Incubator KPIs Worth Reporting, and Three That Mislead
Startups supported, jobs created and funding raised are the numbers everyone reports. Here is what each one hides, and the measures that survive scrutiny.
Most incubator metrics are chosen because they are available
Not because they mean something. That is the whole problem in one sentence.
"Startups supported" is easy to count, so it gets reported. "Did the programme change the outcome for those startups" is hard to establish, so it does not. Over time an entire sector converges on a set of numbers that are easy to produce and nearly impossible to interpret.
You will still have to report the easy ones - funders ask for them. But it is worth knowing what each one hides, and worth collecting a small number that actually tell you whether the programme works.
Three that mislead, and what to report alongside them
1. Startups supported
Counts anyone who touched the programme. It says nothing about depth. An incubator that ran a one-day workshop for sixty companies and one that put twelve through a year of residency both report a number, and the first number is bigger.
Report alongside: startups supported by engagement level. Residency, programme cohort, and light-touch, as three separate numbers. It is a more honest picture and it makes your intensive work visible instead of averaging it away.
2. Funding raised by the portfolio
The most-quoted incubator metric and the least informative. Two problems.
It is dominated by outliers. One company raising a large round makes the total, and the total then says more about that company than about your programme. And it confuses selection with support: an incubator that admits companies already likely to raise reports a big number without having caused any of it.
Report alongside: the median raise, and the count of companies that raised anything at all. A programme where nine of twenty raised something modest is usually healthier than one where one of twenty raised a lot, and the total shows the opposite.
3. Jobs created
Almost never comparable between incubators, because the definitions differ silently. Full-time equivalents or headcount? Do founders count? Interns? Cumulative since inception, or current? At what date?
Report alongside: the definition, every time, in the same line as the number. This is not pedantry - if you change methodology between years, an unlabelled figure makes your own trend meaningless.
The measures that survive scrutiny
Survival at 24 months post-graduation
The best single number, and the least popular, because it is slow and unflattering.
Measure it against the cohort that entered, not the cohort that finished. Measuring survivors of survivors is how a 40% survival rate becomes an 80% one without anything changing.
It is hard to game, which is the point. It also takes two years to produce your first figure, so start now rather than when someone asks.
Progression through your own stage definitions
Define three or four stages - idea, prototype, first revenue, scaling - and record which stage each company is at when it enters and where it is at each review.
This is the closest an incubator gets to measuring its own contribution, because it is about movement rather than absolute state. It is also the measure most likely to change how you run the programme, because it shows you where companies get stuck. If everyone stalls between prototype and first revenue, that is a curriculum finding, not a reporting line.
Time from application to decision
An operational measure that predicts a surprising amount.
Programmes that take eleven weeks to decide lose their best applicants to programmes that take three. You are not competing on decision speed exactly, but the strongest applicants have options and they act on them. It is also entirely within your control, which is more than can be said for most metrics.
Mentor engagement, measured in sessions not sign-ups
"Forty mentors on the panel" is a recruitment number. The useful version is how many mentors actually held a session in the last quarter, and how many companies got at least one.
Most incubators discover, on first measuring this, that a small number of mentors do nearly all the sessions. That is worth knowing before the reliable ones burn out.
Space utilisation, if you have space
Allocated versus occupied, not allocated versus total. A centre that is 95% allocated and 60% occupied has a reclamation problem, not a capacity problem, and only one of those needs a building.
Collect on a schedule, not when the report is due
The reason incubator data is bad is almost never that founders refuse. It is that they are asked ad hoc, months after the fact, and reconstruct from memory.
Fix the mechanics and the data improves without anyone trying harder:
- Ask on a fixed schedule the startups know at intake - quarterly works for most measures.
- Ask for few things. A five-field quarterly return gets answered; a twenty-field one gets guessed at.
- Define each field once, at intake, and do not change it mid-year.
- Track who has answered. Most of the collection problem is chasing, and chasing is only possible if you can see the gap.
That last point is where spreadsheets fail. A returned form goes into a folder; the absence of a form is invisible unless something is holding the expected list against the received list.
A workable reporting set
If you are building a reporting pack from scratch, this covers most funder requirements and stays honest:
Activity - startups supported, split by engagement level; applications received and admitted; time from application to decision.
Outcomes - survival at 12 and 24 months post-graduation against the entering cohort; stage progression; companies that raised anything, with median and total; jobs, with the definition attached.
Operations - space allocated versus occupied; mentor sessions held and companies reached; funding disbursed against committed.
Nine or ten numbers. Fewer than most incubators report, and more than most can currently defend.
The uncomfortable part
Some of these will make your programme look worse than the numbers you report today. Survival against the entering cohort is lower than survival against graduates. Median raise is far less impressive than total raised.
Report them anyway, and report them consistently. An incubator that has tracked the same honest measures for four years can show a trend, and a trend is far more persuasive to a funder than a large number with no baseline.
Frequently asked questions
- What is the single most useful incubator metric?
- Survival at 24 months after graduation, measured against the cohort that entered rather than the cohort that finished. It is unflattering, slow and hard to game, which is exactly why it is worth having. Most reported metrics are chosen because they are available, not because they mean anything.
- Why is 'funding raised by portfolio' a weak measure?
- Because it is dominated by one or two outliers and by selection rather than support. An incubator that admits companies which were already going to raise reports a large number without having caused any of it. Report it if a funder asks, alongside a median and the count of companies that raised anything.
- How should we count jobs created?
- Carefully, and with the definition attached. Say whether it is full-time equivalents or headcount, whether founders count, whether it is cumulative or current, and the date. Most job figures across incubators are not comparable because the definitions differ silently.
- How often should incubator KPIs be collected?
- On a schedule the startups know in advance, quarterly for most measures. Asking ad hoc when a report is due produces the lowest response rate and the least reliable data, because founders are reconstructing rather than reporting.
