- il y a 5 semaines
Reverse Engineering Product-Market Fit: A Roadmap for Founders
Catégorie
🤖
TechnologieTranscription
00:00Hello again, VivaTech, and thank you for joining us.
00:03I'm Chris O'Brien, your interviewer, moderator for this session,
00:07the founder and editor of the French Tech Journal.
00:10And we have a special treat today, because with us all the way from San Francisco,
00:15we have Rahul Vorha, CEO and founder of Superhuman.
00:20So welcome, and thank you for coming all this way.
00:22Absolutely, thank you for having me.
00:24I'm super excited, because this is a company I've followed as a tech journalist
00:27for a long time, from a distance.
00:30So it's great to have a chance to have this conversation with you.
00:33Let's just start with the basics for anyone who doesn't know.
00:37What is Superhuman?
00:39Superhuman is the most productive email experience ever made.
00:43Imagine getting through your inbox twice as fast as before,
00:47replying to important messages one to two days sooner,
00:51and saving four hours or more every single week.
00:55We're also reinventing the future of productivity with AI.
01:00Imagine waking up to an inbox where every email already had a draft reply.
01:05You would simply edit and then send.
01:08And sometimes you wouldn't even have to edit.
01:11Yeah, that is the dream.
01:13So that's great.
01:14So there's a number of things, even just what you said there,
01:18we could talk about in the next 25 minutes.
01:21But we're here to talk about product market fit,
01:24which is really something that you have become a notable thinker and leader on over the years.
01:30So let's just start with that.
01:32What does product market fit feel like?
01:34Well, product market fit is the number one reason why startups succeed.
01:40And the lack of product market fit is the number one reason why startups fail.
01:45What does it feel like?
01:48Paul Graham would say you have product market fit when your users,
01:53or rather when you've made something simply that people want.
01:58Sam Altman would say that you have product market fit
02:01when your users spontaneously tell other people to use your product.
02:06But it is perhaps Marc Andreessen who has the most vivid definition of product market fit.
02:12He would say you can almost always feel it when product market fit is not happening.
02:20Customers aren't quite getting value.
02:22users aren't growing that fast.
02:24Word of mouth isn't spreading.
02:25The press reviews are kind of blah.
02:27And the sales cycle takes too damn long.
02:31But he also says you can almost always feel it when product market fit is happening.
02:39Customers are buying as fast as you can add servers.
02:41You're hiring sales and support as fast as you can.
02:45Reporters are constantly calling you about your hot new thing.
02:48Money is piling up in your checking account.
02:51And investors are hanging out outside of your house.
02:56And that's the most vivid definition I found.
02:58And by the way, a definition I was staring at through tears in the summer of 2017
03:04because it seems so subjective, so inactionable.
03:09What do you do when, by that definition, you don't have product market fit?
03:14So I began to wonder, can you measure product market fit?
03:19Because if you could, then maybe you could systematically, even numerically, optimize it.
03:27Yeah, I mean, that's my thought is that you have this new generation of entrepreneurs,
03:32especially in the Gen AI age.
03:34And product market fit is now certainly a centerpiece of the lingo when you're anywhere you go,
03:40that you're talking about it.
03:40But it is also important and yet can be vague.
03:44What do we really mean when we say something like this?
03:47And then you can have the written definition to say, okay, here's what it means.
03:51Here's how people define it.
03:53You guys, you personally went a step further.
03:56You actually have a quantifiable model called PMF that other entrepreneurs can sort of take
04:06and use as a template.
04:08And one of the things, interesting, I wrote this down to make sure I get it right.
04:12One of the central questions is, how would you feel if you could no longer use this product?
04:18So of all the aspects, when you think about that user interaction,
04:23what made that the most powerful question?
04:26The journey to this question started with an experience I had in 2015.
04:34In 2015, we started the company in 2015 and we started much like any other company by writing code.
04:41And in the summer of 2016, we were still writing code.
04:45And in the summer of 2017, we were still writing code.
04:50And I felt this intense, incredible pressure to launch.
04:56Both from the team, but also from deep down within myself.
05:00After all, my last company had launched, scaled, and been acquired in less time.
05:07And here we were two years in, and we still had not launched.
05:11But no matter how intensely I felt that pressure, I knew that a launch would go really badly.
05:17I did not believe that we had product market fit.
05:20But I couldn't just say that to the team.
05:24These are super ambitious, hyper-intelligent engineers.
05:27They poured their hearts and souls into the product.
05:31So I needed a plan.
05:32I started my search for the holy grail.
05:35I spoke to everybody I could find, read everything written on the topic, searched high and low.
05:41And then I met this guy called Sean Ellis.
05:46Now, Sean ran early growth at companies like Dropbox, LogMeIn, and Eventbrite.
05:51He even coined the term growth hacker.
05:54And Sean found a benchmark that is predictive of success.
06:01One that is even more predictive than net promoter score.
06:05Simply ask your users that question.
06:07How would you feel if you could no longer use the product?
06:11Give them three possible answers.
06:13Very disappointed, somewhat disappointed, and not disappointed.
06:18And so very disappointed means they'd be very disappointed without the product.
06:21They love it.
06:23And measure the percentage that answer very disappointed.
06:26What Sean found is that the companies that struggled to grow,
06:31they almost always had less than 40% very disappointed.
06:36Whereas the companies that grew the fastest,
06:39well, those companies almost always had more than 40% very disappointed.
06:45So this metric is more objective than a feeling.
06:49It predicts success better than net promoter score.
06:53And with this metric, you can actually construct what we did,
06:56what we call a product market fit engine.
06:59And that engine can generate a roadmap.
07:02A roadmap that is essentially guaranteed to take you from where you are today
07:07to product market fit.
07:09Okay, so let's break down the PMF engine a bit.
07:12I know it can be complicated, but there are four primary questions.
07:16So what are those?
07:17Can you kind of walk us through the framework?
07:20Now you're testing me.
07:21Okay, so the four questions.
07:22Email these four questions to every user.
07:25Number one, like we just said,
07:28how would you feel if you could no longer use the product?
07:30Three possible answers.
07:31Very disappointed, somewhat disappointed, not disappointed.
07:34Number two, who do you think this product is best for?
07:40This is a very powerful question because happy users will almost always describe themselves,
07:46but using the words that matter most to them.
07:50So this is the perfect source for product marketing copy
07:54that will speak directly to your happiest users.
07:58Number three, what is the main benefit of the products for you?
08:02One of the most important questions.
08:03I'm sure we'll circle back to that.
08:05And number four, how can we improve the product for you?
08:09So very simple, four questions.
08:11Now, timing is key, and also you shouldn't repeat the survey.
08:15In terms of timing, I would send it as soon as the user has had the chance
08:20to experience the core value proposition of whatever it is that you do.
08:24So for example, let's say that it's Uber or Lyft.
08:28Just after you've experienced your first ride and that frictionless payment,
08:33that would be the perfect time to survey the user.
08:36If it was Airbnb, just after your first vacation stay.
08:40At Superhuman, we wait until you've sent a minimum number of emails,
08:44and after about two or three weeks, that's when we survey the users.
08:49So that leads to my next question.
08:51So in my career as a journalist, and the turmoil in the media industry,
08:55I've been part of a number of projects over the decades
08:57where we've gone out and surveyed customers, we've hired consultants,
09:02we've tried to figure out, gather all this data about changing behaviors,
09:06media consumption, et cetera.
09:08So there's the part of that process that's the framing the questions,
09:14gathering the data.
09:16And then at some point you sit down and you have this stuff in front of you,
09:20and then there's a whole other element of, well, what do you do with that when you have it?
09:25And how do you interpret that?
09:28And how do you get the right insights and move forward?
09:31So in your case, one of the elements, one of the bold decisions you made was,
09:37we're going to ignore the feedback or the people who said they weren't disappointed.
09:43So how did you get there?
09:45How did that seem like the right call in terms of how to use that data?
09:51I'm going to neatly sidestep the philosophical discussion
09:55that naturally comes up here about causality and correlation.
09:58But that point aside, let's say that product market fit
10:02is when you have more than 40% of folks very disappointed without your product.
10:07So we want to increase the size of that pie,
10:10the number of people who'd be very disappointed without your product.
10:14And there are two other segments, somewhat disappointed and not disappointed.
10:19In order to increase the size of that pie,
10:25we need to understand two things.
10:27Number one, why do people love your product?
10:30And number two, what holds people back from falling in love with your product?
10:35And that's where we go to the other questions and we start analyzing the answers.
10:41So to figure out why people love your product,
10:44we go back to question number three,
10:46which is what is the main benefit of the product for you?
10:49Take all the survey results,
10:52filter it down to just those survey results
10:55from the people who were very disappointed.
10:57Remember, these are the people who love it.
10:59And look at their answers only to question number three.
11:05This is where I like to take those answers
11:07and put them onto a gigantic word cloud.
11:09So for superhuman, the answers are going to be things like,
11:12the product's really fast.
11:14I end up saving so much time.
11:15I love the keyboard shortcuts.
11:17I really like the design or the aesthetics.
11:20It makes me so much more efficient and so on and so on.
11:23And so you'll probably have hundreds of those responses.
11:25Stick them on a word cloud.
11:27This is why the people who really love your product love your product.
11:32So now we understand why people love the product.
11:34Then we need to figure out what holds people back
11:37from falling in love with the product.
11:39First of all, you have the set of people
11:42who were not disappointed if the product were to go away.
11:46Now, this is to the point of your question.
11:49There is almost no point acting on their feedback
11:54because even if you built all the things
11:56that they asked for,
11:58they didn't even like your products
12:00for the same reason why the people
12:02who fall in love with it do.
12:04So chances are you'll spend all of this time
12:06building all of that stuff
12:07and it won't make a difference.
12:09So as hard as this is to do
12:10and as weird as it is to have a founder
12:12sitting on stage saying this,
12:14just ignore their feedback.
12:17Similarly, for the crowd
12:18that was somewhat disappointed without your product,
12:22I cannot stress this enough.
12:24do not simply act on their feedback
12:27because again, if you just implement
12:29all the stuff that they ask for,
12:31there'll be a large contingent of those
12:34who still will not fall in love with your product.
12:37The question then becomes,
12:39how do you figure out who to listen to?
12:41And this is the magic part.
12:43Take the somewhat disappointed users.
12:46We're going to segment them into two parts.
12:50The first part will be the set of people
12:53for whom the main benefit was the same
12:55as the very disappointed users.
12:57The second part will be those
12:59for whom the main benefit was different
13:00to the very disappointed users.
13:03That second group,
13:05again, we're going to politely disregard
13:07all of their feedback
13:09because they don't love the product
13:11for the right reasons.
13:13They don't love the product
13:14in the same way
13:16that the people who really love your product do.
13:18so just ignore all of their feedback.
13:21Then there's the specific segment
13:23of the somewhat disappointed users
13:25who love the product for the right reasons
13:27but it's something small
13:28and I would wager something very small
13:30that's holding them back.
13:32Implement all of those things.
13:34And if you do the math,
13:35you can write this down on a piece of paper.
13:37It's pretty simple.
13:38You'll see that you're incrementally increasing
13:41the number of people
13:42who would be very disappointed
13:44without your product.
13:45And if I may,
13:46just to ask a follow-up on that,
13:49that's a very precise,
13:53refined equation,
13:54formula, framework for getting there.
13:56Was this a process of trial and error?
13:58Were there sort of references
13:59where you could say,
14:00oh, aha, this will guide us
14:02toward making those,
14:03slicing up those different tranches of users,
14:06which one to favor, which one not?
14:08Or did it come from as you worked through it,
14:10you realized, oh, this is a wrong path,
14:13we have to go back and focus on these people?
14:15Well, I blame the precision aspects of it
14:18in my background as a trained computer scientist.
14:21It's just my engineering brain being applied to it.
14:24But I was really trying to solve the problem of
14:28you are going to get so many different users
14:30and so many different customers
14:32coming into your product.
14:33How do you figure out who to listen to?
14:36If you just act on everybody's feedback,
14:38you're going to end up with a muddled, confusing,
14:42incoherent product
14:43and a product that sadly won't have products market fit.
14:47We all know that you have to focus.
14:49We all know that you have to pick a direction.
14:52The question is,
14:53how do you pick that direction?
14:55And this is a way that marries, in my opinion,
14:58vision with data
14:59and has very successfully in our case
15:02and with dozens of other companies
15:04I've helped do this
15:05take you to product market fit.
15:08You mentioned a second ago
15:10this notion or the concept of
15:11something's holding people back.
15:13There's that layer of people.
15:14So you get your super fans,
15:16you figure out who's really passionate,
15:17but there might be some blocks to them
15:19really, really adopting,
15:22really feeling that next level of passion.
15:24So when you're looking at the various things
15:26that might be holding them back
15:28like lack of integrations or mobile,
15:30how do you start to address that
15:32without then veering off that sort of core focus?
15:36Great question.
15:37So let's go back, for example,
15:39in our case to 2017's, a long time ago.
15:43The obvious thing at the top of that list
15:44was the lack of a mobile app.
15:46And when you have obvious things like that,
15:48you should just build them.
15:49So of course, we, a long time ago, built them.
15:52But then the list became less obvious
15:55and more interesting.
15:57Things like we need more integrations
15:58or better search or attachments handling
16:01or calendaring or all of the things
16:03you might imagine people wanting
16:04in an email application.
16:07And if I understand the question correctly,
16:10it's how do you prevent yourself
16:12just kind of veering off in this direction?
16:15So here's how we do this.
16:16In every planning cycle,
16:18we do this quarterly,
16:20but as a smaller company,
16:21you might do it more frequently.
16:22We aim to spend 50% of our time
16:27doubling down on the things
16:29that people really love about superhumans.
16:32So that's stuff like the productivity,
16:34the efficiency, the shortcuts,
16:35the speed, the design.
16:37And the other half of our time,
16:41systematically addressing that list of objections.
16:44Remember, these objections came from
16:46from the somewhat disappointed users
16:48for whom the main benefit resonated.
16:52Now, you might be wondering,
16:53why is this 50-50 split important?
16:57Well, imagine you spent all of your time
17:00simply doubling down
17:02on the things that people love about your product.
17:05After all, that's what made you successful.
17:07Why not do more of that?
17:09Well, then the answer,
17:10and hopefully this will feel obvious
17:12now that I've laid it out,
17:14is that by doing that,
17:16you won't increase the size of the pie
17:19that is very disappointed.
17:20You won't get more people
17:22to fall in love with your product.
17:24The only way you'll do that
17:25is by converting some
17:26of the somewhat disappointed people.
17:29Okay, that's why you should do
17:30at least some of that.
17:32But why not spend all of your time
17:33running down that list?
17:35The mobile app, the search,
17:37the calendar, and so on.
17:38Well, if you spend all of your time doing that,
17:42chances are somebody will come along
17:45and do your special source
17:48better than you even do it.
17:49Someone will do the speed
17:51and the shortcuts and the design
17:52and all of that good stuff
17:53better, faster, harder than you.
17:55So you don't want that to happen either.
17:58Which is why we go into every planning cycle
18:00with a roughly equal balance
18:02between doubling down on what people like
18:05and systematically addressing objections
18:07from the user base.
18:09So you get through this process,
18:12at least the initial turn of it.
18:14You've got what you feel very confident about
18:17is real product market fit
18:19validated by this framework.
18:21As the founder and the CEO then,
18:24how do you begin to actually apply that
18:28in the way you're managing, hiring,
18:30setting OKRs,
18:32all the things that then you actually have to do
18:34to start to build the business?
18:37In the very early days of the company,
18:39as we were working on this metric,
18:42we had one core OKR,
18:44one core goal for the company,
18:46which was to get to 40% plus.
18:50Initially, when I came up with this framework
18:53and I was kind of developing it
18:55at the same time as applying it to Superhuman,
18:58the very first result for that number
19:01was, I think, 22%.
19:04You may notice that 22% is not 40%,
19:07it's not even close.
19:09But it was close enough that I thought
19:11by systematically applying this
19:14and by working on the framework,
19:16perhaps over the course of a year,
19:18we could get that number to where it needed to be.
19:21So in quarter number one,
19:22it was 22%.
19:24There's a resegmentation trick that you can do
19:28that we didn't get into,
19:29but if you're interested,
19:31you can search for my name,
19:33Superhuman Products Market Fit Engine,
19:35and I go into it in detail online.
19:37That got that number from 22% to 32%.
19:40And then we just systematically ran this process
19:43every single quarter.
19:44We got from 22% to 32% to 48% to 58%.
19:50So after a year of running this process,
19:53we got to the point where 58% of our users
19:56would have been very disappointed without Superhuman.
20:01Now, the thing about metrics like this
20:04and any business metric
20:06is that they all come down over time.
20:10Your churn rates will go up,
20:11your marketing channels will become less efficient,
20:14and yes, even your product market fit score,
20:16if left to its own devices,
20:18is going to fall over time.
20:20Now, why does this happen?
20:21Because as you're growing,
20:23you're going to be encountering new personas,
20:26new user types,
20:28new people coming in via new channels,
20:31people who are maybe lower intent
20:32than those early adopters
20:34who sort you out more proactively.
20:37And so this number is going to come down.
20:40That's natural, it's normal.
20:41That means that the work is never done.
20:43And so even for the last seven, eight years
20:47that we've been running the company in this fashion,
20:49we still keep on working on this number
20:51and we still keep on driving it.
20:54One of the reasons I think
20:55the timing of this conversation is interesting
20:57is we're in the middle of this hype cycle
20:59around Gen AI startups.
21:02And again, extraordinary for me,
21:0430 years following this stuff,
21:06seeing these stories about two guys sit down,
21:09they code something, they release it,
21:11we have 10 million ARR,
21:12we don't even need venture capital, we're fine.
21:15And as a journalist, of course, I'm skeptical.
21:17I look at that and I say,
21:18okay, well, are they diluting themselves?
21:21How real is that growth?
21:22We're all trying this stuff.
21:24How much are people churning?
21:25So has the current dynamic lowered the bar for PMF?
21:30Has it kind of caused you to rethink
21:32some of the assumptions that you make?
21:34Or is it, well, you know,
21:37are some of these people potentially diluting themselves
21:39just because they see that revenue coming in
21:41thinking, oh, we've got PMF?
21:45I think that there's two fundamental things
21:49that, to your point, AI has really changed.
21:52Number one, you can, with a much smaller team,
21:56much faster, build an application
21:59or a user experience that has a chance
22:02of hitting products market fit.
22:04And number two, you can do so with far fewer people.
22:09So what I'm seeing from my vantage point,
22:11also as an angel investor,
22:13I've invested in over 120 companies at this point,
22:16is that it is easier than ever
22:19to get to products market fit.
22:21But I don't think that's widely known yet.
22:23So we're in the early days of this phenomenon.
22:26Not yet everyone knows how easy it is
22:28compared to how difficult it used to be.
22:31And to put some numbers on that,
22:33it used to be the case in Silicon Valley.
22:36I normally live near San Francisco.
22:39that good for a pre-seed or a seed stage startup
22:43is a company getting from zero, let's say,
22:46to a million dollars of annual recurring revenue
22:48in about a year.
22:50That used to be good.
22:52In a year, it's now normal to see companies
22:56go from zero to four, five, six, seven,
22:59all the way up to $10 million
23:01of annual recurring revenue in one year.
23:05Let's give you another example.
23:08The average company in Y Combinator today,
23:11the average is growing at 10% week over week.
23:15If you go back two years ago,
23:17it was only the top quartile of companies
23:20that was growing at 10% week over week.
23:23So hopefully you can see just how much
23:25AI has changed the game.
23:27It's changed demand.
23:28It's changed how easy it is to build.
23:30It's reduced the number of people you need.
23:32In some of the obvious ways,
23:34like you need fewer customer support people,
23:36but also in less obvious ways as well.
23:38The very best founders are busy automating
23:41every single aspect of their business.
23:44So it's an entirely new game,
23:47and I'm learning as much as everyone else is.
23:50Okay.
23:50I'm going to try to squeeze in one more question here.
23:52So you mentioned at the beginning
23:54there was a journey at Superhuman
23:55to create this framework.
23:57Now you have it.
23:58If you could go back to day one very quickly,
24:01obviously you'd go faster,
24:02but how would you proceed differently
24:05if you had this PMF framework,
24:07the PMF engine in your hand day one?
24:10Well, I think I would apply it from day one
24:13as opposed to just kind of doing stuff for two years.
24:17Although I suppose the phrase du jour
24:19is you can just do stuff.
24:21Doing stuff with a plan
24:22is better than just doing stuff.
24:24Great.
24:25Well, we'll stop there.
24:26Perfect.
24:27Thank you all for joining us,
24:28and please give a warm round of applause for all
24:30for coming all the way to join us.
24:32Thank you.
Commentaires