Value Rules: The Underused Meta Feature That Fixes Your Blended ROAS

Right now, today, Meta is almost certainly overpaying for a chunk of your customers and underpaying for your best ones. And it has no idea it's doing it.

Here's what that actually costs you. Say a fifth of your spend is going to a market or a segment that converts fine on Meta's dashboard but barely repeats, while your highest-value buyers are being throttled because the platform can't see they're worth more. On a $40k/month account, that's the better part of $8k/month chasing customers who'll never come back twice, while the ones who would are getting starved. Your in-platform ROAS looks acceptable. Your blended ROAS quietly bleeds.

The reason this happens is simple. Meta optimises towards the conversion event it can see: the purchase. It cannot see what happens after. It doesn't know that one buyer is worth three of another over the next year. So it treats them as equal, and your money flows accordingly.

Value rules are the feature that lets you correct this by hand. They've been around in rough form since mid-2025 and have got genuinely usable in the last few months. Most accounts I look at still haven't touched them. This is how I'd set them up, and just as importantly, how I'd prove they actually worked.

What value rules actually do

A value rule lets you tell Meta to bid up or down for a specific segment, based on information the platform can't access on its own. Lifetime value, repeat rate, margin: the things that live in your back end, not in the ad account.

You build them inside ad set settings by creating a rule set. You pick a criterion (gender, location, and a few others), then you tell Meta to increase or decrease the bid for that group.

That's the whole idea. You're feeding the algorithm a piece of business truth it was blind to, and letting it adjust spend to match. Two places I'd reach for first.

1. Bid down the geos that hog spend and don't repeat

This is the cleanest win, and it's where I'd start.

If you run a campaign across multiple countries, one big market tends to eat the budget. A large consumer market like the US will often gobble a disproportionate share of spend while sitting on a softer return, simply because it's where the volume is. Value rules let you group more countries into the same campaign and quietly downgrade the bid on the greedy one.

The move: open a rule set, select location, choose the country that's overspending and underperforming, and decrease the bid. Start small. Something like 10% is plenty for a first pass. Don't anchor to my number, 10% or 30% or 50% is yours to find, the point is to nudge in small increments and watch what it does to your spread of spend.

What you're hoping to see is the budget that market was hoarding redistribute towards geos with healthier returns, without you having to chop the campaign into a dozen separate ones. It keeps your structure consolidated, which Meta's algorithm prefers, while still giving you a hand on the wheel.

2. Bid up the segments your back end knows are worth more

The flip side, and the one that compounds over time.

The classic example: you know from your own data that one segment repeats far more than another. Maybe men come back and buy a second time at a noticeably higher rate than women for your product, or the reverse. Meta can't see that second purchase coming. You can.

So you build a rule set on gender, select the higher-LTV group, and increase the bid. Now you're willing to pay a bit more to acquire the customer who's worth more over twelve months, even if the first order looks identical to Meta.

The same logic extends to any segment where you hold the truth and the platform doesn't. The discipline is the same as everywhere else in this account: you're not guessing, you're encoding something you already know to be true from your CAC-to-LTV maths into the bid.

A warning before you get carried away

Value rules are deliberately granular. They cut against the grain of "set it and forget it" full automation, which is exactly why they're useful to us, and exactly why they can become a mess.

Don't build fifteen of them in an afternoon. Change one thing. Let it run. The trap is stacking five rules at once, seeing your numbers move, and having no idea which change caused what. One adjustment at a time is slower and far more honest.

Which brings us to the part most people skip entirely.

How to prove it worked, not just that the dashboard moved

Here's the catch with value rules. They're easy to fool yourself with.

You bid down a geo, your in-platform ROAS ticks up, and you declare victory. But did you actually make more money, or did you just shuffle the same revenue into a corner of the account that reports more flatteringly? Improving the number on the dashboard and improving the business are not the same thing, and value rules make it dangerously easy to confuse them.

This is where Meta's incremental attribution earns its keep. Introduced in April 2025, it changed what we can measure. Any ad set on the highest-volume strategy running maximise number of conversions can switch its attribution model from standard to incremental. Incremental only counts conversions that happened because of Meta: it strips out the purchases that would have occurred anyway, the ones the platform slid in front of at the last second to claim credit.

You don't even have to run your whole account that way to see it. Go to your columns, choose compare attribution settings, tick incremental attribution, and apply. Now you can see your reported purchases sit next to your incremental purchases side by side.

So here's the test I'd run around any value rule change.

• Before you touch anything, note your incremental purchases and your blended ROAS, not just the in-platform number. Blended is your spend against your total store revenue. That's the figure that pays the bills.
• Make one change. Bid down the one geo, or bid up the one segment. Nothing else moves.
• Give it room. A week to ten days minimum. Less than that and you're reading noise.
• Read the incremental column, not the headline. If incremental purchases held or climbed while blended ROAS improved, the change created real value. If your in-platform ROAS rose but incremental purchases fell and blended went sideways, you didn't fix anything. You just moved money to where it photographs better.

That last case is the one to watch for. A value rule that lifts dashboard ROAS while blended stays flat hasn't earned its place. It's cosmetic. Pull it.

The honest version of this work is slow. One rule, validated against incremental and blended numbers over a fortnight, then the next. It's far less satisfying than flicking ten toggles and watching a green number climb. But the green number lies, and blended ROAS measured against incremental conversions doesn't.

The question worth sitting with

Before you open a single rule set, the real question isn't which lever to pull. It's this: do you actually know, with numbers, which of your segments and geos are worth more over a year, and which just look busy on the dashboard?

Because value rules are only as good as the back-end truth you feed them. If you can't yet say with confidence that this market repeats and that one doesn't, that this customer is worth double that one, then the feature has nothing to act on. So maybe the first move isn't in the ad account at all. Maybe it's in your own data, working out where your real value actually sits.