I Cut $100,043 of Google Ads to $1,827 and Doubled Leads

The company had everything a local service business is supposed to have. A WordPress site. A page for each of its 59 cities.

A library of about 158 articles on mold symptoms, tenant rights and inspection costs. And a Google Ads account that had quietly become the business.

At the peak, a single month cost $20,155 in ads. Over ten months the total came to $100,043. When the ads ran, the phone rang.

When the ads stopped, it didn’t. That is not a marketing channel. That is a landlord.

The website, meanwhile, earned close to nothing. Not one of the 59 markets ranked in the local top three. Fourteen were on page one at all.

When I asked four AI assistants about mold testing in the company’s cities, three of them had never heard of it. The site scored 64 out of 100 on my Audit Engine. The content library, years of work, sat there drawing readers who never became customers.

The owner’s question was simple: can we stop paying Google $20,000 a month without the business falling over?

Nobody could answer it, because nobody could see what organic was doing. Bookings weren’t tied to a channel.

The GA4 conversion events the site relied on turned out to be a tracking artifact, counting things that weren’t customers. Turning off a $20,000 faucet meant knowing, in dollars, what the other faucet produced, and that number didn’t exist yet.

So the first job was not content. It was measurement.

I ran the program in a deliberate sequence.

Weeks one and two: make the money visible. Tracked phone links on all 59 city pages, so a call from a Sacramento page was recorded as a Sacramento organic call. A quote-request event we could trust, replacing the broken one. Structured business data on every market page, which is the plain-language information that tells Google and AI assistants who a business is, what it does and where it operates.

Title rewrites. A small file that tells AI crawlers what the site is about.

None of this earned a dollar. All of it made the dollars countable.

Weeks two through twelve: build the pages that book. Two markets a week, in order of demand, each rebuilt against a teardown of the competitors already winning that city. A quote-request button on every market page. Internal links pointed at the pages that produce bookings rather than the pages that produce readers.

All 158 existing articles refreshed. New articles on Monday, Wednesday and Friday, each created with my Article Factory: drafted against the client’s context layer, checked by a 30-point scorecard, verified by code, and edited by a human before it went anywhere near the site.

Weeks four through twelve: become the answer. Where Google was already showing an AI Overview for a question, we rewrote the page so the direct answer sat at the top in a self-contained block an assistant could lift. We added FAQ sections, a public entity record for the business, video, and standardized structured data across 276 indexed pages.

Every fix went to the client’s own developer as a specified ticket with the reasoning attached, bundled in a single drop-in package. He implemented; he didn’t have to investigate. That one habit saved weeks.

The three-a-week publishing cadence ran on my Article Factory, already serving other clients, so it cost the company roughly $12 a month to keep producing rather than the $800 to $2,000 per article an agency would charge. My Optimization Loop ran underneath it: a live dashboard showing leads, cash and rankings by channel, a health check every two days, and a weekly audit that caught what moved and what didn’t. And the client’s developer had a growing folder of tickets he could install on his own schedule.

Three independent systems told the same story.

• Paid search spend fell 96%: from a $20,155 monthly peak to $1,827 in total across March through July.
• Total monthly leads, all channels, went from 212 to 427 over the same window.
• Organic search collected $32,538 in cash in 90 days, from the client’s own CRM.
• Organic leads per month climbed 0, 20, 41, 86, 134 from March to July. Collected cash from organic went $0, $1.9K, $6.3K, $9K, $17.2K a month.
• 279 organic leads, 75 of them booked inspections. Spam: one lead out of 279.
• Search Console clicks went 614 to 2,029. GA4 organic sessions went 188 to 2,391. Average position improved from 16.9 to 11.8. The Audit Engine score went 64 to 78 to 83, same method each time.

The month the ads went off, lead volume went up. There was no holdout group, and mold demand rises with warm weather, so what the data shows is a sequence rather than a controlled experiment. The sequence is this: a business that was spending $20,000 a month to make the phone ring now spends almost nothing, and the phone rings more.

• The niche: Residential mold testing, 59 US city markets, WordPress site.
• The problem: No owned demand. $100,043 in Google Ads over ten months, 0 of 59 markets in the local top three, broken conversion tracking, three of four AI assistants unaware of the brand.
• What was built: An attribution layer (tracked phone links, a trusted quote event), 59 rebuilt market pages with structured data and FAQs, a refreshed 158-article library plus new articles at three a week, answer-shaped content for AI Overviews, a public entity record.
• Implementation: Twelve weeks in three phases: measure, build the pages that book, become the answer. Every fix shipped to the client’s developer as a specified ticket in one drop-in package.
• Results: Paid search from a $20,155 monthly peak to $1,827 for the quarter. Leads 212 to 427 a month. $32,538 collected from organic in 90 days. 75 booked inspections from 279 organic leads. Search clicks 614 to 2,029.
• Built with: my Audit Engine, Article Factory and Optimization Loop.