Inside a $5K MRR Plumbing Business: The Full Diagnosis

Reading time: 8 minutes

Last verified: 11 August 2026

A working plumber doing $5,000 a month in recurring revenue. Solid trade, solid reputation, leaky marketing. Here is the diagnosis, page by page.

Most owners who book a Diagnosis are not in trouble. They are in the awkward middle. The business works, the customers are happy, the growth is flat and nobody can name the leak. Everybody has a theory, the theories cancel out, and the spending carries on.

Here is one of those accounts in full. The client asked to be anonymised, so no name and no city. Everything else is exactly what we found.

→ He blamed the newest and most visible channel, because that is the one with a number attached to it. The two real leaks were sitting underneath it, and neither of them was the ads.

This article covers the trust-signal score, the two leaks on the page his ads pointed at, the review and funnel problems, and the order we fixed them in. By the end you can run the first two checks on your own business.

What was the owner's starting position?

Quick Answer: A US plumbing business, eight years in, three vans, two technicians plus the owner, at around $5,000 a month in recurring revenue.

He had been running Facebook ads for three months, and the leads were not converting. The agency running the ads said his funnel was the problem. He said the leads were poor quality. Neither could prove the other wrong, and his monthly ad budget went out either way. We do not start with the ads. We start with the website and the trust-signal layer underneath it, because no ad platform can fix a page that fails the visitor on arrival.

What did the trust-signal score look like?

Quick Answer: 5 out of 19. The worst gaps were no named author anywhere on the site, broken service schema on three of seven service pages, and a review feed that had not moved in twelve weeks.

The scorecard told us most of what we needed inside fifteen minutes. The 19 signals are the Ambitions scoring framework, built from the editorial patterns we keep seeing in the pages that AI answers do cite. Nobody outside the labs knows what a language model checks, and anyone claiming to have that list is guessing.

Present: a verified Google Business Profile, consistent name, address and phone details across the main directories, and a 4.7 star rating from 38 lifetime reviews.

Missing:

• A named human on any service page. The site described a business, not a person.
• Working service schema. Three of seven service pages had broken markup, and there was no FAQ schema anywhere.
• Press or third-party mentions in the last 24 months. None.
• Citations beyond the basic directory listings. Almost nothing.

A 5 is low, and most of what was missing is editorial rather than technical. That is the useful part: editorial work is cheap.

What was actually wasting the ad budget?

Quick Answer: Two problems on the same page. The broken schema was making the page harder for search and answer engines to read. Separately, the page had no named author, no reviews above the fold and a six-click path to a booking, and that is the part paid traffic was landing on. Two leaks, two mechanisms.

The broken schema was real. We ran Google's Rich Results Test on the destination page and the Service markup did not describe the service the page was selling. Three of his seven service pages had the same fault. Structured data is one of the ways a search engine or an answer engine works out what a page is for, and when the markup contradicts the page you are asking a machine to guess.

What the broken schema was not doing was raising his ad costs. Meta does not read schema.org structured data when it decides who sees an ad or what a click costs. I say that plainly because I have heard agencies claim otherwise this year. Meta also retired the single "relevance score" in 2019, replacing it with three ad relevance diagnostics, so any story that turns on a relevance score moving is about a metric that no longer exists.

What paid traffic actually met sat one layer up from the markup. No named human, no reviews visible before you scrolled, and a booking path six clicks long. A stranger arrived on a page that gave them no reason to trust the business and no quick way to act. Campaign tuning does not fix that.

So the schema fix was about being legible to search and answer engines, and the page and funnel rebuild was about what a visitor does when they arrive. The owner had been arguing about the one channel that sends an invoice.

What was the review problem?

Quick Answer: 38 lifetime reviews at 4.7 stars, and nothing new in twelve weeks. Strong rating, silent feed. He had no system for asking, so reviews arrived only when a customer felt strongly enough to volunteer one.

Reviews carry real weight locally. Whitespark's 2026 Local Search Ranking Factors puts review signals at roughly 19 to 20% of local pack weighting, and in that survey quantity ranks above recency, so I will not tell you a fresh trickle beats a large back catalogue. Google's own local ranking documentation says more reviews and positive ratings can help your ranking. It says nothing about velocity.

As Ambitions' own observed practice rather than a documented ranking rule: a stopped feed reads badly to a human buyer, and a business with no asking system has no way to grow the quantity that does count.

The fix was small. After every completed job, an automated text goes out with a link to the Google Business Profile.

How many clicks did it take to book a job?

Quick Answer: Six, from homepage to confirmed booking. I asked him to book a job on his own site from a clean browser, and he could not complete it.

The path ran homepage, services menu, service category, service page, contact form, confirmation screen, then an emailed link to the scheduler. One of those steps landed in an inbox, so the booking depended on the customer coming back later.

Every extra step is another chance to leave, and a path you cannot describe out loud is a path a stranger cannot follow. His dashboards never flagged it, because analytics tells you where people leave, not that the path itself was the problem.

We rebuilt it to two clicks: homepage to a service page with the scheduler embedded, plus his name, his face and his reviews above the fold. Same traffic, same ads, shorter path, and a page that answered the trust question before it asked for the booking.

Why did the AI engines not know anybody worked there?

Quick Answer: There were no named people on the site. No team page, no author on any service page.

His About page named the business and described the vans. It did not name a human being. A page that does not name a human gives an answer engine nothing to credit and gives a buyer nobody to hold responsible. That is an argument from how attribution works, not a published rule, and no engine publishes what it weights.

The fix took about thirty minutes per page: the owner's name, a photograph, his credentials and years in the trade, plus the lead technician on the deeper pages. Cheapest work on the list, and the one owners resist most.

What was the fix order?

Quick Answer: Schema first, then named authors, then reviews automation, then the funnel rebuild, then citation work as an ongoing job.

• Schema repair across the seven service pages: one day. Ambitions prices a schema install at roughly £400 to £500.
• Named authors and credentials: two days, done by the owner, so no fee.
• Reviews automation setup: half a day.
• Funnel rebuild to two clicks: one day.
• Citation and press work: ongoing, and the slowest to show results.

Schema goes first because it is cheap and it makes everything downstream legible to search and answer engines. Citation work goes last, because there is no point pointing the world at pages that are not yet worth citing.

Should you skip the diagnosis and just buy ads?

Quick Answer: You can, and you will get what this owner got. Traffic delivered to a page that could not convert it, and a growing certainty that ads do not work. Ads work. They multiply whatever the page underneath them already does.

He spent three months of budget before booking a Diagnosis. Ambitions prices a Diagnosis at £497 to £1,500, and the order is the same every time: diagnose, fix the layer underneath, then turn the ads back on.

If you only do one thing after reading this, open your own site in a private browser window and try to book yourself a job. Count the clicks. Then run your main service page through the Rich Results Test. Twenty minutes, no cost.

Liyana's insight

The moment I look forward to in every Diagnosis is the one where the owner stops defending the channel he has been arguing about for months. He blamed the ads because the ads send an invoice. Nobody invoices you for broken markup or a booking form six clicks long, so those sit there for years while everybody shouts about something else. These owners are not careless. This man could talk me through a commercial hot water system in exhausting detail. Nobody had told him where to look.

FAQ

Why was this client anonymised?

He asked to be, and we do not publish a name without the client's agreement. Where figures come from a client, they are self-reported and not independently audited, and we say so.

Could he have found the schema problem himself?

Yes, in about five minutes with the Rich Results Test. Nobody had told him to look. A good part of a Diagnosis is pointing the owner at the right test.

Will fixing schema lower my ad costs?

No. Meta does not read schema.org data when it decides who sees your ad or what a click costs. Schema helps search engines and answer engines understand your page. If your ad costs are the problem, look at the page the ad points at: the offer, the proof, the named human and the number of clicks to a booking.

Will every Diagnosis find leaks this size?

Some are smaller, some are worse, and it is worth running even on a growing business, because growth hides leaks. Two to three weeks from kickoff to the written report and the ranked fix list.

Do I have to buy the fixes from you?

No. The report is yours, with the fix order in it. Plenty of owners hand it to their existing developer, and the named-author work is usually theirs to do anyway.

Sources

• Google Rich Results Test
• Whitespark, 2026 Local Search Ranking Factors
• Google Business Profile, how to improve local ranking