Most outdoor DTC brands tell us the same thing on the first call: "We optimize for ROAS. Our agency manages to a target. Performance is fine, but we can't seem to scale."
The problem is hidden in that sentence. ROAS isn't actually a goal you give to Meta. It's a number you read after the fact. The thing you're actually optimizing for is whichever conversion event you've set as your campaign objective, usually "Purchase," fired by every transaction on the site and valued at the order total. Meta then goes and finds the cheapest possible humans who will trigger that event.
That sounds fine until you look at what it actually means in practice. "Purchase" treats a $42 one-time first-time buyer of a $42 SKU exactly the same as a $180 repeat customer buying your highest-margin hero product. The algorithm has no idea which one matters more to your business, because you never told it. So it goes and finds more of whichever one is cheaper to acquire. Almost always, that's the wrong one.
This is the discipline we've been calling signal engineering internally for the last year, and it's quietly become one of the highest-leverage things we do for clients. The premise is simple: ad networks optimize for what you tell them success looks like, not for what success actually is. If you give them a vague signal, they give you vague customers. If you give them a sharp signal, they give you the customers that actually grow the business.
The three levers left
When iOS 14 broke audience targeting and Meta's automation tools (Advantage+, Andromeda) took over delivery, the lever count for advertisers collapsed. There used to be a dozen things you could tune. Now there are three:
- Budget
- Creative
- The conversion signal you send back to the platform
That's it. Almost every brand we talk to has spent serious time on the first two and almost no time on the third. They're treating the conversion event like a default setting. It isn't. It's the steering wheel.
What "garbage in, garbage out" actually looks like:
Here's a real pattern we see at $10M+ outdoor brands:
The brand sells a mix of SKUs — some entry-level products at $30-50 that anyone might buy on a whim and a hero product line at $180-400; that's where the actual margin lives. Their Meta campaigns are optimizing for "Purchase," weighted by order value.
What does Meta do with that? It finds the cheapest path to any purchase. Which means it floods the funnel with bargain hunters buying the $42 SKU, because those conversions are easier and the math on a $42 sale still shows up as a "purchase" in the algorithm's training data. ROAS looks okay on paper. New customer LTV is quietly tanking. The brand can't figure out why scaling spend kills efficiency and why their best customers seem to be coming from anywhere except paid social.
The signal they're sending says "find me anyone who'll buy anything." That's exactly what Meta delivers.
The fix: engineered conversion events
The work we do isn't complicated, but it requires being deliberate about something most brands have on autopilot. The basic moves:
Fire a custom event for new customer purchases only. Not all purchases. New-to-file customers. This is usually a CAPI event with a flag based on whether the email or customer ID exists in your historical orders. Meta now optimizes against the conversions you actually need, new acquisition, instead of conversions you'd get anyway from existing customers in retargeting overlap.
Fire a custom event for hero / high-margin SKU purchases. Whichever 2-3 SKUs are the real profit drivers in the business. When that event becomes the optimization target on prospecting campaigns, Meta starts hunting for the customers most likely to buy those specific products, which are usually higher-intent, higher-LTV, and far more profitable than the bargain shoppers chasing entry SKUs.
Use those events as your primary campaign objectives. Not "Purchase." Not "Add to Cart." The engineered events. The platform now spends every dollar trying to find the kind of customer your business actually needs more of, instead of the kind that just happens to be cheapest.
The improvement isn't subtle. We've seen accounts cut new-customer CAC by 30-50% inside the first 60 days from this single change, with no creative refresh, no budget change, and no audience restructure. Just a sharper signal going back to the algorithm.
Why this works mechanically
Two things are happening under the hood.
First, you've narrowed what Meta is solving for. The algorithm is extraordinarily good at finding cheap conversions, but it can only find cheap conversions of whatever event you've nominated. By firing a more specific event, you're forcing it to find a more specific (and more valuable) kind of human.
Second, you're feeding the algorithm cleaner training data. Every conversion event teaches Meta something about who responds to your creative. If half those events are low-quality customers, the model learns to find more low-quality customers. If every event represents an actual high-value purchase, the model learns to find more of those. The signal becomes a flywheel.
This is why brands that engineer their signals tend to see compounding gains: month two is better than month one, and month three is better than month two, while brands optimizing for a generic "Purchase" tend to plateau and then slowly decay as the algorithm grinds toward the cheapest possible conversion.
What to actually do this week
If you want to pressure-test this on your own account, the diagnostic is straightforward:
Look at your Meta campaigns and ask three questions. First, what conversion event are you optimizing for? If the answer is "Purchase" with no qualifier, that's the problem. Second, do you know what percentage of your last 90 days of Meta-attributed purchases were new customers versus existing customers in retargeting overlap? If you don't, you're flying blind on the metric that actually matters. Third, do you know which 2-3 SKUs drive the majority of your contribution margin? If yes, are those the SKUs you're optimizing your prospecting toward? If not, you're paying Meta to bring you the wrong buyers.
A real fix takes a working Conversions API setup, server-side event mapping, and someone who knows how to debug the data flow when (not if) events misfire. It's not trivial. But it's also not exotic. It's just deliberate.
The bigger point
Most brands obsess over creative testing—which is right; creative is one of the three remaining levers. But they ignore the lever sitting right next to it. Then, they wonder why their ROAS targets feel impossible to hit at scale.
You don't choose a conversion event. You design one.
The brands that figure this out aren't doing anything magical. They've just stopped accepting Meta's defaults as if those defaults were built for their business. They weren't. The defaults are built for the platform. Your job, or your agency's job, is to engineer signals that point Meta at the customers your P&L actually needs.
H Street is a growth and demand generation agency built for outdoor DTC brands. If you want us to audit your Meta event setup and identify which signals you should actually be optimizing for, reply to this email and we'll set up a call.
