Defining Product-Market Fit

Product-Market Fit is one of those terms that feels intuitive but rarely is. It’s a critically important milestone along the journey to startup success, so we wanted to demystify how we think about it at Connect, and share our best advice for how you can find it fast.

Product-Market Fit (PMF) is achieved when your product satisfies the needs of your target customers. It’s when you offer a better solution than they can get elsewhere.

When you have PMF, growth is cheaper: users refer their friends, and paying customers keep paying. When you don’t have PMF it’s like force-feeding people something that they don’t need, which makes growth more expensive.

This is what makes it such an important milestone in a startup’s journey. The actions you take before and after PMF are likely very different.

Before you have PMF, your whole focus should be on understanding your target customers better, and iterating your product quickly to give them what they need.

After you have some form of PMF, you can focus on acquisition and other growth activities. The below graph comes from Jeff Chang, (former growth engineer at Pinterest).

💡 PMF is neither static nor binary

There’s no magical threshold that you cross when you “have PMF”. You might have very strong PMF with one group of customers, and weaker PMF with another.

It can also change over time. Customer expectations shift as new products hit the market, so you need to work to keep PMF. However in this post, we’ll focus on achieving PMF for the first time.

“Product market fit isn’t a one-time, discrete point in time that announces itself with trumpet fanfares. Competitors arrive, markets segment and evolve, and stuff happens.”
– Marc Andreessen

📖 Further reading

• David Rusenko (founder of Weebly) gave a great talk as part of YC’s “Startup School” on how to find PMF. It covers a lot of what we go through in this doc, but with some extra real-life examples.
• Marc Andreessen on why “market” is the most important part of the PMF trifecta.
• Marc Andreessen on some PMF myths.

You can measure PMF by looking for signals that your customers like what you’ve built. There’s no single right way to measure it. It’s fairly subjective, so you need to work out what’s right for your company and your audience.

At Connect, we believe some measures are more accurate and therefore useful than others. Below are a few different ways to measure it:

1. Retention

Retention is generally the clearest indicator that people like your product.

They’ve used it, seen what it can do and can’t do, and want to keep using it.

Exact retention metrics will differ by company. For B2B SaaS, having customers that keep paying is a great signal. For social B2C, engagement retention might be more useful. There are grey areas so it’s best to figure out what works for your business and product.

Tim Chong, co-founder of Yonder shares how he thinks about retention in the context of the “core user action”.

He asks himself:

1. What problem am I solving for users?
2. How do I measure that my product is solving that problem?
3. This forms my “Core User Action”

Tim says: “Driving ‘dirty’ App opens where someone opens and closes the App isn’t retention, and you get into the world of growth hacking/push spamming.

I would say that the biggest measure is retention/repeat behaviour of your core user action – with the usual flattening retention curve etc.”

Below are some benchmarks from Lenny’s Newsletter for both user retention and net revenue retention. These were collated by asking industry experts what they see as “good” and “great” retention benchmarks.

Definitions:

• User Retention = “using the product, making a purchase, posting a photo”, still active after 6 months.
• Net Revenue Retention = MRR in month X last year / MRR in month X this year from those same customers
• These are broad definitions, as they take in very different types of company. They’re not intended to be very scientific, but more to give general targets.

Ultimately, you’ll need to assess the best retention metrics for your business. Léa Samrani, Product Lead at Uptime, says “It’s essential for companies to understand what people use their product for and build their own custom metric that can tangibly measure both customer behaviour and predict future business success“.

However you do it, make sure to measure by cohort (i.e. splitting users by month acquired) rather than blended. Blended retention mixes together older users with those that are brand new, who haven’t had a chance to churn yet.

Basically what you’re looking for is a retention curve that flattens out. You’ll churn customers that don’t have great fit, and the ones that remain do have it. The graph below from Brian Balfour shows this nicely.

Léa points out that there are some products where retention isn’t a great metric – most notably those where churn is a positive outcome. For example, in dating or healthcare the user might find a long-term partner or get their condition under control, and no longer need the app.

2. Organic referrals

If your customers are bringing in their friends, it’s a good sign of PMF.

Whenever you recommend a product to your friends or colleagues, there’s a bit of social risk. If the product is bad, your friends think a bit less of you. If it’s good, they think more of you.

So if your customers are referring their friends, it’s a sign that the product is so good that they’re willing to take that risk.

A great way to measure your organic referrals is your virality co-efficient.  It’s a measure of how many customers you actually acquire, with each new customer. For example, if every user brings in one of their friends, that’s much more valuable than a user that brings in no friends.

To measure the viral co-efficient you take the amount of users, multiplied by amount of invites sent per user and multiply it by the conversion rate.

Users X invites per user X conversion rate / 100 = virality co-efficient

You can also simply take a poll of users at signup to ask how they heard about you and measure how many were recommended by friends. Measuring direct traffic is also a good indicator of people coming because they hear about you from a friend, especially early on before brand marketing kicks in.

Merci Victoria Grace, former Head of Growth at Slack, cites 50% of growth coming from direct traffic and referrals as a good benchmark.

3. User surveys

Asking customers questions through a short survey can be a useful way to gauge how well you’re meeting their needs. This method is both easy to measure, and easy to benchmark.

A common survey to send out is “Net Promoter Score” (NPS).

You measure NPS by asking a user how likely they are to recommend your product to a friend, out of 10. 9-10 are classed a “promoters”, 7-8 are “neutral” and 0-6 are “detractors”. Then:

(% of promoters) – (% of detractors) = NPS score

I don’t love this method, because it’s affected by external context, like how often people talk about the problem you solve, and how social your customers are. But it’s generally recognised and well regarded in the industry, so is a broadly safe bet.

Another approach is the “disappointed” test.

Dropbox and Superhuman asked users “how would you feel if you could no longer use our product?” and measured the percent who answer “very disappointed”. Sean Ellis, early growth lead at Dropbox, showed that anything above 40% was a strong indicator that a company was ready to grow.

Matt Gallivan, user research lead at Slack and ex-AirBnB, suggested two alternative questions:

• Compared to other ways you do X (X being the function/problem you solve), using this product to do X is generally: much better, somewhat better, neither better nor worse, somewhat worse, much worse?
• The last time you did X, what product or tool did you use to do it, and why? [Leave this one open ended]

Both these approaches can see how well you’re solving your customers’ needs.

Whatever approach you take, make sure to keep your surveys short – ideally just one question to ensure maximum responses. Adding extra questions will help you learn more, but drop-offs will mean you may get bias in the survey.

The downside to survey responses is that they’re not as strong a signal as real, observable behaviour. People generally want to be nice so might inflate their scores a bit.  They’re also open to gaming (i.e. only send out to ‘good’ customers).

💡 Top tip: Survey frequency.

There are a few different ways to survey your customers. You can either do it after a certain amount of time/purchases/visits (e.g. after 2nd purchase), or periodically to all customers (e.g. September aggregate). The former can give you quicker feedback loops, the latter gives you a wider picture.

If you send surveys out without a deliberate frequency, you may lose quality of insight about your PMF.

Sara Chokr says that you can get extra insight by segmenting the survey results to find common themes. The neutral group are typically where you can have the biggest impact in driving PMF – what are those users’ issues and how can you convert them into huge promoters of your product?

The article below is from email client Superhuman who dug into how they could improve the product for users that were on the fence.

💡 There is no one-type-fits-all measurement for product-market fit.

Stitch are a B2B2C medtech company, which means that their product has a slow feedback loop. CEO Jonathan Moshinsky says:

“For us, PMF is the ability to get repeat sales of the same feature set. If the customer expands, with no new features required. If we get two customers doing that, it’s enough to show that something is working.”

“Another great indicator is the sales cycle time reducing. If our 10th sale takes half the time, it’s because people are wanting it faster. Due to our slow product feedback loop, this works better.”

📚 Further reading:

• This blog post from Lenny’s Newsletter gives a fantastic overview of what is good and great churn, split by company types. It’s highly recommended reading.
• This follow up to the above post from Casey Winters (former growth lead at Pinterest) digs in with much more detail.
• Casey Winters also gives a great talk entitled “Do you have PMF? It’s all about retention”.
• A guest post from Matt Gallivan on Lenny’s Newsletter about how to interview customers to find product-market fit.
• Andrew Chen has a short-but-useful post on the viral coefficient.
• Superhuman have a great blog post about how they “built an engine to find Product Market Fit”.
• A great talk from Josh Elman (investor and ex-PM at Twitter and Facebook) about “the only product metric that matters”.
• Measuring “core actions” by Sarah Tavel, (investor and ex-PM at Pinterest).
• Reforge article about the important of measuring retention.
• Amplitude blog post about how to interpret your retention data.

This really is the billion-dollar question and unfortunately there aren’t any cheat codes. However, here are some behaviours that can help you to find PMF faster.

1. Identify your target customers 🎯

Simply put, you can’t serve a set of customers if you don’t know who they are. This is a non-negotiable first step to finding PMF. You can use market segmentation or personas to create a view of who these people are.

Tim Chong says that for a consumer product, your ideal user should be “laughably narrow”. TAM doesn’t matter too much at this stage, more having something concrete.

At Yonder, their “laughably narrow” customer was Stephanie who worked as an Account Director at a marketing agency in London. She’s 28 and moved to London from South Africa for career opportunities and to meet new people.

The best route to finding your target segment can vary, but a good rule of thumb is to find people with the strongest combination of “feel the pain you can solve” and “are willing to pay for it to be solved”.

For your first set of customers, it’s very helpful if they’re generally early adopters. These folks are more willing to try new things, and ideally become design partners.

Katerina Pascoulis, Stripe PM and former founder, says that you have to make sure that your user personas are accurate. Do users exist who fit those abstractions? As you have more conversations with real people and your learning evolves, so should your personas.

2. Talk to your customers ☕️

Getting insights from customers (and non-customers) is a generally under-utilised weapon tactic.

Léa Samrani says “In the most cases there is an emotional or social reason motivating people in using the product that has nothing to do with the product features. Understanding that is key to unlocking PMF”.

Katerina shares that sometimes founders and others are nervous about talking to customers in case they tell you something you don’t want to hear. But it’s better to fail fast and learn your product isn’t right early as then you can work to change it.

Founders sometimes use the Henry Ford “people would have asked for a faster horse” parable to excuse not talking to customers. Or they believe that user interviews are a specific skill set that they don’t have, so prefer to do nothing.

Talking to customers is always, always preferable to not talking to them. You don’t have to build what they ask for, but ignoring them completely is inexcusable.

There are multiple ways to talk to your customers. Some examples:

• Staff the customer support lines once a month
• Find non-customers as you would for regular acquisition, and interview them instead of selling
• Message your customers, and ask them for time

This is not a one-off activity – you should build it into your regular routine.

Tim Chong shares some hacks for how to find people before you have users yourselves:

• Facebook groups where your group of customers hang out
• Friends (+ friends of friends)
• Community groups
• Random people on the street

He also suggests finishing user interviews by asking: “Is there anyone you know that I should speak to?”

And when you conduct an interview, the following tips are useful to get the most out if it:

• Let them speak – just listen and guide with prompts. Ask open ended questions to get them talking. The most useful things they’ll say are the bits they don’t want to say. Every time you ask a question, they reset to the ‘easy’ answer. Let them keep talking. If you want to change direction, you can say “you mentioned XYZ, can you tell me more about that?”
• Record them (with permission) or have a note-taker present. Be fully engaged in the conversation. They’ll see you taking notes and guess what you want to hear.
• Don’t talk about your product or company. At least, at first. Understand their problems, what they care about, what they don’t. It’s much more interesting for building insight.

Sara Chokr shares some example questions to ask::

1. Tell me a little bit about yourself/the problem.
2. What is the main challenge you are facing at the moment in relation to [problem]?
3. How have you tried to solve this before?
4. What has prevented you solving [problem] in the past?
5. If you had a magic wand, how would you ideal solve this?
6. Would you be willing to have another call in a few weeks to test our solution?

📚 Further reading:

• Shreyas Doshi wrote a great Twitter thread about asking your customers to stack-rank their problems, to avoid building something they don’t need. Long but gold.
• The Mom Test by Rob Fitzpatrick focuses doing research for a new company idea, but it equally applies to any ongoing product research. A must-read.
• A good blog post on this topic from the Product Lessons newsletter.

3. Prioritise ruthlessly 🏆

Clear prioritisation is one of your most powerful levers to hitting PMF quickly.

Work out your core set of hypotheses that you want to prove and focus on them. Anything that’s not helping you learn faster isn’t helping you at all.

This is where balancing product quality becomes subjective and unique to each company. While you’re trying to learn how to solve your user’s problems, any attention that you give to making the product gorgeous and shiny may not move you in the right direction.

Léa says that a great way to prioritise is to work out what you definitely won’t be working on.

4. User testing 👀

Showing customers work in progress is a great way to learn about them and to ship product fast. User testing is when you ask people that look like your target market to try out your solutions in front of you, so you can see how they interact with them and how they work (or don’t!). You can user test when you’re building a feature or iterating, or as early as wireframe stage.

User testing at this point is useful for two reasons. Firstly, it helps you validate quicker and refine your designs, so whatever you do build has a better chance of being valuable to your users.

Secondly, it can act as a user research session. When you watch users interact with your product with you asking open-ended questions – get them speaking in a stream of consciousness. This is a great way to learn about your customers, in the context of your product.

User testing can enable you to short-circuit the general plan-design-build-learn flow.

Léa Samrani suggestions making experimentation part of your process early on. This creates a culture with a growth mindset, which will be able to pivot when needed and react faster – making it more likely to find PMF faster.

5. Ship early and often 🛥️

Think of your product team as an engine that creates value for customers.

Each time your ship something new into the hands of your customers, you can learn more about how to best serve them. The premise of a minimum viable product (MVP) is to ship the smallest feature that will help you learn.

Pre-PMF, shipping is less about features and more about validating your product hypothesis.

Katerina says that’s it’s all too easy for startups to feel they need to build something fully featured before launching it. But that runs the risk of building something that customers don’t really want or need.

Instead, ship as soon as you’re ready to learn. Customers are surprisingly forgiving of rough edges. As Reid Hoffman said: “If you’re not embarrassed by the first version of your product, you’ve launched too late”.

Your feedback mechanisms need to be in place however – a way to measure what people use, and a means for talking to customers to get their feedback.

Other than that, it’s all down to speed. It’s just a game of how fast you can go through this process (provided you’re shipping a good enough product to learn from).

6. Develop a set of leading metrics 💯

Ultimately you want to measure PMF, but doing so takes time. It takes six months to discover if users are retained after six months, and even user surveys are lagging indicators.

A useful mitigation here is to give yourself leading metrics that can help you understand the likelihood of a good outcome. Depending on your product, these could include things like:

• Onboarding conversion rates.
• Using the product at least 3 times.
• Having an order arrive within one day.
• Connecting with five friends.
• Getting value from the product.

Katerina shares her technique here: “I see these as activation metrics – it’s basically what has to happen for your customer to see the value of your product, and that period of time should be as short as possible.

These will all help you to understand ahead of time how likely you are to give customers a great experience, which then influences PMF. These are your levers to improving PMF.”

Make sure it’s actually something you want to move – Léa writes:

“Define your north start first which should be a customer behaviour metric that is a leading indicator to future success, then work backward to define what are your leading metrics to reach it.”

📚 Further reading:

• https://miro.com/miroverse/product-metrics-analytics-board/?social=copy-link

7. Don’t waste time ⏱️

If you don’t have it yet, finding PMF should be your #1 goal. It’s a vital milestone that reduces the cost of growth, and is a fantastic proof point for founders and investors. Focus is everything at this point.

A useful way to make sure that you’re shipping fast is to ignore things that don’t matter. These are things that won’t matter if you don’t last long enough to find PMF and include:

• Bugs that don’t count impact your ability to reach PMF.
• Scalability (i.e. don’t slow down shipping and learning just because something may not scale).
• Features that aren’t core to your PMF.

Katerina suggests that sometimes you shouldn’t even worry about building the product. Just ship a concierge product that delivers value with a human-as-a-backend. That will help you learn faster.

She adds that build vs buy is a useful consideration: “For example, Stripe have built a bunch of products to help founders get to market quicker. Outside of payments there are products like checkout and the customer portal where you can set up a pricing page and highly conversion optimised checkout page without code. You could build this yourself but why would you if it’s not core to your offering, especially if you’re pre-PMF.”

📚 Further reading:

• Retool’s lessons from getting product-market fit
• Paul Graham things that don’t scale
• Mom Test Concierge