Meta Ads for Local Service Businesses: The 2026 Playbook

Most local service businesses that try Meta ads conclude the platform does not work for them. They run a campaign for a month, spend a few hundred dollars, get some junk leads and a couple of tyre kickers, and go back to Google Ads or referrals.

The platform works. The way most local businesses are told to run it does not. Meta is fundamentally an interruption channel, which means it behaves nothing like search, and almost every failure we audit traces back to running it as though it were search.

This is the playbook we use for local service businesses: what to run, in what order, what to expect it to cost, and the specific mistakes that waste the first month of budget.

Search and social are opposite problems

On Google, someone types "emergency plumber" plus their city. They have the problem, they know they need the solution, and they are looking for a vendor right now. Your job is to be visible and credible at that moment. Intent is given, and you are competing on presence and price.

On Meta, nobody is looking for you. They are looking at photos of a friend's holiday. Your ad interrupts that. Nobody has a problem in mind, so nothing is given and everything has to be created by the ad itself.

This single difference explains most of what follows. On search you optimise targeting and bids. On social you optimise creative and offer, because the creative is what determines who stops, and who stops is your real targeting.

The order to run things in

There is a correct sequence and running it out of order is the most common expensive mistake.

First, make sure you can convert. If your website does not clearly say what you do, where you do it, and how to book, do not send paid traffic to it. Paid traffic magnifies whatever your conversion rate already is. If that is close to zero, more traffic just costs more money.

Second, run retargeting before prospecting. Everyone does this backwards. Retargeting reaches people who already visited your site, so it is cheaper, converts better, and proves your offer works before you spend on cold audiences. If retargeting cannot convert people who already showed interest, cold traffic will not save you.

Third, run a single cold campaign with broad targeting and multiple creatives. Not five campaigns. One, with enough budget to actually exit the learning phase.

Fourth, scale what works and kill what does not, on a schedule.

Targeting: less is more than you think

The instinct is to narrow the audience: homeowners, aged 35 to 65, interested in home improvement, within five kilometres. This felt right in 2016. In 2026 it usually hurts you.

Meta's delivery system finds people likely to take your optimisation event, and it needs volume to learn. Narrow audiences starve it. The practical guidance from Meta's own performance documentation is to give the system room and let it find the buyers.

For a local service business, this usually means: a geographic radius that matches where you actually work, minimum age if legally or practically relevant, and nothing else. Let the creative do the filtering.

The one exception is genuinely small service areas. If your radius contains 40,000 people, you cannot also layer interests on top without starving delivery. In small markets, geography is your only targeting.

Set the radius by drive time, not distance. A 25 kilometre radius from a shop in the northeast covers very different territory than the same radius downtown. Use the actual area you are willing to service, and exclude the parts you are not, because a lead from somewhere you will not drive is worse than no lead.

Creative is your targeting

If you take one thing from this: on Meta, the ad selects the audience. A video of a bathroom renovation self-selects people renovating bathrooms far more precisely than any interest checkbox.

What works for local service:

Real work, real people, real premises. Before and after shots, the actual crew, the actual truck, the actual clinic. Local businesses have an advantage national advertisers cannot buy, which is that you are genuinely local and can prove it. Stock photography throws that advantage away.

Vertical video shot on a phone. Not polished brand films. A 20 second clip of the owner explaining what a job typically costs and what affects the price outperforms an expensive production because it looks like content rather than an ad.

The first two seconds carry the whole ad. The overwhelming majority of people scroll past. The hook is the entire game: state the problem, show the transformation, or say the specific thing your customer is worried about.

Name the place. Your city name plus "homeowners" in the first line of copy dramatically improves relevance. It also filters out people outside your service area before they cost you a click.

Run at least four concepts at once, not four variations of one concept. Different angles: price transparency, speed of service, a specific problem, social proof. You are looking for which angle resonates, and you cannot find that by testing button colours.

The offer decides whether any of it works

Cold audiences do not book a 6,000 dollar service off a single ad. The offer has to match how cold the audience is.

Offers that work for local service businesses:

• A genuinely free, genuinely useful inspection or assessment with no obligation
• A specific fixed price on an entry service that leads naturally to more work
• A real quote in a stated timeframe, where the speed is the promise
• Seasonal urgency where the urgency is real, such as pre-winter furnace servicing
• An educational lead magnet for high-consideration purchases, such as a realistic cost guide

Offers that do not work: "contact us for a quote," "20 percent off" with no context, and anything requiring a big commitment from someone who met you 4 seconds ago.

The most reliable improvement we make to underperforming local campaigns is not targeting or creative. It is reducing what the ad asks for.

Lead forms versus landing pages

Meta's instant forms load natively and convert at a much higher rate because there is no page load and fields pre-fill. The trade-off is lead quality, because low friction means people who barely read the ad still submit.

Our default for local service: instant forms with friction added deliberately. Use the higher intent form option, add one or two qualifying questions that require actual typing, and set expectations in the form description about what happens next and what things typically cost.

That last point filters more effectively than anything else. If your minimum job is 2,000 dollars, saying so in the form removes the people who were never going to buy. You will get fewer leads and more customers, and cost per lead will rise while cost per job falls. Optimise for the second number.

Use a landing page instead when the purchase needs real explanation, when you need to establish credibility before the ask, or when you want the retargeting audience that a site visit creates. If you go this route, the page has to load fast on mobile data, because Google's research on mobile page speed has long shown that bounce rates climb sharply with load time, and ad traffic is the least patient traffic there is.

Speed to lead is the whole conversion rate

This matters more than anything else in this article and it costs nothing.

A Meta lead is not a search lead. They were not looking for you. Their intent decays by the minute. The difference between calling in 5 minutes and calling in an hour is enormous, and the difference between 5 minutes and the next day is close to total.

Every local service business we work with gets the same first change: automatic text message within 60 seconds of form submission, automatic email, notification to whoever calls leads, and a follow-up sequence over the following week for anyone who does not answer. Most leads are lost to silence, not to competitors.

This is not really an ads problem, which is why ads people miss it. It is an automation and follow-up problem, and it is usually the highest return change available.

Budget and what to expect

Realistic guidance for a local service business:

Minimum viable budget is around 30 to 50 dollars per day per campaign. Below that, you generate too few conversion events for delivery to optimise, and you spend the whole month in the learning phase. Running 10 dollars a day across four campaigns is the worst possible configuration.

Commit to at least 6 weeks before judging. Two weeks is noise.

Expect cost per lead to vary enormously by trade. A local service lead might cost 15 dollars in one category and 150 in another. The number that matters is cost per booked job, which is cost per lead divided by your close rate. A 90 dollar lead that closes half the time at 3,000 dollars is excellent. A 12 dollar lead that never closes is worthless.

Track cost per job, not cost per lead. Every optimisation decision changes if you look at the second number instead of the first.

The mistakes that waste month one

• Boosting posts instead of running campaigns. Boosting optimises for engagement, not leads. It is the fastest way to spend money on likes.
• Too many campaigns at low budgets. Consolidate.
• Editing live ad sets constantly. Every significant edit resets learning. Set it up, then leave it alone for at least a week.
• No pixel or conversions API. Without conversion tracking, delivery is flying blind and you cannot tell which creative works. Install the pixel before spending anything.
• Judging by likes and comments. Some of the highest performing local ads have terrible engagement metrics.
• Never refreshing creative. Local audiences are small, so frequency climbs fast. In a small market you may need new creative every three to four weeks.
• No follow-up system. Covered above, and it is the biggest one.

Attribution: why the numbers never agree

At some point Meta will report 40 leads and your CRM will show 22, and someone will conclude the platform is lying. It is worth understanding why before you make budget decisions on the discrepancy.

Meta reports conversions using its own attribution window, by default counting someone who clicked within 7 days or viewed within 1 day and then converted. It also attributes across devices when it can identify the same person. Your CRM counts what actually arrived. These measure different things, so they will never match exactly.

Two practical consequences. First, view-through conversions inflate the apparent performance of awareness style campaigns, because someone who saw an ad and then searched your name gets counted by Meta even though the search engine did the closing. Second, since Apple's App Tracking Transparency changes, a meaningful share of iOS conversions are modelled rather than observed, which means some of the number is a statistical estimate.

What to do about it: pick one source of truth and manage to it. For local service businesses that should be your CRM or booking system, because that is where a real job exists. Use Meta's reporting to compare ads against each other, which it does reliably, and use your CRM to decide whether the channel as a whole is profitable.

Install the conversions API alongside the browser pixel. Sending conversion events server side recovers a meaningful portion of what browser tracking now misses, and it improves delivery because the system learns from more complete data. This is a one-time technical setup and it is worth doing before you scale spend.

Add a "how did you hear about us" field to your intake, and accept that a portion of people will say "Google" when they first saw you on Facebook. That is not an error in your tracking. That is how multi-touch buying actually works, and it is a good argument for judging channels on total booked revenue over a period rather than on last-click attribution.

A 90 day plan

Days 1 to 14. Install the pixel and conversions API. Fix the conversion path. Build the automated follow-up sequence. Produce four creative concepts. Launch retargeting only.

Days 15 to 45. Launch one cold campaign, broad targeting, four creatives, 40 dollars a day minimum. Leave it alone for the first week. Kill clear losers at day 10, add two new concepts.

Days 46 to 90. Scale winners by increasing budget gradually rather than in jumps. Keep introducing new creative. Start measuring cost per booked job rather than cost per lead. Add a lookalike audience built on actual customers, not on leads.

By day 90 you should know your cost per booked job and whether the channel is viable for you. If it is not viable by then, the usual cause is either an offer that asks too much of a cold audience or a follow-up gap, not the platform.

If you would rather have this run for you, that is what our Meta ads work covers. But the pixel, the follow-up automation, and an honest look at your offer are worth doing regardless of who runs the campaigns, because they determine whether any amount of ad spend can work.