Most small businesses that try social media ads quit within three months. Not because ads don’t work, but because they spent $400 boosting posts, got a few hundred likes from people who will never buy, and concluded the whole thing was a scam.
Boosting a post is the single most common mistake, and Facebook puts that big blue button right there because it’s the easiest way to take your money.
This guide is about doing it properly: picking one platform, running real campaigns in the ads manager, spending small until the numbers tell you to spend more, and building something that keeps working when you stop paying.
Decide what a result is before you spend anything
“Brand awareness” is what people say when they haven’t decided what they want. Pick one thing the ad should make someone do. Fill in a quote form. Book a call. Buy the $39 product. Join the email list.
Then work out what that action is worth to you. A plumber who closes 1 in 4 quote requests at an average job of $300 can afford to pay maybe $20 to $30 per lead and still come out ahead. A SaaS company with a $50/month plan and a 2% trial-to-paid rate has completely different math. If you don’t know these numbers, you can’t tell whether a campaign is working, and the platform’s dashboard will happily tell you it’s going great.
Write the target down. Something like “20 quote requests in 30 days at under $25 each, $500 total budget.” That is the whole strategy on one line.
Pick one platform, and make it the boring one
Everyone wants to be on TikTok. For most businesses selling to adults with money, Meta (Facebook and Instagram through the same ads manager) is where you should start, and you should probably stay there for six months before touching anything else.
The reasons are practical. The ad platform is the most mature, the targeting is the most forgiving of beginner mistakes, the audience is enormous across every age bracket, and there are ten years of free tutorials for every problem you’ll hit. Instagram ads run through the same system, so you get both placements for one learning curve.
LinkedIn is the exception. If you sell to businesses and the job title of your buyer matters (HR directors, dental practice owners, ecommerce managers), LinkedIn’s targeting is worth its much higher cost per click, which regularly runs $5 to $10. Just don’t go there with a $10/day budget. You’ll get four clicks and learn nothing.
TikTok, Pinterest, X, Reddit, YouTube: all valid, all second platforms. Master one first.
The disadvantages of social media marketing are real and worth reading before you commit, mostly because they explain why the platform you pick should never be the only place your customers can find you. More on that at the end.
Use the ads manager, not the boost button
Boosting a post optimises for engagement. The algorithm goes looking for people likely to like or comment, and it is extremely good at finding them. Those people are not buyers. They’re people who like things on Facebook.
In Meta Ads Manager you choose an objective, and for almost every business the right one is Leads or Sales, with the Pixel (or Conversions API) installed on your site so the platform knows when someone actually did the thing.
Now the algorithm goes looking for people who fill in forms and buy stuff. Same budget, completely different audience, and this one change accounts for more of the gap between “ads don’t work” and “ads work” than anything else in this article.
Installing the Pixel takes twenty minutes if you’re on WordPress or Shopify. Do it before you launch anything.
Targeting in 2026 is mostly letting go
Five years ago you’d stack interests, exclude age ranges, and layer behaviours until your audience was 40,000 people. That era is over. Meta’s Advantage+ audiences now outperform hand-built targeting for most advertisers most of the time, because the machine has more signal than you do.
What still matters:
Location, obviously. A dentist in Cluj doesn’t need to reach Bucharest.
A rough age floor if your product clearly excludes teenagers.
And, more than anything, a custom audience built from your existing customers. Upload your email list (even 300 addresses helps), tell Meta to find people who resemble them, and let it run broad from there. Your best cold audience is almost always “people like the people who already paid you.”
If you have no list yet, run broad with just location and age, and let the Pixel data teach the algorithm over the first two weeks.
The ad itself: what actually stops the scroll
The creative is where you should spend your time, because it’s the only part of this the algorithm can’t do for you.
A few things that hold up across almost every account I’ve seen:
Video beats static images for cold audiences, and it does not need to be polished. A 20-second phone clip of the owner explaining one specific problem they fix will outperform a stock photo with a logo on it every single time. Put captions on it. Most people watch with sound off.
The first line of copy has to name the problem, not the business. “Boiler making a banging noise?” gets read. “ABC Plumbing, serving Manchester since 1998” gets scrolled past.
Show the price, or a price range, if you can. It filters out people who were never going to buy and improves lead quality more than any targeting setting.
One ad, one offer, one button. Not “learn more about our range of services.”
Make four or five variations of the same ad before launch. Different opening lines, different thumbnails. Meta will spread spend across them and within a week you’ll know which one to keep. You are not smart enough to guess the winner in advance. Nobody is.
Budget: small, then boring, then bigger
Start at $10 to $15 a day for a single campaign. Not less, because below that the algorithm can’t gather enough data to exit its learning phase, and not more, because you’re about to make mistakes and there’s no reason to make expensive ones.
Then leave it alone for seven days. This is the hard part. The temptation to tweak on day two, when you’ve spent $30 and got one click, is enormous. Every edit resets the learning phase. Set a calendar reminder for one week out and go do something else.
At the seven-day check, look at exactly three things: cost per result (lead or sale, not click), the frequency (if the same people have seen it more than three times, the audience is too small), and which creative variant is winning. Kill the losers. If cost per result is under your target, increase budget by 20% and wait another week. If it’s double your target, the problem is almost always the creative or the landing page, not the targeting.
A campaign at $15/day that produces leads at $18 each is worth far more than a $100/day campaign you don’t understand.
Where the click lands matters as much as the ad
Sending paid traffic to your homepage is throwing away roughly half your budget. The person clicked because of one specific promise. The page they land on should repeat that promise, show one form or one button, and load in under three seconds on a phone.
For most service businesses this is a single page with the offer, a short form, two or three reviews, and the phone number. For products it’s the product page, not the category page. Check the page on your own phone on mobile data before launching, because that’s how 90% of your clicks will experience it.
Keep the organic side running, cheaply
Paid ads don’t work well on a dead page. When someone sees your ad and taps through to your profile, they should find recent posts, a filled-out bio, and evidence that you exist. This doesn’t mean posting daily. Two or three times a week is fine, and most of it can be batched on a Monday morning and scheduled. We compared the best free and cheapest social media schedulers for exactly this; the free tiers of most of them handle a small business comfortably.
Build the thing that doesn’t stop when you stop paying
Here is the uncomfortable truth about social ads: the moment you pause the campaign, the leads stop. Instantly. You’ve built nothing that lasts. And every year the auction gets more crowded, so the same $15 buys a little less.
The businesses that do well long-term run ads as a top-up on a foundation they own. That usually means an email list (collect it from every lead, even the ones who don’t buy) and a website that ranks in Google for the searches your customers actually type.
Search traffic compounds. A page that ranks for “emergency plumber Bristol” keeps producing leads in month 14 without you spending a cent that month. If your site has been neglected on that front, our SEO packages cover the technical fixes, content, and links needed to get it ranking, and it pairs well with paid social because the retargeting audience you build from organic visitors is the cheapest audience you’ll ever advertise to.
Run the ads. Just don’t let them be the only thing holding the business up.
