Meta ads for software companies: when they work and when they do not

Why would a software company advertise on Meta at all?

Because the people who should buy your software are not searching for it, and on Meta you do not need them to.

Search advertising is a harvest. It reaches people who have already decided they have a problem and started shopping for a solution. That is the best traffic there is, and for most software categories there is nowhere near enough of it, either because the category is young or because buyers describe their problem in words they would never type into a search box.

Meta reaches the same people earlier, while they are still living with the problem and have not yet framed it as something to go and solve. That is a harder sell and a much larger pool.

The trade is straightforward. Search buys intent that already exists and runs out. Social manufactures intent, more slowly and less predictably, and does not run out.

If your buyers already search for what you sell in real volume, run search first. Meta is for the demand that has not formed yet.

What has to be true before it works?

Four things, and they are conditions rather than preferences. A campaign missing any one of them produces the same failure.

You need an offer specific enough to interrupt somebody with. Not a demo, which asks for thirty minutes in exchange for nothing, but something that is worth the click on its own terms. You need creative that names the problem in the buyer's language rather than the category's.

You need somebody who answers a lead the same day, because a lead generated by interruption goes cold faster than one generated by a search. And you need an agreed definition of a qualified lead, in writing, before launch.

The last one sounds like process rather than performance, and it is the one that decides whether anything else you do is measurable.

Check all four before spending. Each is cheaper to fix now than after a quarter of data has been generated against the wrong definition.

1. An offer worth clicking on its own, not just a demo request.
2. Creative that names the problem the way the buyer says it.
3. Same-day follow-up by an actual person.
4. A written definition of a qualified lead, agreed with sales.

What should the first campaign actually test?

One question, and the question should be about the message rather than the machinery.

The most useful first test is two framings of the same problem for the same audience with the same offer. Hold everything else still. If one framing wins clearly, you have learned something about your buyer that will improve every asset you own, including your homepage and your sales calls. If neither wins, you have learned that the problem you named is not the one they feel, which is more valuable still.

Resist testing audiences first. At a small budget, splitting spend across audiences puts each below the volume the platform needs to optimise, and you get three unreadable results instead of one readable one.

Write the question down before launching. A test whose question was decided afterwards is a description of what happened, not a test.

How do you judge the first eight weeks?

On the quality of conversations, and on nothing the ad platform can tell you by itself.

The platform will report cost per lead and it will look either encouraging or alarming, and in the first weeks it means very little either way. What matters is what sales says about the people who arrived. Did they have the problem? Were they the right size? Did they know what they were signing up for?

So the reporting that decides this is a conversation with your salespeople, not a dashboard. Ask them to tag every lead by whether it was worth their time, and read the tags rather than the cost per lead.

Expect the first month to be exploratory and the second to be readable. A verdict formed in week three is a verdict about the learning period.

Cost per lead in the first month is a description of the learning period, not a result. The readable number is what sales says about who arrived.

When should a software company not do this?

More often than an agency will tell you, and the cases are recognisable in advance.

Do not start if nobody can answer leads within a day, because you will pay to generate conversations you then let go cold. Do not start if you cannot produce new creative on a rhythm, because social creative fatigues and a campaign starved of it decays no matter how well it was built. Do not start if the budget cannot clear the platform's learning threshold for any conversion event worth optimising towards.

And do not start if you need pipeline this quarter. Meta is a channel that gets better over quarters, and judged over six weeks it will look like a failure and get switched off just before it starts working.

None of these are reasons the channel does not work. They are reasons it will not work for you yet, which is a different and more useful thing to know.