Case study · Google Ads takeover

A locksmith spending $10,380 a month. 34% less spend, and calls that pay.

An established mobile locksmith in the Orlando area handed us Search campaigns that were busy, expensive and measuring the wrong thing. This is the 30 days before we touched it against the 30 days after, 22 August to 20 September 2026, with the owner's own dispatch log beside the Google numbers.

Book a Free Consultation What the jobs paid

Search campaigns only, Local Services Ads excluded Booked jobs from the owner's dispatch log

After the takeover, 22 August to 20 September

130 measured leads on $6,883, down from $10,380 the month before

A lead here is a first-time phone call from the website, a call from the ad itself that lasted at least 60 seconds, a submitted form or a text. Nothing softer is counted.

34%less ad spend

$10,380 in the 30 days before, $6,883 in the 30 days after.

130measured leads

76 first-time calls, 48 ad calls of 60 seconds or more, 5 forms and 1 text.

$53cost per lead

$6,883.29 divided by 130.

$1.60collected per $1 of ads

From the owner's dispatch log, 1 to 17 September. Cash on the day only.

Not counted as leads in the after window: 12 repeat calls from people who had already called, 18 taps on a phone button, 3 ad calls logged under the old rule with no minimum length, and 2 Google forwarding calls. Local Services Ads run in the same account, are managed separately, and are left out of every figure on this page.

Where it started

The scoreboard said $33 a conversion. Most of them were not calls.

In the 30 days before the takeover the Search campaigns spent $10,380 on 957 clicks. The conversions column looked healthy, because it was adding together four different things.

What the account was counting214 taps

214 taps on a phone button and 6 page views were scored as conversions, beside 81 calls from ads with no minimum length and 17 Google forwarding calls. Google's bidding was being trained to find people who tap, not people who hire a locksmith.

What it counts now130 real leads

First-time calls matched to the click that caused them, ad calls only when they last 60 seconds, forms and texts. Repeat callers are tracked and kept out of the total. The bidding system now learns from the same thing the owner gets paid for.

The honest limit of a before and after: the old setup could see 98 phone calls in its 30 days, which works out to $106 per visible call, but it almost certainly missed calls made from the website, so the true before figure is unknown. That is why this page leads with spend and with the dispatch log, not with a before and after cost per lead.

What we changed

Six changes in the first two weeks

Nothing here needed more budget. Every change either stopped paying for a call the company could not serve, or fixed what the account was learning from.

Rebuilt conversion tracking around real calls

Phone-button taps and page views were demoted. First-time calls, repeat calls, texts and forms are now recorded separately, and only first-time contact counts as a lead.

A 60-second floor on calls from the ad

A call that ends in under a minute is a wrong number or a price check. Those no longer teach the bidding system anything.

Repaired a counter that had gone dark

Calls made straight from the ad stopped registering for a week in late August. It was fixed on 31 August, and 48 ad calls have been recorded since that would otherwise have been invisible.

Six negative keyword lists, 404 terms

Out of area (229 terms), do-it-yourself and job seekers (100), competitors and roadside clubs (37), services not offered (19), price shoppers (12) and European vehicles that need a dealer (7).

Eleven location exclusions

The dispatch log showed technicians being asked to drive to towns a long way down the state. Those areas, and other metros that kept leaking in, were excluded outright.

Let impressions fall on purpose

Impressions dropped 41 percent. The account stopped showing for browsers, other cities and cars the company does not cut keys for, while click-through rate rose from 4.08 to 4.33 percent.

Side by side

Thirty days before, thirty days after

Same campaigns, same service area, same company answering the phone.

Search campaigns	23 Jul to 21 Aug	22 Aug to 20 Sep	Change
Ad spend	$10,380	$6,883	-34%
Clicks	957	595	-38%
Impressions	23,433	13,746	-41%
Click-through rate	4.08%	4.33%	+0.25 pts
Average cost per click	$10.85	$11.57	+7%
Phone taps scored as conversions	214	0	removed
Measured leads	not comparable	130
Cost per measured lead	not comparable	$53

Cost per click went up 7 percent across the full window and we are not hiding it. Cheap clicks from the wrong towns were the first thing removed, so the clicks that remain are the contested ones.

Week by week

Weekly Search spend, before and after

Monday to Sunday. The red bars are the four full weeks before the takeover.

$2,43827 Jul

$2,7193 Aug

$2,26610 Aug

$2,18217 Aug

$2,16524 Aug

$1,31331 Aug

$1,2727 Sep

$1,78114 Sep

The week of 24 August is the changeover: tracking was rebuilt that week and the negative lists went in. The week of 14 September is the highest since, and still 26 percent under the average of the four weeks before the takeover.

The part Google cannot see

What the jobs paid, from the owner's dispatch log

The company logs every lead its dispatchers handle and what the technician collected. From 31 August to 17 September that log shows 87 leads from Google Ads, 36 booked jobs and $5,782 collected, against $3,653 of ad spend from 1 to 17 September.

41%of leads became jobs

36 of 87 in the dispatch log.

$101ad cost per booked job

$3,653 of spend across 36 jobs.

$42cost per logged lead

Lower than the Google figure because the log also catches calls Google cannot attribute.

1.6xcash against ad spend

Money collected on the day. Repeat customers and reviews are not in it.

Job type	Leads that booked	Average ticket	Why the rest were lost
Car keys and programming	about 50%	$245	No budget on the day, or a European car that needs the dealer
Car lockout	64%	$115	Customer did not have the $60 to $80 when the technician arrived
House lockout	63%	$170	Cancelled while the technician was on the way
Safes, motorcycles, car alarms	0%	none	Services the company does not offer

This is why the next round of negative keywords came from the dispatch log and not from Google: motorcycle keys, car alarms and door repair were costing clicks and could never book.

The honest part

What is still imperfect

Two things this page does not claim

It does not claim a before and after cost per lead, because the old tracking makes the before number unknowable. And 1.6x is not a return on ad spend in the usual sense: it is cash collected on the day against 17 days of spend, from a log the owner keeps, not from Google.

Some out-of-area searches still get through. In the after window two clicks came from a search for a locksmith in a town several hours away, and both were scored as conversions. When somebody searches a plain phrase and then asks the dispatcher to open a safe, no keyword can filter that. It has to be filtered on the phone.

Spanish-language searches are converting on English ads and an English page. A Spanish ad group and landing page is the next build, and it is not reflected in any number here.

Straight answers

Questions about this case study

Why is there no before and after cost per lead?

Because the account was counting 214 taps on a phone button and 6 page views as conversions in the month before the takeover, and it probably missed calls made from the website. Any before figure would be a guess, so the page compares spend, which is exact, and uses the owner's dispatch log for results.

Are Local Services Ads included in these numbers?

No. The company runs Local Services Ads in the same Google Ads account and manages them separately. Every figure on this page is Search campaigns only.

What does $1.60 collected per $1 mean?

The owner's dispatch log shows $5,782 collected by technicians from 36 booked jobs between 31 August and 17 September, against $3,653 of Google Ads spend from 1 to 17 September. It is cash on the day. It leaves out repeat customers, reviews and anything paid later.

Why did cost per click go up if the account improved?

The cheapest clicks were the ones from towns the company does not serve and searches it cannot fulfil. Removing them raises the average cost of what is left by 7 percent while cutting total spend by 34 percent.

Does a locksmith need Advanced Verification to run these ads?

Yes. Google requires Advanced Verification for locksmith advertisers in the United States and Canada before Search ads can serve. It is a separate check from the screening for Local Services Ads.

Is the client named?

No. The page publishes their ad spend and their takings, so the company is described by trade and area only.

Think your account is counting the wrong thing?

Send us read access. We will tell you what your conversions column is really counting, where the spend is leaking, and what the account looks like once only real calls are scored.

Book a Free Call

See the other case studies