Book a free review →

178% more conversions. 57% lower CPA. The first thing we fixed was what counted as a conversion.

78% of their conversions weren't conversions. That was just the beginning.

Client Exterior cleaning, UK
Results after 3 months of management
178% ↑ up Quote requests
57% ↓ down CPA
20% ↑ up Ad spend

The account had been optimising on numbers that didn't exist.

This Yorkshire-based company offers professional exterior cleaning — roofs, driveways, patios, gutters. The kind of work homeowners put off until something forces the issue, and then want done quickly by someone they trust. The account had been running for a year. Spend was relatively stable, CPA had been around £20 during the summer, but had crept up to £49 by the time they came to us.

The problem was that the account was measuring the wrong things and spending on the wrong searches. 78% of reported conversions were clicks to call or clicks to email — a tap on a phone number, a click on an email address. No call connected. No message sent. No job discussed. These were being counted identically to quote requests and callback forms, which meant the reported CPA of £70 in Search was masking a real CPA closer to £94.

Underneath that: smart bidding paying up to £35 per click on irrelevant searches, DIY-intent keywords absorbing hundreds of pounds with zero real conversions, and a best-performing keyword only showing on a quarter of available searches because its bid was too low. One ad running in every ad group, unchanged, with no keyword relevance and no CTA.

"Smart bidding without clean conversion data doesn't optimise. It guesses. And when it guesses with £35 clicks on searches that were never going to book a roof clean, the budget disappears before it can find the searches that would have."

Six issues. The CPA problem was the least of them.

01
78% of conversions were low-value clicks Click-to-call and click-to-email were counted as primary conversions alongside genuine quote requests. The gutter cleaning campaign looked like it had a CPA of £102. Its real CPA was £323. The account had never seen that number.
02
Smart bidding paying the wrong amounts for everything With no real conversion data to learn from, smart bidding was spending up to £33.97 per click on a competitor's business name, £28.37 on "patio cleaner pressure washer", and £24.20 on "gutter cleaner camera". None of these were going to book a job.
03
Keywords pulling in DIYers "Clean my patio" was triggering searches like "screwfix patio cleaner" and "how can I clean my patio". "Driveway cleaning" had spent £452 with no real conversions. "Block driveway cleaning" was attracting people looking for tips and products. Budget going to people who wanted to do it themselves.
04
Best keywords being underbid "Roof cleaning near me" had a real CPA of £38 — well below account average — but was only showing on 25% of searches because the bid was too low. More budget was flowing to the wrong end of the account than the right one.
05
One ad per ad group, no testing, no keyword relevance Every ad group had a single ad running unchanged. Headlines made no mention of the service being searched for and included no CTA. Google was assembling whatever combination it liked.
06
Location performance ignored Manchester had a CPA of £194. York had a CPA of £33. No bid adjustments had ever been applied. The account was paying the same for a click from Hull as a click from Lincoln, despite a 4x difference in what those clicks were worth.

Tracking first. Then everything else.

TRACKING
Tracking rebuilt from scratch Click-to-call and click-to-email removed from the primary conversion count. Proper conversion tracking set up for quote requests, callback forms, and calls from the website above a meaningful duration threshold. The account now had an honest picture of what was actually happening.
BIDDING
Smart bidding replaced with manual CPC Bids set manually based on keyword relevance and historical CPA data where it existed. High-intent keywords like "roof cleaning near me" given significantly higher bids to capture the impression share they'd been missing. Low-intent and unproven keywords bid down or removed.
KEYWORDS
Negative keywords expanded and DIY intent blocked Product names, chemical brands, and how-to terms added as broad match negatives at account level. The keywords that were attracting them stayed — "driveway cleaning" and "clean my patio" are valid searches — but the irrelevant traffic they were pulling in was blocked. DIY-product searches, competitor names, and material searches cut off at the root.
ADS
Ads rewritten across all ad groups Keyword in headline one. Clear CTA in headline two. Two ads per ad group minimum, with headlines and descriptions pinned to prevent Google serving irrelevant combinations. Service-specific copy replacing the generic "one stop shop" messaging that had been running since launch.
LOCATIONS
Location bid adjustments applied Bids increased for high-performing cities, reduced for poor-performing ones. Manchester bid down. York, Liverpool, and Lincoln bid up. The account now spent more where conversions were cheaper and less where they weren't.

The growth in conversions came almost entirely from efficiency, not from putting more money in.

Spend increased by only 20%. The CPC reduction reflects what happens when smart bidding is replaced with manual control grounded in real conversion data: Google stops paying £34 for clicks on competitors' names and starts finding the searches that actually book jobs.

The 57% reduction in cost per conversion is the compounded result of fixing tracking, removing wasted spend, and bidding more on what was working. Each change individually would have moved the number. Together they transformed it.

178% Quote requests
57% Lower CPA
20% Spend increase
“

The account had the ingredients for good performance — the right services, real demand, a business that does the work properly. What it was missing was an honest read of what was actually converting, and bids that reflected that. Smart bidding without clean conversion data doesn't optimise. It guesses. And when it guesses with £35 clicks on searches that were never going to book a roof clean, the budget disappears before it can find the searches that would have. Three months of clean data changes what the account can do next.

Melissa Ros · Founder
More Results
Management · Insurance · UK 160% Revenue up 160%. ROAS up 192%. We changed the goal first.
Read more →
Audit · Exterior Cleaning · UK 57% 178% more conversions. 57% lower CPA.
Read more →

Want an honest look at your account?

20 minutes. Enough to spot the major issues, understand what you need, and give you a realistic idea of what we'd do and what it would cost.

Get a free review →