How to measure Product-Market-Fit
By Peter Clark
Having being the CEO of startups and also the Product Manager creating new products within existing companies, I've spent quite a lot of time ™ stressing about Product Market Fit metrics.
What is Product Market Fit?
As a quick reminder, Product Market Fit (PMF) is the binary event that dictates if a product resonates with users to the point it'll be a successful product. You can read more about it at either YC or a16z.
Many people recommend measuring and tracking PMF via surveys, specifically:
how would you feel if you could no longer use the product?
Surveys and NPS
This question is essentially an equivalent of the "Net Promoter Score" — wonderful tools like Satismeter exist to help founders measure both PMF and NPS. (I was an incredibly early user of Satismeter, and I recommend all startups use it!)
After you've answered the survey (either the PMF "how disappointed…" or the NPS "how likely to recommend to friends…") you're asked to give feedback, and that feedback is incredibly useful.
I believe this feedback is foundational to building a healthy product organization, you can see how critical I believe this feedback is in a blog post I wrote in 2015.
I believe product teams should report on this metric monthly — tracking how it adjusts over time — however I no longer believe it should be the primary way that startups measure PMF.
The problem with these kinds of surveys is that they are qualitative. When users use a product, they're not only influenced by the problem you're solving but a vast array of other factors that influence how they feel about a product:
Are they friends with the founders?
Are they paying?
Is the brand cool and hip?
Is the UI fancy?
Are there trendy features like AI within it?
Imagine if Brad Pitt invited you to his random SaaS product. Now imagine if some loser like myself invited you to his random SaaS product. Both are identical, which product will get better survey results?
Here is the primary way I believe PMF should be measured:
Do users keep coming back?
You should create a cohort view of user logins and users taking a core product action and plot that over time, for example:
3 months after activation, how many users were logging in at least once a week? how many were taking the core product action of <creating a document>?
Products that do not yet have PMF have terrifyingly poor retention — similarly, products that do have PMF have shockingly good retention.
Put it this way, unless you are in a very specific industry (finance, APIs, etc) you do not and cannot have PMF without users logging in at least once a week. Broadly speaking, a product that users love to use once a month just isn't viable.
Users simply do not have the brainpower to remember to use products on such an infrequent basis. Of course, there are exceptions to the rule (such as expense tracking, to pick a random example) but those are far less product centric and far more sales/workflow centric.
The great thing about this metric is that it cannot lie. There's no way for a user to mislead or be polite to avoid hurting your feelings, because it's tracked purely quantitatively.
Never ever underestimate the power of retention
I have found that founders and product managers greatly underestimate the value of retention. They'll see stagnant growth and think "we need more top of the funnel" — I know this for a fact, since I receive emails from YC founders every month asking for help with Product/Growth and they always think their problem is acquisition, not retention.
I have also had the (unfortunate?) pleasure of being tasked with fixing retention for companies with high churn — my thinking on this has also evolved, and I now believe that in a huge number of cases you cannot drastically fix retention if your initial cohorts are terrible. It's one thing for users to activate, stick around for a few months, and then abandon — there's a bunch you can try to do there. But if a user activates and bounces in the first month, your problem isn't retention it's PMF.