Phase one of Google’s Local Services Ads migration starts this month. Select home and storefront service advertisers in the United States get their LSA campaigns automatically upgraded to Performance Max pay-per-lead campaigns inside Google Ads. No opt-in form. No migration window you pick. Google flips it. If you run a plumbing shop, a roofing crew, a locksmith, or a cleaning company that has been buying leads through the standalone LSA dashboard for years, the box you log into is about to change, and so is the bidding logic underneath it.
That was the biggest operator-facing story of the week, but it was not the only one. Here is what else landed between August 1 and August 7, and what to actually do about it.
Google Is Moving Your LSA Money For You
The July 6 policy rewrite renamed the Local Services platform policies to “Local Services Ads requirements,” which sounded like housekeeping. It was not. It aligned the rulebook with the new badge system and cleared the way for the PMax migration now hitting in August. Two things change at once for you: where the campaign lives, and what signals decide your cost per lead.
Standalone LSA gave you a narrow set of controls. Job types, service area, weekly budget, and the Google Guaranteed badge did most of the work. PMax pay-per-lead pulls in creative assets, audience signals, and conversion feedback from your account. That means the quality of what you feed it starts to matter in a way it did not before. Shops with clean conversion tracking and real asset libraries will get better lead costs. Shops that never wired up call tracking or uploaded a decent photo set will drift.
Three moves before your account flips. First, confirm your conversion actions in Google Ads are correct and that phone calls from ads are counted as primary conversions, not secondary. Second, upload at least ten real job photos and five short text assets covering your top three job types. Third, export your last twelve months of LSA lead-level reports now. Google already cut historical LSA reporting once during this transition, and you do not want your only record of cost per booked job living inside a dashboard that is being retired.
One more wrinkle worth flagging. Starting August 2, Local Services Ads no longer allows AI-edited or AI-generated images in the European Union and in New York state. If you operate in New York and you have been running upscaled or AI-cleaned job photos, swap them for unedited camera originals this week. This is a narrow rule today. It is the kind of rule that spreads.
AI Will Cite Almost Anyone. It Recommends Almost No One.
The most useful research of the week came out August 5 from Semrush and Growth Memo. They tracked 1,094 US categories monthly inside ChatGPT from January through June 2026, with five prompt variants per category. The finding cuts against how most local businesses are chasing AI visibility.
Being cited and being recommended are different outcomes with different requirements. In categories close to a brand’s core expertise, brands were cited 74% of the time, named 44% of the time, and got both 34% of the time. In distant categories, appearances were 50% citations and only 25% mentions. So an AI assistant will happily pull a fact off your page while recommending a competitor by name in the same answer. Citation is cheap. Recommendation is earned through repeat presence across many prompt variants in one category.
That reframes the work. Publishing thirty pages across nine loosely related service topics buys you citations you cannot bank. Publishing twelve pages that all orbit one job type in one metro, and getting mentioned in reviews and directory listings for that same job type, is what moves you from cited to recommended. Pick one category. Own it before you spread.
The numbers underneath explain the urgency. A SparkToro analysis of Similarweb data put zero-click Google searches at 68% in early 2026, up from roughly 60% in 2024. When an AI Overview shows, 83% of searches produce no click anywhere. In AI Mode, 93%. Google said at I/O that AI Mode passed one billion monthly users with query volume more than doubling quarter over quarter. Ranking on a page nobody clicks is not a win.
Your Google Business Profile Is Now the Feed, Not the Listing
HousingWire ran a piece on August 5 making the case that listings data has become the marketing asset, and the supporting detail is worth repeating for every local operator, not just agents. Google’s “Grounding with Google Maps” reached general availability this year, connecting Gemini to more than 250 million verified places. Gemini treats the Business Profile as authoritative on local queries. AI Overviews use Profile data as the structural base of local recommendations. ChatGPT reaches the same information through Bing Places, Foursquare, and verified directories.
The practical consequence is that inconsistency is now a ranking penalty in the answer layer. When these systems cross-reference your name, address, phone, and hours across Google, Bing, Yelp, Foursquare, and your own site and find mismatches, confidence drops and citation frequency falls with it. Nobody sends you an email about this. Your recommendation rate just quietly declines.
The audit takes an hour. Pull up your GBP, your Bing Places listing, your Yelp page, your Apple Business Connect entry, and your website footer side by side. Check that the business name string matches character for character, including any suffix. Check the phone number formatting and, more importantly, that it is the same number. Check holiday hours, which are the most common stale field in August. Then front-load your GBP description with your entity name and primary city in the first sentence, because these systems weight opening text hardest when deciding what a business is.
We covered this exact mechanic on Wednesday in Google Cites Itself 8.4x More. Your GBP Card Is Your Homepage, and the week’s data only strengthened the case.
Real Estate Ran Out of Holdouts. Zillow Cut 500 Jobs.
Delta Media released an analysis on August 5 finding that only 2% of real estate brokerages say they will not adopt AI in 2026. The adoption gap between mid-size and the largest firms closed entirely, both at 100%. The interesting shift is in buying behavior. Brokerage leaders rating all-in-one AI and automation platforms 8 or higher out of 10 rose to 51.5% from 41.6% in 2025. Operators are consolidating rather than stacking point tools. If you are an agent choosing between six subscriptions and one platform your brokerage already pays for, the market is voting for the platform.
Meanwhile Zillow cut just over 500 roles on August 4, roughly 7% of its global workforce and its second round of cuts this year after about 200 in January. CEO Jeremy Wacksman cited a flat housing market and cost discipline. Zillow told GeekWire the cuts were not driven by AI, though the company describes itself as AI-native and has been retraining staff to build AI into their work since spring. Read it however you like. The operator takeaway is that portal economics are tightening while portal AI features keep shipping, and lead prices rarely fall when a seller is under margin pressure.
Two smaller items worth your attention. New research reported August 5 found ChatGPT strongly favors English-language pages compared to Google AI and Copilot, which matters if you serve a bilingual market and have been building Spanish-language service pages, a topic we covered for trades on July 23. And Gemini is now replacing Google Assistant on additional devices, which quietly expands the surface where your Business Profile data does the talking for you.
On the small business side, Q2 data showed 93% of owners expecting moderate to significant growth over the next year, with inflation reclaiming the top concern spot at 34% ahead of cash flow at 30%. Among small businesses using AI, marketing leads adoption at 63%, and 87% of AI users report positive impact. Optimism is holding. Costs are the pressure point.
Here is the question to sit with this weekend. If an assistant were asked to recommend a business like yours in your city right now, would it name you, or would it quote your website while recommending someone else? Those are two very different outcomes, and this week’s research says most operators are chasing the wrong one. Pick your one category. Get named in it.
Atlas Unchained builds the AI systems that make local operators the answer, not the footnote. If you want the week’s moves without reading eight newsletters, subscribe and we will keep sending them.
About the Author
Trevor Kaak is the founder of Atlas Unchained, a portfolio of products and services helping local businesses run leaner with AI — from custom websites to vendor-bidding marketplaces to vertical SaaS. He writes about marketing, automation, and the craft of building software for operators who’d rather work on their business than in it.