The metrics that close loops

The most important metric in growth marketing isn't CAC. It's not ROAS. It's not LTV. It's the metric that closes a loop you're actually trying to test.

Most teams worship the holy trinity — CAC, ROAS, LTV/CAC — and miss the metric that would actually change their model. Those three numbers tell you the scoreboard. They don't tell you the play.

The trinity is necessary, not sufficient

The scoreboard metrics

• CAC — What you paid to acquire the customer. A floor, not a strategy.
• ROAS — What you got back, fast. Usually overstated by last-click; usually understated by long payback windows.
• LTV / CAC — What you'll get back, eventually. The ratio every operator quotes and almost nobody stress-tests across cohorts.

All three are essential. None of them, on their own, will tell you why you're winning or losing. They are output metrics — they describe the result, not the cause. If those are the only numbers you're reading, you're driving by the speedometer with the windshield painted over.

Three layers of metrics

The metric stack inside a serious growth program has three layers.

The three layers

• Input — Things you control. Creative volume per week, bid logic, audience segmentation depth, budget pacing, landing-page test cadence. These are levers, not scoreboards.
• Output — Things you measure. CAC, ROAS, payback period, conversion rate, contribution margin. The scoreboard.
• Learning — Things that change your model. Incrementality lift, MMM contribution, geo-holdout deltas, brand-search lift, cohort retention shape. The why-machine.

Most teams over-index on output. They report it daily. They under-invest in input — they don't track creative throughput or test velocity. And they ignore learning entirely — incrementality and MMM live in slide decks, not in operating reviews.

This is upside down.

Stage matters

The right metric stack depends on what stage the business is in.

Pre-PMF. Acquisition cost is irrelevant if the product doesn't retain. Track activation rate, time-to-first-value, qualitative-interview volume, and the shape of the retention curve. CAC is a distraction here.

Scaling. Now CAC matters — but blended CAC hides the truth. Track CAC by channel, marginal CAC at the next dollar, contribution margin by cohort, and payback period as a hard ceiling. Set a payback rule and don't break it.

Mature. The scoreboard metrics start lying louder. Last-click attribution flatters paid channels. ROAS overstates incremental return. The stack moves toward incrementality lift, MMM-allocated contribution, brand-search rate as a leading indicator, and CLV cohort curves you can defend to a CFO.

The right metric isn't the loudest one. It's the one that closes a loop.

The metric most teams underweight

Incrementality.

ROAS lies. Last-click lies. Multi-touch attribution lies more politely but still lies. The only honest answer to is this spend working is a properly-designed incrementality test — geo-holdout, conversion-lift study, switchback test, ghost-bid experiment, whatever fits the situation.

Most teams don't run them because they're slow, expensive, and politically dangerous. The first incrementality result usually embarrasses someone. That's the point.

Run them anyway. The cost of pretending your attribution numbers are real is always higher than the cost of running the test.

The actual takeaway

The right metric isn't the loudest one. It's the one that closes a loop.

Pick the question first. Are we acquiring profitably? — CAC, payback. Is the channel actually adding incremental customers? — geo-holdout. Is the creative working harder than the audience? — input metrics.

The metric follows the question. The question doesn't follow the metric. Build the stack from there.