Putting it together
North stars
One number the whole team can point at, chosen carefully enough that improving it genuinely means the business improved.
A north star is not your most important metric. It is the metric that, if it improves, means almost everything else improved too — and which is hard to move without actually doing the work. Choosing it badly is worse than not having one, because a team will optimise whatever you put on the wall.
The metric hierarchy #
Confusion here causes a specific and common failure: reporting a north star to a team that cannot influence it, and a vanity metric to a board that needs to make a decision.
| Level | What it is | Who watches it | Cadence |
|---|---|---|---|
| North star | The single measure of whether marketing is working | Board, executive team | Monthly, quarterly |
| Guardrails | Constraints that stop the north star being gamed | Executive, marketing lead | Monthly |
| Input KPIs | The things you can actually change that move the north star | Marketing team | Weekly |
| Diagnostics | Everything else — CTR, bounce, CPM, position | Whoever is optimising | Daily, as needed |
Putting diagnostics on a board slide. A board cannot act on click-through rate, and presenting it invites questions that waste everyone's time. If a number would not change a decision the audience is empowered to make, it does not belong in front of them.
Leading versus lagging #
A lagging indicator tells you what happened. A leading indicator tells you what is about to happen. You need both, and you need to know which is which.
Lagging — closed revenue, CAC payback, retention
Trustworthy and final. Also too late to act on. By the time closed revenue tells you the quarter went badly, the quarter is over.
Leading — qualified pipeline, trial starts, branded search volume
Actionable while there is still time. Also noisier, and gameable if you are not careful — which is exactly what guardrails are for.
The practical arrangement: your north star is usually lagging, because it needs to be trustworthy. Your input KPIs are leading, because the team needs something to steer by this week. The relationship between the two should be explicit and, ideally, tested — if your leading indicator has never predicted your lagging one, it is not a leading indicator, it is just a number.
Candidate north stars by business model #
| Business model | Reasonable north star | Why it works | Essential guardrails |
|---|---|---|---|
| Ecommerce | MER — total revenue ÷ total marketing spend | Uses only numbers you can defend: real revenue, real spend. Immune to attribution disputes and platform self-reporting entirely. | Contribution margin, new-customer share, return rate |
| B2B / considered sale | Qualified pipeline created, or closed-won revenue by first-touch source | Connects marketing to the number sales is measured on. Survives long cycles better than lead counts. | Pipeline-to-close rate, average deal size, sales cycle length |
| SaaS | New ARR from marketing-sourced accounts, or CAC payback period | Payback in particular ties acquisition to unit economics, which is what actually constrains growth. | Net revenue retention, churn, gross margin |
| Lead generation / services | Cost per qualified lead, with qualification defined by sales | The word doing the work is qualified. Cost per raw lead is trivially gamed by lowering quality. | Lead-to-sale rate, revenue per lead, sales capacity |
| Local / appointment-based | Booked appointments attributed to source | The closest countable event to revenue, and it happens fast enough to steer by. | Show-up rate, average job value |
| Content / media | Returning engaged readers | Rewards depth over reach, and it is very hard to buy. | Total reach, subscription conversion |
Why MER deserves special mention #
Marketing efficiency ratio — total revenue divided by total marketing spend — is the most under-used metric in performance marketing, and its virtue is what it refuses to do.
It does not care about attribution. It does not consult Meta's opinion of its own contribution. It does not double-count. It uses two numbers you can defend in any room: what you actually earned, and what you actually spent.
When platform ROAS says 4.2× and MER says 1.8×, the gap is the double-counting, the modelled conversions and the generous windows. MER is the number that survives contact with your accountant. Use platform ROAS to decide which campaign to scale; use MER to decide whether the whole programme is working.
The obvious limitation: MER cannot tell you which channel is working, so it cannot allocate budget on its own. That is what incrementality testing and MMM are for. MER tells you whether the machine is profitable; those tell you which parts to feed.
Guardrails: what stops a north star going wrong #
Every metric can be improved in a way that damages the business. Guardrails are the constraints that make gaming it visible.
| North star | How a team games it without meaning to | Guardrail that catches it |
|---|---|---|
| Leads generated | Lower the barrier — remove qualifying questions, run cheap broad campaigns | Lead-to-opportunity rate, revenue per lead |
| ROAS | Cut all prospecting, spend only on branded search and retargeting | New-customer share, total revenue growth |
| Traffic | Chase high-volume irrelevant keywords and cheap traffic | Conversion rate, engaged sessions |
| MER | Cut spend entirely — MER improves as marketing shrinks | Absolute revenue growth, market share |
| CAC | Acquire only the cheapest, lowest-value customers | LTV:CAC ratio, average order value |
| Pipeline created | Inflate stage definitions, mark unqualified deals as opportunities | Pipeline-to-close rate, stage ageing |
Nearly every guardrail is a quality measure protecting a quantity target, or an absolute measure protecting a ratio. If your north star is a ratio, you need an absolute guardrail. If it is a count, you need a quality guardrail. A north star without a paired guardrail will eventually be optimised into something nobody wanted.
Choosing yours #
- Start from how the business makes money, not from what your tools can measure. The convenient metric is rarely the right one.
- Test it against the gaming question: “how could a clever, well-intentioned team improve this number without helping the business?” If you find an easy answer, you need a guardrail — or a different metric.
- Check you can actually measure it from a source you would defend. If it depends on platform-reported conversions, it will not survive its first serious challenge.
- Pair it with two or three guardrails. Not ten. Guardrails that nobody reads are decoration.
- Define the input KPIs that move it, and give those to the team. The north star is for direction; inputs are for work.
- Write down the definition — exact formula, exact source system, exact date treatment — and put it where the report lives. Most metric arguments are definition arguments that nobody realised were definition arguments.
- Review it annually, not quarterly. A north star that changes often is not a north star.
The scored self-audit asks whether you have a north star, whether it is guarded, and whether anyone acts on it — alongside thirty-odd other questions about collection, tagging, consent and revenue linkage. It recommends a starting north star for your business model and orders the rest into a list of what to fix first.
North star: revenue from new customers, sourced from the CRM and reconciled to finance. Guardrails: MER, and lead-to-sale rate. Inputs: qualified leads by channel, cost per qualified lead, and conversion rate at each funnel step. That structure fits more businesses than any clever alternative, and every number in it is defensible.
Common questions #
Can we have more than one north star?
Then you do not have one. The value of a north star is that it settles arguments — when two courses of action conflict, you check which improves the number. Two north stars means the argument continues, just with more slides. If you genuinely have two irreducible goals, you have two businesses or a sequencing decision you have not made yet.
What if we cannot measure revenue attribution reliably?
Use MER, which needs no attribution at all — only total revenue and total spend. It is available to every business from day one, it is impossible to argue with, and it is a far better starting point than an attributed number nobody trusts. Add channel-level detail later, once collection is solid.
Should the north star be the same for marketing and sales?
Ideally they share a lagging outcome — revenue — with different leading indicators underneath. When marketing is measured on leads and sales on closed deals, the two teams are structurally set against each other, and every quarter contains an argument about lead quality. Measuring marketing on qualified pipeline or closed-won by source removes most of that conflict at the definition level, which is the only place it can actually be removed.