Case study · a real account, not a worked example
Google Ads takeover: an enquiry from £52 to £26 in the first full month
A vehicle tuning and DPF cleaning workshop on Merseyside had rebuilt its own Google Ads and was paying more for every click than before. We were given access on 21 September 2026 and made the first changes three days later. In October, the first full month, an enquiry through the website form cost half what it had, and there were twice as many.
Written by Daniel Stoychev, who runs the account. Figures from the client's Google Ads account, read on 2 November 2026. The owner asked for the business not to be named, and it is not a Birmingham business.
£52£26
paid for each form enquiry from the adverts
25 Aug to 21 Sep against October 2026
1.32.1
form enquiries a day from the adverts
about 32 in 25 days against about 66 in 31
£1.50£1.08
the price of one click
7 to 20 Sep against October 2026
£71£56
spent a day, for more clicks a day
7 to 20 Sep against October 2026
The starting point
A busy account that could not say what it was buying
The workshop had run its Google Ads from one old campaign. On 7 September the owner replaced it with five campaigns of his own on a shared budget. The adverts were clicked more readily than before, but the price of a click went from about £1.13 to £1.50 and the spend from under £58 a day to over £71.
Two things were hidden behind that. The website the adverts led to was built on a site builder and counted nothing when a visitor tapped the phone number or the WhatsApp button, so a large share of real enquiries never reached the account. And the conversions that were imported were counted about twice for every real form. The bidding was being steered by a number that was wrong in both directions.
The negative keyword list had been bought ready-made and pasted in. Some of the words on it were the words this trade's customers type.
The log
What was changed, on which date, and why
Measurement first. Until every enquiry is counted once, a change to a bid or a budget is a guess.
| Date, 2026 | What changed | Why it comes at this point |
|---|---|---|
| 21 September | We are given admin access. Nothing is changed for three days: the account, the change history and the search terms are read first. | The first fix should be the one the reading points to, not the one that is quickest. |
| 24 September | Our own tags go live on the website, counting forms, phone taps and WhatsApp taps once each. The same day, the first 42 fixes: the negative list cleaned, four keywords added, thirteen sitelinks, one typo. | Counting starts before anything that depends on the count. |
| 1 October | The account begins bidding on the new counts. Calls of twenty seconds or longer and the tagged website actions become the conversions; the old doubled import is kept for reference only. | A week of clean data was in hand, enough to switch on. |
| 5 and 6 October | Prices in the adverts corrected to the real ones, the adverts switched off between midnight and five on weekdays, more negatives from the first fortnight of search terms. | Nobody answers at three in the morning, and a wrong price brings the wrong caller. |
| 7 October | A new website replaces the site-builder one and every advert and sitelink is pointed at its matching page. Mobile speed score 63 on the old home page, 98 on the new. | The page is where a paid click becomes an enquiry or does not. |
| 15 October | Each service is given its own daily budget, so the busiest one can no longer use up the money meant for the others. | A shared budget hides which service pays. |
| 25 October | Two adverts paused, and one added that is written around the fault a customer describes rather than the name of the service. | By now there were four weeks of clean counts to judge an advert on. |
Like for like
The same account in three windows
Clicks, spend and click price are counted by Google the same way in every period, so these three rows compare fairly. The number of clicks a day barely moved. What they cost did.
| Window | Who ran it | Clicks a day | Spent a day | Price of a click | Click rate |
|---|---|---|---|---|---|
| 7 to 20 September (14 days) | The owner, on his own rebuild | 47.6 | £71.30 | £1.50 | 9.9% |
| 21 to 30 September (10 days) | Handover: our first fixes from the 24th | 52.7 | £57.00 | £1.08 | 9.6% |
| 1 to 31 October (31 days) | Us | 52.2 | £56.20 | £1.08 | 8.8% |
The click rate eased from 9.9 to 8.8 per cent as the adverts were shown for a wider and cleaner set of searches. It is printed because a report that shows only the lines that went the right way is not a report.
How £52 and £26 were counted
Two figures, and how far each can be trusted
- Before: about £52
- From 25 August to 21 September the adverts cost £1,659. The website's inbox held about 41 real form enquiries in that time, and analytics put about three quarters of the site's visits down to the adverts: about 32 paid form enquiries in 25 days.
- After: £26
- In October the adverts cost £1,742 and the tag on the form's thank-you page counted about 66 enquiries from them in 31 days. Visitors who refuse cookies are not counted, so the true number is a little higher and the true cost a little lower.
- What else October showed
- With every route counted, October's adverts produced 221 contact actions in all: forms, WhatsApp taps, phone taps on the website and calls from the adverts themselves. None of the taps had been visible before, so there is no earlier figure to set beside it.
- How sure we are
- The before figure is built from an inbox count and an analytics share, the after figure from a tag, so the comparison is sound in direction and size but not to the pound. It is the one line that compares the same thing in both periods, which is why it is the one printed.
For your own account
Four checks from this account that a Birmingham business can run tonight
None needs us. Each takes a few minutes in an account that is already spending, and each was wrong in this one.
Read ninety days of search terms, sorted by cost
Count the rows that could never have bought from you. On this account the negative list was the problem in the other direction: it blocked words that buyers use.
Negative keywords for a West Midlands radiusRing your own advert and fill in your own form
Then look for both in the conversions column the next day. If either is missing, or the form shows up twice, the bidding is learning from the wrong numbers.
Call tracking for Birmingham adsOpen the page the advert lands on, on a phone, on mobile data
If it takes longer than you would wait, so does the click you paid for. The page is part of the account.
Landing pages for Birmingham adsDivide the spend by the enquiries, not by the clicks
The price of a click fell by 42 pence here. The figure that mattered to the owner was the £26 beside each enquiry.
Why your Google Ads are not converting
Want the same reading of your own account?
Leave a name and a number or an email. Daniel Stoychev sends a link request from our manager account, reads the account himself and writes back within 72 hours with six findings. We change nothing in the account.
- The check is free and written by the person who would run the account
- £195 a month flat after that, rolling monthly, no set-up fee, no VAT to add
- One business per sector in each area: the reply says whether yours is open