The seven marketing metrics that actually matter in B2B
Most marketing reports in small companies contain twenty numbers, none of which changes a decision. Seven are enough — provided you also state which question each one answers and when it lies.
In short
- Seven metrics cover the steering: enquiries, enquiry source, close rate per stage, cost per enquiry, cost per customer, time to close, and customer value over the relationship.
- Every number needs three things attached: the question it answers, the collection rhythm, and the condition under which it misleads.
- Three widely used metrics can be dropped by small companies with a clear conscience.
- The most common misreading: monthly figures with long sales cycles. What was spent in March closes in September.
There are two kinds of metric: those that can change a decision, and those you merely report. The second group is the larger one — and it costs time every month.
The test is simple: if this number doubled or halved, what would you do differently? If the answer is hard to give, it does not belong in the report.
The seven
1. Qualified enquiries per month
Answers: is enough coming in at the top?
Rhythm: monthly, viewed across three months.
Misleads when: "qualified" is undefined. Without a fixed condition — "fits the audience and has a stated need", say — the number measures the weather.
2. Enquiries by source
Answers: where is further work worthwhile?
Rhythm: monthly.
Misleads when: only the last touch is counted. Someone who arrives via a written piece, searches three weeks later and then calls shows up as "direct" — and the piece gets no credit.
3. Close rate per stage
Answers: where exactly does it break down?
Rhythm: quarterly.
Misleads when: only the overall rate is looked at. "12 per cent" does not say whether the problem is at first contact or at the quote — and those are completely different jobs.
4. Cost per enquiry
Answers: what does the inflow cost?
Rhythm: quarterly.
Misleads when: your own working time is left out. A channel that appears to cost nothing but eats six hours a week is the most expensive one you have.
5. Cost per customer won
Answers: does this add up?
Rhythm: half-yearly.
Misleads when: the delay is ignored. With a six-month sales cycle, first-quarter costs belong to third-quarter closes. Calculated monthly, this number is nonsense.
6. Time to close
Answers: how long does a case tie up resources?
Rhythm: half-yearly.
Misleads when: the mean is used. A single large deal running fourteen months shifts it substantially. The median is the honest figure here.
7. Customer value over the relationship
Answers: how much may a customer cost?
Rhythm: yearly.
Misleads when: calculated too early. Without several years of history this is an assumption, not a metric — and it is almost always set too optimistically.
Three metrics you can drop
Email open rate. Since mailbox providers began pre-loading images, it measures technology as much as interest. The click rate on a specific link is considerably more informative.
Social network reach. It fluctuates with decisions you do not influence. If that channel matters, what counts is how many people come from there to your page — not how many were shown something.
Total page views. Without a breakdown by source and page, this is a number that rises when a piece happens to get shared. It only becomes useful per page and per source.
Worth knowing
With sales cycles over three months, monthly reporting is misleading for every outcome metric — not imprecise, but systematically wrong. The costs fall in one period and the close in another.
In practice this regularly produces an expensive mistake: a channel gets cut after two weak months, even though the cases from that period cannot possibly have closed yet. With long cycles, report inflow figures monthly and outcome figures quarterly — rolling over twelve months.
The monthly report on one page
A report longer than one page does not get read. Three blocks are enough.
| Block | Content | Size |
|---|---|---|
| Inflow | enquiries in total, by source, change against last quarter | 3 numbers |
| Outcome | closes, close rate per stage, median duration | 3 numbers |
| Decision | what we will do about it, what we will stop | 3 sentences |
The third block is the most important and is missing from most reports. Without it, a report is documentation rather than steering.
A recurring pattern: a channel delivers few enquiries but the ones with the highest close rate. Viewed on inflow alone it looks weak and gets cut — three quarters later the good customers are missing, and nobody connects that to the decision.
That is why every source belongs in the report with two numbers: how many enquiries it delivers and what becomes of them. Judging a source by volume alone is the most common expensive mistake in B2B marketing.
What honestly cannot be measured
Part of the effect cannot be attributed, and saying so openly is better than an invented attribution.
- Referrals. Someone arriving through a conversation may have read one of your pieces a year ago. That chain cannot be reconstructed.
- Mentions in AI answers. There is no console for it. Only referral traffic is measurable, and that is the tip.
- The effect of doing nothing. Nobody knows what would have happened had you not worked the channel.
The only sound proxy is the question in the first conversation: "How did you come across us?" It is imprecise, but it is closer to the truth than any automatic last-touch attribution.
Help me cut my marketing report down to what matters. Our situation: - Industry and audience: [details] - Typical time from first contact to close: [details] - Channels we work: [list] - Numbers we report monthly today: [list] Tasks: 1. Assign each of our current numbers to one of three groups: changes a decision / documentation only / misleading at our cycle length. Justify each assignment in one sentence. 2. Propose which numbers we should collect instead, and for each state: the question it answers, the rhythm, and the condition under which it misleads. 3. Design a one-page report with three blocks: inflow, outcome, decision. 4. Tell me which effects we honestly cannot measure with our setup. Do not invent industry benchmarks.
In closing
More numbers do not lead to better decisions, only to longer meetings. Seven metrics cover the steering, provided each one states which question it answers and when it lies.
The most important part of the report is not the numbers anyway, but the three sentences underneath: what will be done about it and what will be stopped. A report without those three sentences can be deleted outright.
Common questions
Which marketing metrics really matter in B2B?
Seven: qualified enquiries per month, enquiries by source, close rate per stage, cost per enquiry, cost per customer won, median time to close, and customer value over the relationship. Everything else is either derived from these or changes no decision.
How often should marketing metrics be collected?
Inflow figures such as enquiries monthly, close rates and cost per enquiry quarterly, cost per customer and cycle length half-yearly, customer value annually. With sales cycles over three months, monthly outcome figures are systematically misleading, because costs and closes fall in different periods.
Which metrics can be dropped?
Email open rate, because mailboxes pre-load images and it therefore measures technology rather than interest; social network reach, because it depends on decisions you do not control; and total page views, as long as they are not broken down by page and source.
Why is the median better than the average?
For time to close, a single very long case shifts the average substantially, whereas the median shows the typical case. For planning and capacity questions, the typical case is the more useful figure.
How do you attribute enquiries to the right source?
Automatic attribution usually captures only the last touch and credits long-acting channels such as written pieces or referrals to "direct". The most honest proxy remains asking in the first conversation how someone came across you — imprecise, but closer to the truth.
Marketing that sets itself up
The Studio Engine beta is live. Claim your spot and help shape it from the start.
Join the beta →