At some point in your career, someone senior will ask: “What’s our most important marketing metric?” And you’ll feel the pressure to give a single, confident answer — CAC, ROAS, pipeline, MQLs, whatever sounds most sophisticated in the room.
Here’s the uncomfortable truth: there is a right answer. But it isn’t the same answer for every company. The metric that should be running your marketing decisions depends almost entirely on the stage your business is at right now — and getting this wrong is one of the most expensive, least-visible mistakes a marketing team can make.
Not because tracking the wrong metric produces bad reports. But because the metric you obsess over becomes the metric you optimise for. And optimising for the wrong thing at the wrong stage can look like growth — until the quarter it doesn’t.
Why ROI and ROAS Are Not the Answer (Most of the Time)
ROAS and ROI are not bad metrics. They’re just metrics that answer a very specific question: “Did this spend generate more revenue than it cost?” That question is critical — but only once you’ve already solved the more fundamental problems underneath it.
Think about what ROAS can’t tell you. It can’t tell you whether the revenue it’s measuring is coming from the right customers — the ones who stay, expand, and refer. It can’t tell you whether your acquisition engine is building a brand or just renting visibility. And critically, it can’t tell you whether the funnel it’s measuring has a leak somewhere between the click and the close that’s making even a strong ROAS number misleading.
The point isn’t that these metrics are wrong. It’s that every metric has a shadow — the thing it doesn’t measure — and at certain stages of a business, that shadow is more important than the number in the light.
The Stage-by-Stage Breakdown
Here’s the framework. Three stages, one primary metric each — not because the others don’t matter, but because one metric should be the north star that everything else is subordinate to. When in doubt, this is the question that gets asked first.
At seed stage, you don’t have enough data to optimise. You don’t have enough volume to make CAC or ROAS statistically meaningful. What you do have are early users — and the single most important thing you can learn from them is whether the people who try your product actually experience value from it. That’s activation: the moment a user first gets what you promised them. If activation is low, no amount of marketing spend fixes what is fundamentally a product-message mismatch. You’re pouring water into a bucket with a hole in it.
In their early days, Airbnb’s growth team identified that guests who completed at least one successful booking were dramatically more likely to book again — and that hosts who received their first booking within 21 days of listing were far more likely to remain active on the platform. Both were activation milestones, not acquisition metrics.
Rather than optimising for traffic or sign-ups, their early marketing and growth work focused on improving the rate at which new hosts and guests hit those first-experience moments. They famously went to hosts’ homes to take professional photographs — a manual, unscalable intervention — purely to improve the activation rate of new host listings. The metric driving that decision wasn’t ROAS. It was % of new hosts receiving a first booking within 21 days. The photography programme was a marketing decision masquerading as an operations one.
At growth stage, you know your product works. Activation is healthy, you understand your ICP, and you’re ready to pour fuel on the engine. This is where most teams default to CAC as their north star — but CAC alone is incomplete. A $200 CAC is great if your customer pays back that cost in month two. It’s a business-threatening number if payback takes 18 months and you’re burning cash to scale. Payback period — the time it takes to recover customer acquisition cost through gross margin — is the metric that tells you whether your growth is actually sustainable or just impressive on paper.
HubSpot’s early growth stage is a masterclass in engineering short payback periods through channel design, not just acquisition efficiency. Their inbound marketing model — creating educational content that attracted buyers already in-market — wasn’t primarily a cost-cutting exercise. It was a payback-period engineering exercise.
Customers who found HubSpot through their own content were already partially sold by the time they signed up. They activated faster, expanded to higher tiers sooner, and churned less. The result was a payback period that was materially shorter than the industry average for comparable SaaS products of the era — not because HubSpot spent less to acquire customers, but because the time-to-value after acquisition was compressed by the quality of the inbound lead. Their content investment paid back not through organic traffic metrics, but through its effect on payback period across the funnel.
At scale, the acquisition engine is running. Payback periods are understood. The new question is whether the revenue you’ve already acquired is growing — or slowly leaking. Net Revenue Retention (NRR) measures how much revenue you retain and expand from your existing customer base, after accounting for churn and contraction. An NRR above 100% means your existing customers are growing faster than others are churning — which means your business can grow even if you acquire zero new customers. Below 100%, you’re on a treadmill: running hard on acquisition just to stay flat.
Slack’s legendary growth story is often told as an acquisition story — the viral product, the word-of-mouth, the rapid spread through tech companies. But what actually made Slack a $27 billion acquisition target was its NRR, which was consistently reported to be well above 130% during its growth and scale years.
The mechanism was elegant: teams started on a free tier, hit a threshold of value, and upgraded. Then the product spread within the organisation, floor by floor and department by department, without Slack needing to re-acquire the same company. Their marketing at scale wasn’t primarily about new logos — it was about expansion within existing accounts, driven by a bottoms-up adoption model that made their NRR self-reinforcing. When Salesforce acquired them, they weren’t buying a customer base. They were buying a retention engine.
The Framework, Consolidated
| Stage | North Star Metric | Core Question It Answers | Stop Over-Indexing On |
|---|---|---|---|
| Seed / Early | Activation Rate | Are the people who try us experiencing value? | Traffic, MQL volume, ROAS |
| Growth | Payback Period | Is our acquisition sustainable enough to scale? | CAC in isolation, total MQL count |
| Scale | Net Revenue Retention | Is our existing base growing or slowly leaking? | New logo volume, total pipeline created |
One More Thing: Know When You’re Between Stages
The most dangerous position is the transition. A company moving from growth to scale that still runs on payback period as its north star will under-invest in customer success, over-invest in acquisition, and not notice the NRR problem until it’s significant. The metric shift has to happen slightly before the stage shift — not after.
A practical signal: if more than 30% of your revenue is now coming from expansions and upsells within existing accounts, you’re probably already at scale stage whether you feel like it or not. Start measuring NRR like it’s your most important number — because it probably is.
Not sure which stage you’re at — or which metric to run on?
Book a free 15-minute call. Tell me where your business is, what you’re currently tracking, and I’ll give you a clear read on whether your north star metric is aligned to the stage you’re actually at — and what to do if it isn’t.
Where to Go Next
Knowing your north star metric is the first decision. The second — equally important and equally mishandled — is how you present it to the people who control your budget. A growth-stage company with a clear payback period story still loses budget conversations when it can’t communicate why that metric matters and what trade-offs it implies.
That’s exactly what the article below covers.