Who Meta Ads actually works for: from e-commerce to local business

The question is usually asked backwards. People ask whether Meta Ads works, as though the platform were a machine with a pass rate. It works for a fairly predictable set of businesses and fails for another fairly predictable set, and which one you are in has almost nothing to do with budget or creative talent.

Start with what the platform is. Google is a search engine: somebody types a problem and you answer it. Meta is a discovery engine: nobody typed anything. A person is scrolling with no purchase in mind, and delivery decides that this particular ad is worth interrupting them with, based on what it has learned about people who behaved like they do.

That single difference decides everything else. On Meta you are not capturing demand, you are creating it, which means you pay for attention before anyone has admitted they want the thing. For some businesses that is the cheapest customer acquisition available. For others it means paying to educate an audience that will then go and buy from whoever ranks first in search.

Two questions that settle it

Before any of the vertical-by-vertical discussion, two questions do most of the work.

Can the product be understood in three seconds by someone who was not looking for it? Not fully explained. Understood well enough to want to know more. A visible physical product clears this instantly. A service with an obvious pain behind it usually clears it. Enterprise software with a nine-month evaluation cycle does not, at least not with a direct-response ad.

Is the audience broad enough that random interruption is not wasteful? Meta's delivery is good at finding narrow audiences inside broad ones, but the maths still has to work. If your buyer is one in fifty thousand and the deal is worth two hundred euros, the impressions spent on the other forty-nine thousand nine hundred and ninety-nine are the actual cost of the channel.

Where it works well

E-commerce and D2C

The native case. Visual product, impulse-compatible price, purchase completed in one session, and a conversion event the pixel can see immediately. Catalogue campaigns and retargeting do genuine work here rather than the theatrical work they do elsewhere. The constraint is margin, not policy: what a customer costs against what they are worth over a year, and whether the tracking reporting either number can be trusted. That is the argument on our e-commerce page and it does not get easier with a bigger budget.

Local services with visible results

Restaurants, delivery, salons, clinics, gyms, home services. Meta's real strength for these is not targeting sophistication, it is geography plus a picture of the outcome. A person four streets away who was not thinking about dinner can be given a reason to think about dinner. What breaks these accounts is usually not the ads: it is a delivery radius nobody configured, or a business that buys first orders while the repeat rate quietly leaks.

Consumer apps and subscriptions

Low friction to try, decision made on the phone, event tracking that can be built properly. The trap is optimising towards the install or the free trial, which is cheap to measure and tells you almost nothing. If the platform is taught to find people who install and never open, it will do exactly that, faithfully and at scale.

Info products and courses

Works, and is also the category where policy bites people who assumed they were mainstream. Income claims, before-and-after framing, and anything implying a guaranteed outcome fall under rules that are enforced by machine and applied to the landing page as much as to the ad.

B2C services with an emotional trigger

Legal, immigration, insurance, education. These sit oddly: the demand is real but often already articulated, which usually means search. Meta earns its place when the trigger is emotional rather than procedural, or when the person does not yet know the service exists. It also runs into special ad category rules in employment, credit and housing, where detailed targeting is simply removed and the creative has to carry everything.

Where it works badly

Not "never". Badly, meaning the same money produces more elsewhere.

• Long-cycle B2B with a buying committee. The click is affordable and the pipeline is not attributable for two quarters. Meta can support this as an awareness layer, but judged on last-click leads it will always lose to search.
• Emergency and immediate-need services. A burst pipe is a search query. Nobody discovers a plumber on Instagram at two in the morning.
• Products nobody can picture. If explaining it takes a diagram, the interruption is over before the explanation starts.
• Businesses with no measurable conversion event. Meta's optimisation is only as good as the signal it is given. An account where the goal is "brand awareness" and nothing is instrumented is buying impressions and calling it strategy.
• Genuinely prohibited categories. Explicit adult content is not restricted, it is banned, and no operational discipline changes that. The honest plan there is specialist networks and organic, which is what our adult page says.

The reach argument, honestly stated

Meta reports more than three billion people using its apps daily across the family. That number gets quoted to prove the platform suits everyone, and it proves the opposite: at that scale, almost everybody's audience is in there somewhere, so the question is never availability. It is whether the interruption is worth paying for in your particular arithmetic.

If both questions at the top came out yes, the channel is a fair bet and the rest is execution. If they came out no, a bigger budget on Meta buys a more expensive version of the same answer, and the money belongs in search or in organic. We would rather say that in the first conversation than in month four, which is also the point of the channel comparison we published earlier.