Should 20% of Your Meta Budget Optimize for Something Other Than Purchases?

A brand I was chatting to last year had a quarter where Meta just stopped working, so they did the thing that feels obviously right and was completely wrong.

Their purchase numbers had gone soft. Cost per acquisition was climbing, ROAS was sliding, and the gut reaction was "we need to warm people up first". So they carved off a big chunk of the account, pointed it at view-content and add-to-cart events, and let it run. Within a few days the dashboard looked incredible. Cheap add-to-carts everywhere. The mid-funnel campaigns were "crushing" on cost per result.

And the business did not make a single dollar more.

That's the trap I want to talk through, because the question of whether a slice of your Meta budget should optimise for something other than purchases is a genuinely good one. I think the answer is sometimes yes. But the brands that get a real win from it and the brands that quietly waste a quarter are doing the exact same thing on the surface. The difference is underneath.

The signal you give Meta is the signal it optimises for

Here's the thing most people skip past. Meta is not trying to grow your business. It's trying to get you the cheapest possible volume of whatever event you asked for.

If you tell it to find purchases, it goes and finds the people most likely to buy. If you tell it to find add-to-carts, it goes and finds the people most likely to add to cart, which is a different and much larger crowd, and plenty of them will never get their wallet out. So the moment you switch a campaign to a mid-funnel event, you have changed what "winning" means. The algorithm will happily hand you a flood of cheap actions that look like progress and don't pay rent.

I keep coming back to a line that's worth tattooing somewhere: the signal you give Meta is the signal it optimises for. If what you actually want is profitable new customers, asking the machine for add-to-carts and hoping the purchases follow is a bit like ordering a salad and being surprised it's not a steak.

So mid-funnel optimisation isn't good or bad on its own. It's a tool with a very specific job, and the whole game is knowing whether you've got the job it's for.

When it's a real reach play

The case where I genuinely like it is when you've diagnosed a reach problem, not a conversion problem.

Picture a brand spending ~$60k a month, creative that's already proven it can sell, healthy purchase campaigns. But growth has stalled. When you look at rolling reach, you can see it plainly: in the good months they were getting in front of a lot of new people, and in the flat months that fresh reach dried up. The purchase-optimised campaigns had basically picked the low-hanging fruit and were now circling the same warm pool over and over.

That's the situation where pointing some budget at an upper-funnel event earns its keep. You're not using it as a crutch for weak ads. You're using it to make Meta cast a wider net and pour new people into the top, people the purchase objective was too narrow to ever reach. The mid-funnel event there is a reach instrument, plain and simple.

And the value it creates often doesn't even show up inside Meta. I've seen upper-funnel reach pull serious lift in branded search and direct traffic, the stuff that lands as "no attribution" in your reporting. One brand running a view-content reach play saw a roughly 35% jump in branded queries over the test window, which makes complete sense. You put your product in front of thousands of new people, a chunk of them go and look you up later. Meta never gets the credit, but the till still rings.

When it's a bandaid for weak creative

Now the other case, which is the one my opening brand was actually in.

If your purchase numbers are sliding because the creative has gone stale, mid-funnel does nothing but hide it. You'll get cheaper add-to-carts, sure, because add-to-carts are easier to buy than purchases. The dashboard turns green and everyone exhales. But the people clicking through still hit the same tired ads and the same offer, and they still don't buy. You haven't fixed the leak, you've just moved your eyes to a metric that doesn't measure the leak.

This is the bit I'd push back on hardest if you brought it to me. Mid-funnel optimisation cannot manufacture demand. It can widen the net or shift the optimisation target, but if the ads aren't earning attention and the offer isn't compelling, a softer event just lets you lose money more cheaply and feel busier doing it.

My blunt test: if you turned the mid-funnel campaigns off tomorrow, would you expect blended new-customer revenue to drop? If the honest answer is "I'm not sure, but the cost per add-to-cart was great", you're using it as a bandaid.

The exclusions trap that fakes a win

There's one more thing that has to be right before any of this means anything, and it's the least glamorous part of the whole job: your exclusions.

The brand I mentioned at the top had a quiet problem underneath the creative one. Their mid-funnel campaigns weren't properly excluding recent purchasers and existing customers, so a big share of those lovely cheap add-to-carts were just... existing fans, adding things to cart like they always do. Of course the cost per result looked amazing. The campaign was optimising against the easiest audience on the platform, the people who were going to engage anyway.

When you fix that and force the campaign to actually find new people, two things happen. The reported cost per event gets uglier, because you've taken away the cheat. And your real number, cost per incremental order, the orders that genuinely happened because of the ad, starts to mean something. I've watched a mid-funnel test go from "looks brilliant, does nothing" to a real ~25% improvement in cost per incremental order purely because the exclusions got cleaned up first. Same campaigns. The only thing that changed was who Meta was allowed to count.

So before you judge a mid-funnel test at all, make sure it's hunting for strangers, not feeding you your own customers back at a discount.

How I'd actually decide

If a founder asked me whether to put 20% of their Meta budget on something other than purchases, I'd walk through this before touching the account.

• Is it a reach problem or a conversion problem? Pull rolling new-customer reach across your good and bad months. If reach falls off a cliff in the flat months, mid funnel meta ads optimization is worth a real look. If reach is steady and conversion is what's sliding, fix the creative first.
• Is the creative already proven? Mid-funnel is for scaling what works to more people, not for rescuing ads that have stopped working. If you wouldn't bet on these ads in a purchase campaign, don't promote them.
• Are your exclusions watertight? Recent purchasers and existing customers out, properly. Otherwise the test will lie to you on day one.
• Are you measuring the right thing? Judge it on incremental new-customer orders and blended results, not on in-platform cost per add-to-cart. The cheap event is the bait, not the goal.

If those four line up, I'd run it as a clean split: carve out around 20% of the budget, maybe 10% on a view-content event and 10% on a higher-intent event like a quiz completion if you've got one, and leave the rest on purchases. You usually don't need new creative to start, which is what makes it such a fast test. Give it a couple of weeks, hold everything else still, and read the blended numbers, not the vanity column.

What you're really doing there is asking a precise question: does giving Meta a wider signal bring me net-new people I couldn't reach otherwise? Sometimes the answer is a clear yes and you've found a fresh vein to scale into. Sometimes it's a no, and the kinder thing is that you found out in a fortnight instead of a quarter.

If your mid-funnel campaigns are sitting there looking like your best performers and you can't quite say whether they're adding real revenue, that's usually a sign the measurement and the exclusions deserve a proper going-over before you scale a single dollar further. We do that kind of teardown for clients all the time, and more often than not the "winning" mid-funnel campaign is the first thing the numbers quietly throw into question. So before you crown it: which of your campaigns is actually bringing you strangers, and which is just charging you to talk to people you already had?