The short answer
Facebook ads reach people who were not looking for you, so they suit seasonal offers, long-consideration work, hiring, and remarketing. Use the Leads objective rather than the Boost button, install the pixel first, keep the audience wide, and give it a month.
Facebook does not have demand. It interrupts it.
Somebody with a burst pipe opens Google. Nobody scrolls Facebook hoping to find a plumber. That single difference decides most of what follows, and it is the reason a boosted post about emergency service usually produces likes and nothing else.
What Facebook does well is put an offer in front of people who were not looking. That is worth real money for some jobs and close to nothing for others, so start by deciding which one you are.
When is it worth running?
Four situations, in our experience. A dated seasonal offer, like a pre-season tune-up with a deadline. Work with a long consideration window, such as a kitchen remodel, orthodontics, or a roof replacement with financing, where nobody searches on the day they decide. Remarketing to people who already visited your website. And hiring, which is the quietest good use of the platform, because trades fill open positions there faster than on job boards.
If none of those describe you right now, put the money into search and your Google Business Profile and come back to this later. That is not us being modest, it is just where the return is.
Set it up so you can tell what happened
Put the Meta pixel on your site and add the Conversions API if your site supports it, because browser tracking alone now misses a meaningful share of events. Then define an actual conversion event: a form submission or a call, not a page view. If you skip this step, the platform optimizes toward the only thing it can see, which is clicks, and you will be paying for the cheapest clicks available.
Give it a few days to register traffic before the first campaign goes live. A pixel with no history is starting from nothing.
Then stay out of the Boost button. Boosting a post is the default path and it optimizes for engagement, which means the system finds people who like and comment. Those are not the same people who fill in a form. Open Ads Manager and choose the Leads objective, or Sales if you sell online.
Set the campaign to optimize for the conversion event you defined. Then keep the structure small: one campaign, one or two ad sets, three or four ads. Splitting a small budget across eight ad sets means none of them ever gets enough data to learn from.
Targeting: wider than feels right
The instinct is to narrow. Homeowners, aged 35 to 65, interested in home improvement, within four miles. That audience is too small to deliver against, and the interest categories are guesses the platform makes about people anyway.
Set a geographic radius or a list of towns you genuinely serve, set the age floor if it matters, and stop. Let the system find the buyers, because it is better at that than the interest checkboxes are. If you have a customer list, upload it and use it as a source for a lookalike audience. That is the one targeting input worth the effort.
Budget, and the mistake everyone makes in week one
The platform needs a steady flow of your conversion event before delivery settles down. At a small daily budget with a low lead rate, that can take weeks, which is why a campaign judged on day three always looks like a failure.
Pick a budget you can run for a full month without flinching, then leave it alone. Every edit to targeting, budget, or creative restarts the learning. Owners who check the account daily tend to edit daily, and the campaign never gets past the noisy phase. Look on day seven, change nothing unless something is clearly broken, and make your real judgment at the end of the month.
The creative that works is boring
A photograph of your actual crew, your actual truck, or the actual finished job. Not stock. Video of the work in progress, shot vertically on a phone, usually beats anything produced. The first line of text should say what you do and where, because it is the only line most people read.
Make three or four versions and let them run against each other. You will be wrong about which one wins often enough that guessing is not worth defending.
How to tell whether it worked
Count booked jobs, not leads. Social leads convert at a lower rate than search leads because the person was not looking for you, so a cost per lead that looks great can hide a cost per job that is worse than search. Ask every new customer where they heard about you and write it down, because platform attribution and your booking calendar rarely agree.
If a month of steady spend produces no booked work, stop. The channel is not right for the offer, and running it another month at a higher budget will not change that.
Follow-up questions
Enough to run a full month without stopping, since anything shorter tells you nothing. Decide the total you can lose without regret, divide by thirty, and run that daily. Half a month at twice the budget is the worse version of the same test.
Placements are set in the same campaign, so leaving both on is usually fine and lets the system spend where it performs. Check the placement breakdown after a few weeks and turn off anything spending real money with no results.
Because boosting optimizes for engagement, and the platform delivered exactly what it was asked for. Rebuild it in Ads Manager with the Leads objective and a conversion event, and the same budget will find different people.
Let’s build something.
Twenty minutes on the phone and you’ll know whether we’re the right fit. Either way you’ll leave with something useful.