How to run effective PMF sprints
Part 1 of the PMF Sprint Playbook
If you're reading this then you are probably still chasing product-market-fit or you are exploring new ideas. You’ve got some sort of belief or hypothesis about the problem you’re trying to solve and you’ve understood that the best way to move forward is to actually simply talk to people about it.
I’ll break down what we found to work best when trying to figure out whether a specific problem is worth trying to solve. There’s a lot to talk about so I broke this down into multiple parts. Today, in part 1, I’ll discuss all of the preliminary work and give you a basic structure so you can maximise learnings.
Why a PMF Sprint?
A PMF sprint is a a 2/3 week, interview-heavy block of time solely for testing a problem-market hypothesis.
Like most things in life, you’ll get better results if you exclusively focus on a specific problem. That’s the idea behind the ‘sprint’. Dedicate at least 2 weeks of your and possibly another founder’s time on this one specific thing. Don’t get distracted. Stop everything else. Commit everything to this and treat it like a sprint, not the usual product building marathon.
The Secret Sauce
The first step is to have a concrete idea of what it is you want to validate.
How do you form a solid hypothesis of a problem to solve?
Todd Jackson from First Round Capital breaks this down into 3 approaches
Market first: Start with a market or space that interests you, then look for a specific problem.
Experience ripe for improvement: Look for areas where you believe there should be a better consumer experience than what currently exists, and iterate from there.
Problem first: Start with a problem you’ve experienced firsthand and figure out if enough other people have the same problem.
There are loads of examples of successful founders that have done this. I won’t go into detail here. Quickly you’ll start to realise that your ability to talk to people and extract the right information is the most important skill you’ll need to develop at this stage.
4 Ps in a Pod
To begin with you should already have thought about or heard of a specific pain point and thus identified the first P: the Problem.
Next, before we can actually go out and start talking to people we need to narrow down who most commonly faces this problem: the Persona.
You don’t need to do these in this specific order. You might have started with a market-first approach in which case you’ll have figured out your Persona before the Problem. Once you combine those two you’ll have some sort of idea of what it is you can solve. That’s the Promise.
Finally to deliver on this Promise you’ll end up with the Product. Notice that what you are building is the last step and the culmination of everything else in your strategy. The product comes last. It merely crystallises the problem, persona and promise you’ve already nailed, which is the core idea behind sustainable PMF growth.
A solid hypothesis is very simply a document. Something that articulates an offering based on those principles.
Don’t spend a huge amount of time on this. At most half a day thinking deeply about it and writing the document and getting feedback from peers.
I sometimes like to add a fifth P to frame it considering existing products you might be competing with: Positioning. For example:
Unlike [competitor product], our offering [differentiator]
Script
Now that you have a better idea of who to target and why, you can start coming up with the right questions. Here we’re starting to enter the realm of how to develop the skill of talking to potential customers.
Rob Fitzpatrick wrote essential reading on this: The Mom Test.
After reading this book I made the classic mistake of trying to structure the perfect interview script with lots of great unbiased questions I could ask. As a software engineer by trade I have a habit of going deep with structured thinking. This was mostly a waste of time. You’ll never get through those questions like you would in an interview since you want to keep things as casual as possible to get people to open up and stay engaged in the conversation. I still got value from doing this however since I ended up diving really deep into the persona and hypothesis, but it’s definitely not necessary.
The key is to boil it down to 3 questions that will make or break your hypothesis. These should be elephant in the room, ‘scary’ questions that will help you focus in on the critical information you want to get out of these conversations once you’re in the thick of it. Having a limited number also makes these easy to remember and simple to casually introduce into the conversation.
AI will do a pretty good job of helping you here. You can paste your hypothesis doc and any additional context with the following example prompt:
Based on the following hypothesis what would be your top 3 questions to other potential [persona] to validate whether they would pay for solving [problem]. These should be “elephant in the room”, scary questions that will make or break the hypothesis. Avoid obvious opinionated questions. Use “The Mom Test” book as a reference for good questions to ask.
Notes
The final piece of preparation you’ll want to do before you start jumping on to calls or talking to people is what you’ll do after these calls. The idea behind all of these initial steps is to lay solid ground work beforehand so once the calls start coming in you can solely focus on those and getting more in the pipeline since that’s the bulk of the work.
The best argument for taking notes is so that you can share the learnings with your team. It’s the best way to get everyone aligned and on the same page. The other argument is that, in this day and age, it’s dead simple to transcribe a call, put it into an LLM and get it to come up with a great summary takeaway. Even if you’re having a casual conversation you can record yourself talking through the key points after the discussion and get a similar outcome.
Luckily for us we also have that built-in to our product. You can start a new note, transcribe and still write down any critical information you want to during the call.
The only thing you’ll actually need to prepare here is a solid prompt for getting a consistent output. What we found works best is to start with direct quotes and then have sections based on factual information surrounding your hypothesis. In Superthread you can create templates that will automatically format your notes after your call in this way and this saves me a huge amount of time but any other tool will work as long as you can pull the transcript of the call and paste in your prompt into your favourite LLM model of the day.
If you’re a solo founder without a team or if you’re super early in your journey you probably wouldn’t need to actually do this. As long as you trust yourself and your brain not to forget the actual learnings or care about leaving a paper trail and having something to come back to.
Next steps
That’s your prep completed. You can get all of this done in under a day’s work and you’ll have a really solid foundation for what comes next: scheduling and talking to people. We’ll cover those in the next post. Stay tuned!
Go draft your one-page hypothesis doc, block 2 weeks in your calendar and reply or comment ‘PMF’ to this if you want to receive my templates.