be counted.

Finding two: views are inflating from surfaces that never convert

While calls fell, maps impressions across the book rose 82% in June alone — and on the largest brand we manage, views ran 48% above prior year on identical calendar windows while interactions stayed flat, with maps impressions tripling from an abrupt onset in March. Engagement per hundred views fell by a fifth book-wide. The incremental views are AI-era surfaces — answer panels, embedded map modules, assistant results — that display a listing without ever handing over the click. Visibility is not falling. What a "view" is worth is falling.

Finding three: there are two kinds of law firm profiles now

When we classified every search term across the book, the firms split cleanly. The established brands are brand-recall machines — as much as 98% of the searches that surface their profiles contain their own name. Their call volume tracks their advertising, not their map position, and they are maximally exposed to AI answer-absorption. At the other end, discovery-native firms — including one at 97% unbranded discovery — live entirely on searchers who typed a problem, not a name.

That spectrum is the strategy. For the brands: protect the counting, then build the unbranded capture they have never had. For everyone: the searcher with no name in mind is won or lost in the Map Pack, square by square — and that machine can be measured.

Finding four: we dissected the winners. It is not what the industry sells.

Fifteen forensic dossiers on the firms dominating our clients' maps — Google records, page-level backlinks, review anatomy, content audits — and most popular theories died on contact:

• Page-level link authority? Dead. One winner sweeps all 49 grid squares of its market with exactly one referring domain on its landing page.
• Content volume? A tiebreaker. In contested markets the winner's page runs 3,300 words naming its city 53 times against 1,800-word templates naming it 11 — but other markets are swept with thinner pages than the losers'.
• What every winner shares: a rating of 4.9 or higher — review counts ranged from 38 to 917; the rating did not — a tightly focused local entity, and signals proportionate to their market. June's core update enforced that last one in front of our instruments: a satellite listing carrying 350 five-star reviews in a small town — wildly out of scale — was cut from 41 map squares to 3 overnight, while the same firm's flagship, with fewer reviews, kept its metro.

Seventeen of the twenty-nine offices we manage sit below 4.9 today. That number, not a posting calendar, is the first project in every metro.

Finding five: the map re-sorts monthly. Position for it.

In one June week our grids recorded two competitors collapsing in a single market and the best-positioned firm standing absorbing more than thirty map squares in seven days — more movement than months of optimization produce. Volatility is the new weather. Across the book, hundreds of positions sit at ranks 4–10 with a named opponent in every square. The next update is the cheapest ranking gain available — for whoever is built for it.

The part most agencies would not print

Partway through this work we found a flaw in our own ranking scanner — multi-office brands sharing one name broke its matching logic, and some early numbers we reported ran high. We rebuilt the matcher around Google's verified location IDs, re-scored all 576 historical scans, flagged 351 positions we could no longer defend, and told the affected client before they could find it.