The account was running paid social because the prior agency said it had to.
The business: a home services firm in Atlanta, 14 employees, $4M annual revenue, six service lines, primarily residential customers. The prior agency had been running $11,000 per month split across Meta ($6K), TikTok ($3K), and Pinterest ($2K), plus a $3,500 Google Search campaign as “bottom-funnel capture.”
The reported numbers: Meta and TikTok were producing lead form submissions at $35-50 per lead. Pinterest was producing $80 per lead. Google Search was producing $110 per lead.
The reported dashboard looked balanced. The actual business was choking. Booked jobs (the only metric that mattered to the owner) had been flat for nine months. The lead-to-booked-job conversion rate was 8% on Meta, 5% on TikTok, 3% on Pinterest, and 41% on Google Search. The cheap leads were not turning into work. The expensive leads were.
The math the prior agency had not run
CPL had been the agency’s headline metric in monthly reports. The owner had been told the account was efficient because CPL was low.
The real math on booked-job cost:
Meta: $35 CPL × 0.08 conversion-to-booked rate = $438 per booked job. TikTok: $40 CPL × 0.05 = $800 per booked job. Pinterest: $80 CPL × 0.03 = $2,667 per booked job. Google Search: $110 CPL × 0.41 = $268 per booked job.
The cheapest channel by reported CPL was the second-most expensive channel by actual booked-job cost. The most expensive channel by CPL was the cheapest channel by booked-job cost. The dashboard had been telling a story that was true at the lead level and false at the revenue level.
Total monthly booked-job cost on the existing mix: roughly $670 weighted average. Total booked jobs per month: 21. Total spend: $14,500 including agency retainer.
The recommendation that took a meeting
We recommended killing all three paid social channels and consolidating the budget into Google Search and Local Service Ads.
The marketing person at the firm pushed back. The narrative had been that the brand needed multi-channel presence to be modern and credible. The Meta page had been built up over two years. Killing paid social felt like abandoning an asset.
The owner pushed back differently. The owner asked the right question: would killing paid social mean losing the brand-awareness benefit those channels provided? The honest answer: the brand-awareness benefit was real but small, and could be re-captured later if the math justified it. Right now, the math did not.
The agreement: 90-day test. Kill paid social entirely. Move the freed budget to Google Search and Local Service Ads. Measure on booked jobs, not on leads.
What happened in 90 days
The reallocation:
Meta, TikTok, Pinterest: paused. Google Search: lifted from $3,500 to $8,500 monthly. Local Service Ads: launched at $3,500 monthly. Total monthly spend: $12,000, down from $14,500 (the prior agency retainer came off the books).
The booked-job results after 90 days:
Google Search produced 32 booked jobs per month at $266 per booked job, with $8,500 in spend producing more booked-job volume at lower per-job cost than the entire prior account had produced.
Local Service Ads produced 9 booked jobs per month at $389 per booked job.
Total booked jobs per month rose from 21 to 41, a 95% increase at lower total spend. The cost per booked job dropped from $670 to $292 1.
The owner’s revenue numbers in month 4 of the new arrangement were 47% above the trailing-three-month average pre-test. The growth has held through subsequent quarters.
Why the paid social spend had failed
Three structural reasons that generalize across service businesses.
The first reason is intent. A homeowner who needs an emergency plumber tomorrow is searching on Google. A homeowner scrolling Instagram is not in-market for the service that day. Meta and TikTok lead forms collect contact information from people who are mildly interested at the moment of the ad; few of them convert to scheduled work because the underlying intent was low.
The second reason is sales-cycle friction. Service businesses sell on phone calls and scheduling. A Meta lead form requires follow-up, qualification, scheduling, and conversion. A Google Search click goes directly to a contact form or a phone call from a buyer who is already in-market. The conversion math reflects the difference in friction.
The third reason is geographic targeting precision. Local Service Ads charge per booked call, not per click, and only serve to users in the service area searching for the service. The unit economics are structurally aligned with the business in a way that Meta’s broad-targeting model is not.
The pattern this illustrates
This is a service-business variant of the more-channels piece. Concentration on the channels whose mechanics match the business beats diversification across channels whose mechanics do not.
The pattern repeats across service categories. Roofers, HVAC firms, plumbers, electricians, landscapers, cleaning services, attorneys, accountants: the buyer’s intent is mostly captured at the moment of search, not at the moment of scrolling. Google Search and Local Service Ads are the channels whose mechanics match. Paid social is, for most service businesses, a tax on the budget that produces leads which sales cannot close.
The honest measurement to run on any service business account: pull the leads from each channel for the past 90 days, mark which ones became booked jobs, and calculate the cost per booked job by channel. The dashboard CPL is the wrong metric. The cost-per-booked-job is the right one. The two numbers usually disagree by 3-10x in favor of Google Search and Local Service Ads.
What this is not saying
Paid social is not useless for service businesses. There are categories where it earns the spend: home services with strong visual appeal (interior design, custom renovations, landscape design), services with long consideration windows where awareness compounds, and brands building local recognition above and beyond capturing immediate demand.
For most service businesses spending under $20K monthly, none of those conditions is the dominant one. The dominant condition is: how do I get more booked jobs this quarter at acceptable cost. The answer for that condition is concentration on Google Search and Local Service Ads.
The firm’s position on this, written into the /atlanta page for local Atlanta service businesses, holds across geography and category. Match the channel to the business. Most service business operators have not done the math. The math usually says the same thing.
The booked-job number is the only number the owner cares about. The agency’s job is to make that number larger. The honest read of which channels do that is the read that survives the test.