Your Google Ads bill went up 15 percent this year, and the clicks got worse. That is the headline number from Forbes’ July 12 small business tech roundup: average cost per click rose 15 percent between June 2025 and June 2026 while advertisers reported a sharp decline in efficiency. Same budget, fewer booked jobs. If your lead cost crept up this spring and you assumed it was your landing page, it probably was not your landing page.
Google raised the rent while shrinking the apartment
Here is the AU take on the CPC number. Paid and organic are getting squeezed by the same hand at the same time. AI Overviews now answer more local queries directly on the results page, which trims the organic clicks you used to collect for free. Then the auction charges you 15 percent more to buy back the visibility the answer box removed. Renting attention from a single landlord was always a concentration risk. Now the rent is rising while the apartment shrinks, and the lease renews daily.
The move is not pausing ads on Monday. The move is measuring like a skeptic. Pull cost per booked job, not cost per click, for the last 90 days and set it next to the same window last year. If the trend bends the wrong way, your next marginal dollar belongs in channels you own: your list, your review pipeline, your site, your booking flow. Wednesday’s post on owning your own numbers laid out the exact tracking setup. This week’s pricing news is the reason to actually build it.
Two out of three buyers now start in an AI answer
The July dispatch from AI and Realtors puts hard numbers on the search shift real estate has felt all year. 67 percent of buyers now start their property search on generative AI, up from 17 percent eighteen months ago. Meanwhile 91 percent of agents never appear in those AI answers at all, and 1 percent of professionals collect 47 percent of the citations. Visibility did not vanish. It concentrated, violently.
Pair that with Tuesday’s look at Zillow Pro and its My Agent lock-in and the picture sharpens. The buyer conversation is moving into chat windows, and the platforms are racing to be the database those windows quote. If you sell anything locally, not just houses, the realtor numbers are your preview. The professionals getting cited publish specific, answerable content backed by clean structured data. The other 91 percent are waiting for a referral that now goes somewhere else.
The fix costs a weekend, not a budget. Publish one page per neighborhood or service area that answers the questions buyers actually type. Mark it up with schema. Make sure your name, license, and specialty read identically everywhere, from your site to your portal profiles. AI engines cite sources that are specific, consistent, and machine-readable, and they skip generic “why choose us” pages every time. June’s citation-share playbook has the full checklist.
The smart money says installation, not models
TechCrunch reported Wednesday on Ode, the $1.5 billion AI implementation company Anthropic launched with Blackstone, Hellman & Friedman, and Goldman Sachs. The thesis: the next trillion dollars in AI gets earned by wiring models into actual businesses, not by shipping smarter models. When the lab that builds the model puts a billion and a half behind the installation trade, believe the signal. The bottleneck moved from the technology to the adoption.
It is the enterprise version of the gap AU flagged on Monday: 66 percent of small businesses use AI, and most teams are winging it with no training and no process. The constraint was never access to the model. It is the unglamorous work of picking one workflow, wiring the tool in, and teaching the team to trust it. Monday’s one-week training fix is the small-shop version of what Blackstone just funded at $1.5 billion.
Microsoft made the same point in reverse. After sustained customer backlash to Teams Facilitator, an AI that monitors meetings, Microsoft shipped controls that let organizers switch AI features off during a live meeting. The biggest software company on earth just conceded that AI works when people opt in and revolts when it is imposed. Remember that before you bolt an AI onto anything your customers touch.
The trades got an AI-first contender
Roooster launched this month: an AI-first field service platform covering quoting, scheduling, dispatch, route optimization, invoicing, and payments across more than 20 home service verticals, from HVAC and plumbing to pressure washing and garage doors. One more credible rival means pricing pressure on ServiceTitan, Jobber, and Housecall Pro, which pays you whether you switch or not. Get a Roooster quote and bring it to your next renewal call.
One caution before anyone rips out a working system in August. Roooster is new, and new platforms earn trust with uptime, support response, and payment-grade reliability, none of which shows up in a demo. The play for most shops is not switching this quarter. It is letting the newcomer sharpen your incumbent’s pencil. Ask your current vendor which AI features are now in-plan, what they cost, and what the renewal looks like if you walk away. Price benchmarks work on software, too.
ServiceTitan’s own 2026 State of AI in the Trades report, surveying more than 1,000 contractors, says 48 percent of home service pros actively use AI today. Of those users, 62 percent report measurable efficiency gains, many saving three or more hours a week. The top use cases: customer communication at 52 percent, then estimates and pricing at 51 percent. That second number is exactly the ground Thursday’s stale price book post covered. Half your competitors are repricing with AI help. Are you?
5.6 million new competitors and one weekend question
PYMNTS reported this week that U.S. business applications hit 5.6 million in 2025, up 24 percent since ChatGPT launched, with the strongest growth in the sectors most exposed to AI. The mechanism matters more than the count. AI lowers the minimum efficient scale of a business, the revenue level where hiring a specialist used to become necessary. A solo founder with good tools now ships what took a staff of five in 2019.
The poster child is extreme: Medvi, a telehealth startup, launched in late 2024 with $20,000 and zero employees and posted $401 million in first-year sales. Your market will not produce a Medvi. It will produce three lean competitors who answer every call with a voice agent, reprice weekly, and publish content the AI engines can quote. Your moat is not that they cannot copy your work. It is the relationships, reviews, and decade of local proof they cannot fake, provided you put those assets somewhere the machines can read them.
So here is the question to sit with over the weekend. If Google doubled your click prices tomorrow morning, which channel you actually own could carry the load: your list, your reviews, your referrals, your site? Whatever came to mind first, that is your Q3 investment. If nothing came to mind, that is the finding. We will keep tracking the CPC creep and the AI citation race and report back. If this Brief saved you a Friday of scrolling, subscribe below and get next week’s in your inbox.
