Meta ads interrupt people who weren't looking. Google Search ads answer people who were. For a service business — the plumber, the accountant, the repairer — that difference is everything: "hot water not working Preston" is a person with a problem, money, and a deadline. Search is where you buy that moment.
It's also an interface built to spend your money on Google's terms. Here's the setup that keeps the intent and declines the padding.
The structure: one service, one ad group, one page
Small account, small structure:
- One campaign per service line you actually want more of (emergency call-outs; tax returns; screen repairs) — not one campaign for "everything we do".
- A handful of keywords per ad group, in phrase or exact match.
"emergency plumber","hot water repairs",[blocked drain plumber]. Broad match hands the steering wheel to the machine before it has any conversion data to steer with — it can earn a place later, not on day one. - Every ad group points at a matching page — not the homepage. The ad promised hot water repairs; the page delivers hot water repairs, with the phone number at the top. (The page is half the account's performance.)
Negatives: the half of the account nobody builds
The fastest way to waste search budget is paying for almost-right searches. Before launch, load the obvious negative keywords:
free,DIY,how to,course,salary,jobs— researchers and job-seekers, not buyers- Brands you don't service, suburbs you don't serve
cheap, if you're not — a click from someone shopping on price alone is a click you bought for a competitor
Then, weekly for the first month, open the search terms report and read what you actually paid for. It's reliably humbling, and every bad term becomes a negative. Ten minutes a week here does more than any bidding cleverness.
Settings where the money leaks
The defaults favour spend, not you. On a new small account:
- Search partners and Display expansion: off. Both extend your "search" budget onto lower-intent placements. Buy pure search first; expand only from proven ground.
- Location: "presence" not "interest". The default includes people interested in your area — which for "plumber Melbourne" includes someone in Manila researching. Target people physically in your service area.
- Ad schedule matched to reality. If nobody answers the phone after 6pm, either stop paying for after-6pm calls or point them at a booking form and say "we'll call at 7:30am".
- Call reporting on, conversions defined. Calls from ads, quote-form submits, booking clicks — marked as conversions before launch. An account optimising toward nothing optimises toward clicks, and clicks are Google's product, not yours. (Same discipline as GA4.)
Bidding, in plain terms
Start with Maximise Clicks with a firm CPC ceiling, purely to gather data. Once the account has a few dozen conversions, switch to Maximise Conversions — the automation is genuinely good when it has your data to learn from, and directionless before that. Skip the exotic strategies; a small account never generates enough signal to feed them.
The number that governs everything is the same one from the budget post: what's a customer worth, and what does an enquiry cost? A $90 cost-per-call sounds dear until the average job is $600; a $9 click is cheap until none of them ring. Judge the account monthly, at the enquiry level, and adjust the one variable at a time.
Search vs Meta: which first?
Simple test: do people already search for your thing with buying intent? Emergency trades, repairs, professional services, "near me" categories — search first; the demand exists, you're just standing where it looks. Products people don't know exist, visual impulse categories, new brands — Meta first; you have to create the want before anyone can search for it. Budget for one channel done properly before splitting to both — concentration beats coverage at every small-budget scale.
FAQ
What about Performance Max? Later. PMax bundles search, display, YouTube and more, optimising across them with limited visibility for you. It can work with strong conversion data; on a fresh account it's a black box eating a small budget. Earn your data on plain search first.
Should I bid on my own business name? If competitors do, yes — those clicks are cheap and defensive. If nobody's bidding on it, your organic and map pack listings already catch it; spend elsewhere.
Google's rep keeps calling with "optimisation" suggestions. The recommendations tab and the calls are sales, not audit. Some suggestions are fine; auto-applying them is how tight accounts go broad. Decline auto-apply and change one thing at a time, on your own read of the search terms report.
Want the account built like this — structure, negatives, pages, and call tracking — as a reviewable task? Send it over.