"What should we optimize our campaigns for?" sounds like a settled question with a settled answer — ROAS, obviously. It isn't settled at all. The right metric changes as the company grows, and copying a metric from a company at a different stage is one of the more expensive mistakes a growth team can make.
Early stage: cash flow is oxygen
At pre-seed and seed, the binding constraint isn't lifetime value — it's runway. You cannot fund a twelve-month payback out of nine months of cash. So the metric that matters is near-term: first-week or first-month ROAS, contribution margin on the initial purchase, how quickly a euro spent comes back as a euro earned. It's not that lifetime value doesn't exist yet; it's that you won't be around to collect it if the near-term math doesn't work. A pre-seed startup optimizing for d180 ROAS is a startup optimizing for its own funeral.
Scale-ups: the payback period can stretch
Once you've proven retention and raised real capital, the calculus changes. Now you can afford to wait, because you have evidence the money comes back and the runway to let it. This is where longer-horizon metrics earn their place: CAC payback measured in months, LTV:CAC ratios, d90 and d180 ROAS. You're deliberately accepting a slower return per cohort in exchange for more cohorts and a bigger base. The number you optimize for gets patient because the balance sheet can afford patience.
Established companies: portfolio and ARR thinking
At scale, no single campaign metric is the whole story. You're managing a portfolio — some spend for immediate return, some for long-payback growth, some for brand that won't show up in a last-click report at all. This is the altitude where marketing mix modeling, incrementality, and frameworks like Rule of 40 and ARR growth belong, because the question is no longer "did this campaign pay back" but "is the whole engine growing efficiently." ROAS is still there, just with a much longer window and a lot more context around it.
The metric encodes your risk tolerance
Here's the thing worth internalizing: a d7 ROAS target and a d365 LTV:CAC target aren't two ways of measuring the same thing — they're two different companies with two different appetites for risk. A short-window target says "I need the money back before I can breathe." A long-window target says "I can wait, and I'd rather own the bigger base." Neither is more sophisticated than the other. One is just honest about survival, the other about scale.
Pick the metric that matches your runway
So before you argue about which ROAS window or which LTV model is "correct," answer a more basic question: how long can you afford to wait for the money to come back? Then choose the target that matches that reality — not the one in the case study from a company three funding rounds ahead of you. The best growth teams don't optimize for the metric with the nicest name. They optimize for the metric that keeps the company alive at the stage it's actually in.