Do Meta ads work for B2B? Yes, but not for the reason you have been told

Why is the usual targeting advice out of date?

Because Meta has spent years reducing detailed targeting and moving budget toward broad audiences its own system picks. Advice built on stacking job titles and employer interests is describing a control panel that keeps getting smaller.

The direction of travel has been consistent for years now. Sensitive interest categories went first. Detailed options kept getting folded into each other after that, and the recommended setup drifted toward handing the system a wide pool and letting it go find the responders on its own. You can still narrow. The platform increasingly performs worse when you do.

This matters because the entire case against Meta for B2B rested on targeting. The argument went: LinkedIn knows where people work, Meta does not, therefore LinkedIn. If precision targeting is no longer the mechanism on either platform, that argument needs rebuilding rather than repeating.

The honest framing is different. Meta sells attention at a low price to a very wide audience. Whether that is useful to you depends almost entirely on how efficiently your creative can throw away the people you do not want.

What is fundamentally different about the click?

There is no intent behind it. On search someone described a problem and you answered. On Meta you interrupted a person who was looking at something else, so the ad has to create the interest it then captures.

This one distinction explains most of what goes wrong. Teams take the Google Ads playbook, point it at Meta, keep the same landing page and the same demo request, and conclude the channel does not work for B2B. What actually failed was asking a stranger mid-scroll for the same commitment a searcher was already prepared to make.

It also explains why Meta looks terrible under last-click reporting and often looks fine when you measure it properly. Interruption channels create demand that gets captured later by branded search or a direct visit, and the model hands the credit to whoever the buyer touched last.

So the workable version of Meta for B2B is either much earlier in the journey than your search campaigns, or much later. It is rarely a straight substitute for either.

Which B2B companies does it actually suit?

Ones with a wide buyer definition. A short decision helps, and so does having something visual to show, or an audience already worth retargeting. The worst fit is the narrow committee purchase carrying a six figure price tag, which is also the most common disappointment.

The pattern is that Meta rewards breadth. If ten thousand companies could plausibly buy from you, being able to reach a hundred thousand people cheaply is an asset. If forty companies in the country can buy from you, that same cheapness is just waste with good margins for the platform.

How does creative do the qualifying?

By naming the audience and the situation out loud in the first two seconds, so the wrong people scroll past without costing you anything. On a broad audience the creative is the targeting.

This inverts the usual instinct, which is to write the broadest possible hook so that more people click. On a channel where you pay per click and your buyer pool is a small slice of the audience, a broad hook is an expensive mistake.

A useful test before launching: if a plumber, a student, and a chief financial officer all saw this, would two of them keep scrolling. If everyone would click, the ad is going to be costly.

• Name the role or the trade in the opening line, not in the body copy
• Name the specific situation, not the category, so it reads as recognition rather than an offer
• Put a price or a scale marker in early if it filters the audience usefully
• Show the actual work, since a real artifact repels people who wanted something else
• Let the wrong audience self-select out before the click, because clicks are what you pay for

Why do the numbers look worse than they are?

Browser and app level tracking has been progressively restricted, so a meaningful share of conversions never gets attributed back. Server side reporting through the Conversions API recovers some of it, and last-click models still undercount interruption channels by design.

Apple's tracking prompt changed what is observable on iOS, and browsers have continued tightening third party cookies since. The practical result is that some of the pipeline Meta influenced shows up as direct traffic, branded search, or nothing at all.

Two responses are worth the effort. Send conversions server side so the platform learns from events the browser dropped, and add a how did you hear about us field to your form. Self-reported attribution is imprecise and it catches things no model does, particularly for a channel people see and act on days later.

The strongest test remains a holdout. Turn the channel off in one region or for one period and watch what happens to total inbound. It is blunt, it costs real money, and it answers the question honestly, which no dashboard will do for you here.

Should you use instant lead forms?

Only with friction added on purpose. The default form is prefilled and takes two taps, which produces a large volume of leads at a low cost per lead and a quality level that will embarrass you in front of the sales team.

Meta gives you the choice between a volume-optimized form and a higher intent version, plus the ability to add qualifying questions. Take the friction. Ask for the company, ask for the situation, ask something a casual tapper will not bother answering.

Then feed the outcome back. If a lead form is the conversion event, the system will find you more people who complete lead forms, which is the same trap that hollows out Google Ads accounts. What you want it learning from is the leads that turned into conversations.

For most B2B offers we would rather send the click to a page that explains the thing properly. The form is faster and cheaper per lead, and cheaper per lead is not the goal.

How should you test it without wasting a quarter?

Retarget first, prove the pipeline exists, and only then go cold and broad. Most failed B2B Meta tests started cold, spent thinly across four audiences, and learned nothing they could act on.

Start with people who already visited the site or watched a video. That audience is small, cheap, and it tells you whether creative and offer can move anyone at all before you pay to build an audience from scratch.

Then commit properly to one cold test. One broad audience. Three to five creative angles that are genuinely different from each other, not four crops of the same picture. Enough budget to get out of the learning phase, and a date when you stop. Splitting a small budget across many audiences guarantees nothing gathers enough data to be judged, which feels careful and is the most reliable way to waste money here.

Write down what would count as success before it launches. Qualified conversations, not leads. Then hold the test for the length of your sales cycle before deciding, because deciding earlier means deciding on noise.