The Clicks Come, The Calls Don't: Where a London Repair Business's Ad Budget Actually Goes

The scenario below is not one company's case file. It is a composite: a pattern that repeats across appliance, boiler, locksmith and emergency repair accounts often enough to be predictable. If some of these lines describe your own account, that is exactly why this was written.

Every trade business that starts advertising in London ends up saying the same sentence: the clicks are easy, the calls are not.

What sits behind that sentence is usually competition. Thirty or forty other businesses are bidding on the same words, in the same postcodes, at the same hour, and not every click they push your way comes from a real customer. This piece looks at how that competition shows up inside your ad account, what order to investigate it in, and which parts of it you can control.

Say you have spent eleven years fixing washing machines and fridge-freezers inside a manufacturer's service network, and you have now gone out on your own. Walthamstow, Leyton, Hackney, as far as Ilford on a good day. You know the work inside out. What you do not know is advertising.

You also intend to do this properly. You buy a decent domain (short, memorable, with your trade in the name) and pay a premium for it, because the obvious ones are long gone. You commission a website. Signage, van livery, business cards, uniforms. Every line of it costs real money, and you pay it, because this is the business you are building. You approach the advertising budget with the same seriousness.

Then the first week goes like this:

➔ Monday: budget exhausted mid-afternoon. Two calls. One is a landlord asking whether you sell used machines.
➔ Wednesday: budget exhausted before midday. No calls at all.
➔ Friday: thirty-odd clicks, the whole budget, one small job booked.

At the end of the month, the jobs that came from a neighbour's recommendation outnumber the jobs that came from the ads. You reach the obvious conclusion, and it is the wrong one: advertising doesn't work for a business like mine.

What is usually going on: demand is not the issue. Fridges do not break less often in East London. What is squeezing you is competitive pressure, and it reaches your account through two separate doors: part of it is legitimate auction pricing, and part of it is traffic that was never going to become a customer.

Before looking for someone to blame, it helps to understand the auction you have walked into.

High intent means high price

Someone typing fridge repair London at 9pm on a Sunday has a broken fridge and food going off. That is close to the highest commercial intent a search engine ever sees, and every competitor within fifteen miles knows it.

Here is the part that gets missed: that click is priced by an auction, and your rivals are the ones filling it. The bids of thirty or forty businesses chasing the same customer are what pull your cost upward. The high CPC on your screen is not a fixed fee. It is a thermometer reading of how crowded your trade is.

The cost ladder for search clicks looks roughly like this:

The three multipliers stacked against you

➔ Category: repair and emergency services sit near the expensive end of the ladder to begin with.
➔ Geography: London CPCs run above the UK national average across most verticals.
➔ Quality Score: a thin, slow website with no clear call to action pushes the price up further. A well-optimised account can pay noticeably less per click than a poorly managed one bidding on identical keywords.

So your real budget was never "£60 a day." At, say, £8 to £10 a click, it is roughly six to eight genuine chances to win a job per day. Lose two or three of those to traffic that was never going to call, and the maths stops working immediately.

The infrastructure problems arrive before the advertising problems. They are worth naming one by one, because almost every small service business passes through the same doors.

The Website

Quoted at three weeks, delivered in seven. It looks acceptable on a laptop and fails everywhere that matters:

➔ The phone number is an image, not a tappable link, on mobile, where most emergency repair searches happen.
➔ The homepage leads with "About Us" instead of what you fix, where, and how fast you get there.
➔ There are no service-area pages, so washing machine repair Leyton has nothing specific to land on.
➔ Page load on 4G is over six seconds.

The Agency

The agency that takes over the ads charges a monthly fee plus spend. The report shows impressions up, click-through rate up, top impression rate "strong." The phone stays quiet.

➔ No account access. You can see the reports but not the account. You cannot pull a search terms report yourself.
➔ Broad match everywhere, with the Display Network and Search Partners left switched on by default.
➔ Slow response cycle. A question sent on Thursday is answered the following Wednesday, usually with "we'll optimise it Monday."
➔ The metrics reported are the metrics that always look good. Impressions and CTR are activity. Booked jobs are outcome. Nobody is reporting outcome.

The lesson here is not "agencies are bad." Plenty are excellent. The lesson is that an advertiser who cannot see their own click data cannot diagnose their own problem, and a report written by the party being evaluated rarely surfaces that problem unprompted.

Once you have account access and spend a weekend actually reading it, four patterns typically come out in this category.

1. Clicks at hours when nobody has a broken fridge

A visible share of clicks arrives between 01:00 and 04:00. Not impossible, since emergencies happen, but your analytics show these sessions lasting two to four seconds, with no scroll and no tap. A genuinely panicked customer behaves in exactly the opposite way.

2. The same handful of IPs, over and over: the part of competition nobody writes about

Neither Google Ads nor GA4 shows you visitor IP addresses, but server logs or a click-tracking tool do. And what they often reveal is a single address clicking your ads eight or nine times inside one week. Always weekdays, always during working hours, never converting. A real customer clicks once and either rings you or closes the tab.

One explanation comes up again and again in this trade, and it is unpleasant but simple: competitor clicking. The logic behind it is brutally rational. Your daily budget is fixed. The moment it runs out, your ad disappears for the rest of the day, and the only name left on that keyword belongs to your rival. A competitor does not need to outbid you to remove you from the auction. They only need to exhaust your budget early.

To be straight about it: not every invalid click is a competitor. Bots, scrapers, accidental taps inside mobile apps and click farms all sit inside that volume too. But in high-cost local service categories (appliance repair, locksmiths, emergency plumbing, vehicle recovery), competitor-driven clicking is the line item everyone in the trade knows about and few put in writing.

3. Traffic from far outside the service area

The location setting sits on the default option, which reaches people in or interested in the area, not only people in it. You are paying London-tier prices for clicks from cities you do not drive to.

4. Search terms you never meant to buy

Broad match quietly buys appliance repair course, fridge repair jobs, how to fix a washing machine yourself, appliance repair near me salary. Every one of those is a real person. Not one of them is a customer.

This is not a rare misfortune

Independent measurement has been putting numbers on this for years. Lunio's Global Invalid Traffic Report 2026, based on 2.7 billion clicks across six ad platforms, estimated that 8.51% of all paid ad traffic is invalid, roughly one click in twelve, amounting to about $63 billion wasted globally over the study period. The same research found that lead-generation businesses see invalid traffic rates around 32% higher than ecommerce, and lead generation is precisely the category a repair service falls into.

Measured on Google Ads alone, rates can run much higher. In ClickSambo's Invalid Click Report Q2 2026, 40.1% of tracked Google Ads clicks in Turkey were invalid. The exact figure varies by methodology, industry and country. The direction never does.

The compounding problem: wasted spend is only the visible half. The invisible half is signal corruption. Smart Bidding learns from the clicks and conversions it sees. Feed it junk leads and fake form fills, and it will confidently spend more of your money in exactly the wrong place. That is why clean conversion signals matter as much as blocking.

Do not rebuild everything at once. Working in cost-of-effort order, cheapest first, is both lower risk and the only way to see which change did what.

Step 1: Location targeting, presence not interest

Switch the location option to people in your service area only. This single change removes a surprising share of unqualified spend for local trades.

Step 2: Scale back what is underperforming, rather than switching it off

Break performance down by network before touching anything. Wherever returns are weakest and suspicious click patterns show up most (usually Display placements and Search Partners in this trade), pull the budget back first and move it to what is converting.

The word to hold onto is gradually. These networks are not worthless; they are simply the hardest to control at the start, when you have no data and no exclusion lists. So give them a small allowance rather than a full switch-off, watch how they behave once the traffic is being filtered, and open the tap further only as efficiency improves. A channel that earns its way up is far more durable than one you switched on at full budget and switched off in frustration.

Step 3: Match types and negatives

Replace broad match with phrase and exact for your core terms, and build a negative keyword list directly from the search terms report: course, jobs, salary, DIY, how to, free, parts only, manual, second hand.

Step 4: Ad scheduling

Pause ads between 01:00 and 06:00 to begin with. Reopen a narrow late-evening window later, once you can see which hours produce real calls.

Step 5: Fix the landing experience

Tappable phone number at the top. A one-line promise above the fold: same-day fridge and washing machine repair across East London. Separate pages for your main boroughs. Load time under two seconds.

Step 6: Manual IP exclusions, and their ceiling

You can add the worst offenders by hand, and you should. It works, briefly. Then it hits hard walls: Google Ads allows up to 500 IP exclusions per campaign, exclusions apply going forward rather than refunding what you already paid, and anyone on a mobile connection or a VPN can have a new address within minutes. Manual blocking is a bucket under a leak, not a repair.

Step 7: Automated, rule-based blocking

This is where a dedicated layer stops being optional. What it changes is response time: instead of finding a bad pattern three weeks later on a Sunday night, the pattern is detected on the click and excluded before it repeats. Tools such as ClickSambo's Google Ads protection handle this with:

➔ Click threshold rules that block an address on its second or third click within a defined window.
➔ Device fingerprinting that recognises the same device even when its IP changes.
➔ Bot, VPN and proxy detection at the click level.
➔ A click log with timestamps, geography and behaviour, so an invalid-click refund request to Google can be evidenced rather than asserted.

Your daily budget can stay exactly where it was. What changes is where it goes, and the chain runs like this.

1. Blocked sources stop costing you

The most direct gain. Every worthless click is money already deducted from your budget that returned nothing. Once a source is blocked, its repeat clicks stop eating into your budget, and more of it starts reaching real people.

2. The same budget buys more real clicks

Your budget stops emptying before noon. You stay live all day, including the evening hours, when many emergency searches happen.

3. Your cost per booked job comes down

The mechanism matters here: blocking fake clicks does not lower the auction price, and your cost per click may stay much the same. What changes is what each click is worth. When more of your clicks come from real people in your area, more of them call, and the cost of each booked job falls. A cleaner account also makes it easier to work on Quality Score properly, through relevant ads and a better landing page, and that is what brings click prices down over time.

4. More qualified leads for the same money

This is where the fall in cost per booked job actually comes from. Thirty clicks now means thirty people in East London with a broken appliance, not three browsers, five students and ten bots.

5. The phone starts ringing

The most convincing evidence is not a line on a chart. It is the WhatsApp messages waiting when you open up in the morning, and calls arriving back to back in the afternoon. Your phone tells you the ads are working before any report does. For a real-world example, see how an appliance repair business went from one call a month to 140 leads a month.

6. You move on to planning the next jobs

Once demand becomes steady, the question stops being "why aren't the ads working." It becomes when to buy the second van, whether it is time to take on another engineer, which borough to add to the service area.

The rest is your operation. No software books an appointment for you, dispatches an engineer, or turns up on time. All it can do is help your budget reach real customers. Everything after that is the trade you spent eleven years learning.

And here is what does not happen: the keyword itself does not get cheaper. Fridge repair London is expensive and will stay that way while the trade is this crowded. Nobody removes that cost. You can only stop paying it for people who were never going to call.

None of this means changing agency, either. Most of the time the agency was working with what it had: no exclusion lists, no clean conversion data, and a brief measured in impressions. Give them a cleaner account, full visibility and a target measured in booked jobs, and the relationship tends to improve on its own, usually because good agencies wanted to be judged on that all along.

Open your account and answer these seven questions. You do not need an agency to do it.

➔ Location setting: "Presence" or "Presence or interest"? If you cannot answer, it is almost certainly the latter.
➔ Search terms report, last 30 days: how many terms would you never have bought deliberately?
➔ Hour-of-day report: is there spend at hours your customers are asleep?
➔ Repeat click patterns: does any single source appear far more often than a real customer plausibly would? (This one needs click-level data from your server logs or a click-tracking tool.)
➔ Display Network and Search Partners: do you know what each is costing you and what it is returning, separately from Search?
➔ Mobile landing page: is your phone number a tappable link, above the fold?
➔ Reporting: does your monthly report contain the number of jobs booked, or only impressions and CTR?

If four or more of those answers make you uncomfortable, your problem is not that advertising doesn't work for repair businesses. Your problem is that in an auction that is already fiercely contested, you are paying full London rates for traffic you cannot see. The competition is expensive enough on its own. You do not also have to fund the clicks that were never going to call.