Airtable’s path to product-market fit — co-founder Andrew Ofstad on building horizontal products

 Yeah. I'm excited in today's show to go really deep into the early days of Airtable, specifically on product design and go to market. You know, there's so much that gets glossed over. When you, when you hear about founding stories and the founders I'm working with today, really want to hear the tactics, how to think about their first MVP, how to figure out their ideal customer profile, how to identify the right growth model.

And I think Airtable is just such an interesting example to learn from because it's just so difficult to build, a big, horizontal, powerful product. And you've obviously had a lot of success with it over the last 10 years. So to rewind back to the beginning, Andrew I've, I've heard you tell the story of how Airtable got started about how you met your co-founders in college.

Uh, but for those who aren't familiar, could you give a quick recap.

 Sure. Yeah. So kind of going way back. We, uh, we met at duke, myself and Howie and Emmett, we were all kind of the tech nerds there. And this is kind of back in the days. I think we probably met in 2005 when the established path was very much, you know, banking consulting or med school 

And we were sort of the weirdos that like to talk about programming and video games and this place called Silicon valley that like, wasn't sort of, you know, on the radar of most duke students and, you know, kind of like started reading hacker news when that first came out and we'd get lunch and jamming ideas and just kind of were kind of good friends 

And so, you know, after college, we kind of went our separate ways. Uh, I got this job at Accenture. They had like this R and D department. They called tech labs and. So my job there was kind of building demos and prototypes of, you know, iPhone apps. And, uh, actually, like, I think that might have been even been before the app store was out, but just kind of like demos of cool things for our customers.

And I actually, how he was a year below me at duke and I got him his first job at Accenture, I, you know, put in a good word. And my manager was like, all right, this guy looks solid. Let's hire him. And the first day he actually didn't show up to work. Like I just had this empty desk next to me and I'm like, oh my God, where's where's Howie.

And he called me like kind of frantic. And he is like, Hey, you know, I decided like last minute I really wanna do this startup. And so I'm not coming to work. So kind of, uh, burn some bridges there, but, you know, uh, that was the first attempt to work together. Um, and he actually tried to convince me to start work them on that company, but I just started this job and, and so didn't make the leap. And, uh, later on, Howie's company was bought by Salesforce. I was the AP at Google and, and the sort of PM occasion program there. I, worked on Android for a year and maps for a few years. And eventually we were kind of both into place where we were at Salesforce and Google respectively as product managers, you know, decided we wanted to kind of take the leap together and start working on air table and, the rest is sort of history from there.

But, uh, you know, it took us another year to convince Emmett who was our third friend from duke, And he was the best programmer we knew. So we eventually convinced him and he joined us about a year later, but that, that's the short, short backstory on how we got started.

 cool. So, Andrew, can you talk about where you got the idea for air table? Like, what was the early vision that

 yeah, the early vision was very much, you know, to try to democratize software creation. And for us, we had like this crazy superpower of being able to build software that we had sort of just really, uh, you know, loved and, and kind of had given us all these advantages in our careers.

 And you could just have this tremendous influence through software. You can, you know, uh, build something and deploy it to people. And it's very cheap to do. And within an organization, you can have a lot of influence, even if you're not in a leadership position.

So for us, that was kind of a lot of the broad vision was how do we make. Software creation easier for non-programmers. And a lot of the inspiration too, was kind of thinking over the course of computing history. And there's very much the early days of computing, which was all about command lines and doing these sort of arcane commands.

And that's how you operated a computer. And it was only through the sort of hard work of like the early, uh, you know, Xerox park and the gooey and apple Mac to kind of take that computing and make it more accessible to everybody. So very much we're inspired by that. And like, how do we take this thing, uh, of, of software creation and programming and really make that more accessible to a much larger audience.

 so early on is very much like an intellectual pursuit of like, you know, wanting to, to learn more about the history of, of software and, uh, how the Goey was invented and so on and so forth. And we all had had a passion for that and wanted to do the same thing for, for, um, software to kinda give everybody a so software stack, they can build useful software on top of.

And so, you know, there's a lot of founders I'd say just getting started with, with their companies today and attempting to validate the idea they're they're pursuing. Did you three do anything in the early days of air table to kind of help prove to yourself that it was a good idea?

 We, we were a bit trained in that way. I would say, I think at the time is very much about the lean startup and getting early customer validation and failing fast. And a lot of companies were pivoting every three months and, you know, kind of like there was this book making the circles called, like the seven steps to the epiphany that very much like espouse this kind of lean company development model where you just, you know, put out a super, uh, rough prototype to get to know the customer and kind of, uh, quickly pivot.

If it it's not looking like it's gonna stick. So we kind of took more of a first principles approach and, uh, didn't really wanna just kind of throw idea hit at the wall with the company. And like I said before, we were like really interested in this general space of software creation. And how do you take a database and programming and make it way easier and more accessible.

And I think just intellectually we're super, uh, interested and, and excited about, about kind of, you know, going down that path. And so I think, we spent a lot of time early on just doing research and almost like on a sabbatical, like reading all this prior art of, of old, uh, computing pioneers, like the Douglas Ingle Barts and the bill Atkinsons, uh, who kind of, you know, helped invent the, the gooey, uh, reading lots of essays on that playing with old products, like HyperCard, that kind of had flavors of software creation for everybody.

And, uh, you know, even kind of like more contemporaries, like Brett Victor kind of read a lot about how to visualize complex systems. So for us, it was a passion of like learning and, and sort of, uh, understanding the prior art. And I think simultaneously, you know, how he was at Salesforce, where, uh, his company was acquired and really like, can learn more about the enterprise software market and saw that most business applications out there are essentially just kind of databases and views and business logic on top.

And there's a lot of reinventing the wheel and, and sort of vertical software. So really we sort of like saw the business potential there where, you know, Salesforce was this giant platform where it's more or less just a flexible database, but you have to sort of be a advanced admin to UN understand how to configure it and to build it.

So we saw the potential of opening up that type of software creation to a much larger market and really kinda spent a lot of time thinking about like, okay, what is the market for this? Let's think about spreadsheets. they've been around forever, but they are like the Kell app, but most of the time, the way people use them is to kind of track objects they're tracking people, companies, simple tables, and they're not doing modeling and number crunching, which is where they're invented for.

So, you know, all is to say like, I think the point is the, we had like this long, early gestation period where we kind of did our research both on the, technological challenge, the sort of market and kind of like the, the precedents be before us. And I think really sort of saw that we could build a big company here and, uh, that could kind of put us in the right mindset to say like, Hey, this might take a long time to do.

Let's make sure we're dedicated to it. Um, and sort of informed a lot of the ways we kind of built the company in the early days.

 it. So like, almost from like, from the very beginning, you were committed to this idea of software creation for everybody. You, you had done the research, you knew it was a really big market and you were, you were just sort of in it for the long haul from the beginning.

 Yeah, definitely. And I, I think it, it, would've been hard to get that level of conviction without having that period of just kind of like going deep on the problem without having like formally broken ground on kind of building the product and trying to like get customers and, and so on and so forth. So we, we kind of, uh, took our time early on to make sure that it's something we were committed to.

And I think that was super helpful. And obviously it helped that we were friends, But I think, uh, if you're gonna sort of do, like we did and kind of tackle a horizontal product that is pretty hard to kind of market and get started and, and just like has a, uh, kind of a lot of stuff that has to be built to get to MVP.

I, I sort of recommend, you know, taking that time to make sure that you're dedicated and make sure you're super interested in the problem space and like the people you're working with.

 Yeah. So related to that, I think I've heard you say that you optimized in the early days for, for raising larger amounts of money, even if it meant more dilution. Um, and simultaneously you kept your burn rate down because you knew, you know, air table was gonna be this long journey, big product to build.

Can you tell us about that strategy and, and maybe what advice you'd give to founders who are looking for ways to stretch their runway?

 Yeah for us, it was very much, at any given point we were trying to figure out what we had to de-risk in order to kinda get to the next stage for the company. So I think, uh, initially like very early on it's like, uh, what, how do we actually make this the database accessible to normal people and, and non-programmers, and, uh, so, so it was all about just sort of.

Can we actually take this and make it easy to use iterating on a prototype, getting lots of feedback and kind of trying to de-risk that. And I think that's something that's really hard to paralyze. Like it, it just kind of takes a team of, uh, a few people just daily kind of working on it and, uh, getting feedback and iterating.

And you can't really like hire a bunch of engineers to kind of make that go faster. So, um, so, so I think like the early challenges we had kind of necessitated having a small focus team and actually it would've been counterproductive to try to scale it up a lot more because we sort of needed to nail that foundation of the database.

Everything else was kind of built on top of that. So if we would've hired a bunch of people to kind of build stuff on top of the early foundation, we had it, would've slowed us down those iterations and it would've been sort of harder to be nimble and change the direction of that foundation. So really, I think for us, the long runway was, was both a function of, we raised a lot of money, but also we wanted to be super lean and focused on, uh, de-risking the main risk for the company as early as possible.

And, and sort of only paralyzed once we kind of knew that we had a good foundation to build on build on, and that meant that we just had like a lot, um, longer runway to you.

 So what did the first version of air table actually look like? Like what, what were you trying to accomplish with, with the early prototypes?

 Yeah. So the, the first version was very much like I kind of talked about before trying to prove that we can make a database easy to use. And so it was actually just very much like a prototype. Like it was a lot of spoken mirrors. We focused a hundred percent on sort of like the product and UI and the interactions, and we didn't really have like a real backend, it was just kind of persisting to local store and the browser just enough to make sure we could sort of test the interactions and kind of get to a point where people could actually use it and play with it and say like, Hey, I'm confused.

How do I create like a, a view or like linked record relationship and just something where we could kind of put it in front of people and ask them to try to organize something and see if they could actually do it. So that was very much the goal of the, the very early version, uh, in terms of what it looked like.

I think the earliest version, it, it looks a lot like a spreadsheet. it was like a database underneath, obviously, but, you know, I think we, we took this approach of coop and the spreadsheet interface, which is, uh, familiar to most knowledge workers and built a lot of the same interactions. But, you know, we had typed fields and we had tables.

You could link between and foreign key relationships. but it took us a lot of iterations to kind of differentiate it and make it feel different and look different and communicate to users right away that this is not just a spreadsheet, but, uh, very much like the early version was like this grid and, and, uh, not too different from what it looks, looks like today in our grid view, but a lot of subtleties of, of how you, uh, set up, you know, things like queries and, uh, foreign key relationships.

Those are the things we really had to iterate on to make it obvious to, to early users. How, how you actually set those up, uh, uh, kind of as a departure from the spreadsheet.

 so you had some of those concepts that you still have today, like views related rows. All of those things were in the initial prototype.

 Yeah. Yeah, definitely. I think they went through a lot of iterations. I would say that, you know, like the very early version, um, views were very much kind of done in a, a sort of a, a different way. And, and linked records were kind of confusing. it took us a long time to, uh, kind of make those things more obvious.

 So one example is, you know, most databases, they have foreign keys and the way you sort of join two tables together is you write a query.

 Uh, so that was, that was kind of abstract and hard for people to figure out. So we took a very different approach of thinking to them more as like here's a field where you link to records in a different table. And we went through a lot of iterations to make that very visual. So for example, instead of, uh, when you link to a, a different record making it more of a grid, we made it more of like each record as a kind of its own card to really reinforce that, you know, this isn't just a, a sea of cells in like a spreadsheet, but instead each row represents an object in the real world in making those very visual and kind of like showing that they're almost like physical cards or objects.

Like we, we did a lot of sort of, uh, almost like bordering on skew morphic things to kind of reinforce the, the database structure to differentiate us from the kind of grid, uh, spreadsheet, uh, model. So, so a lot of like subtle iterations like that, that helped convey those points to our early customers.

 Yeah. And so when you, when you put this prototype in front of people, cuz there were a lot of database concepts sort of baked in from the very beginning and I'm curious, like, did people know what to do with this? their early users, what, when you put it in their hands, what did they, what did they do with it?

 It, uh, it kind of de depended. Like, I, I would say early on, you know, nine out of 10 people would look at it and say like, Hey, it's a spreadsheet. I know how to use a spreadsheet, but then they just kinda get confused when it sort of didn't match what they'd expected to do. Or they just like, didn't know what to do with it.

Like, oh, should I organize this or that? So for a lot of people, it just didn't click early on. And it kind of took us doing things like building templates and kind of translating the product into. A use case they could understand for it to click with them, but we definitely did have like an early adopter, you know, kind of crowd initially that sort of just understood the structure of a database more and it kind of clicked with them.

And didn't, didn't even really mean that they were sort of programmers or database people. Like, I think it was just sort of like this persona of tinker, uh, a person that's very structured in their thinking that sort of clicked with air table and they would say, oh, this is awesome. Now I can organize, you know, my apartment hunt, or I can sort of start tracking my, uh, contacts here for like a very lightweight CRM.

So for some people it's just kind of a certain mindset where it just clicked and they kind of thought in this more structured, almost like database way And so very early on, those were sort of the customers. We would put it in front of them and they'd say, oh awesome.

I'm gonna start keeping track of this thing and that thing. And for example, we had a friend Naji who, uh, was a video producer and he saw it. He's like, oh my God, I'm gonna start tracking my video productions here. So we kind of created a table of all as cast all as crew link them together. And, a lot of 'em kind of needed hand handholding to, to tell 'em what to do with it.

But for some that just kind of clicked and we saw how they used the product and really leaned into their feedback for how to improve it from there. So the.

 cool. And so Andrew, tell us about that path from like, you know, these early prototypes to the, to the first version of air table that you considered launch ready, or like what, what did your first launch look like?

 We had kind of a, a slow path to launch. I would say we, we kind of separated the concept of trying to get feedback from customers and to refine the product from the concept of like public launch. So very early on, we had just kind of a private, you know, alpha where we just get, you know, friends to use it and we ourselves would use it.

And we'd very much like sit down with those customers and, uh, just kind of, you know, see what they're using the product for and, And eventually we, we felt like

 was that like 10 people, 20 people at that time. Okay.

 It's probably like, I would say a hundred people or so, but it was, you know, it, the, the thing with the horizontal product is that you build, you know, obviously this, this very generic thing.

And then the cool part is seeing how people use it. So we kind of built this very horizontal product and then, uh, kind of rolled it out to, to more and more people over time. We had a beta eventually. Um, and so with each one of those iterations where we kind of get more customers, we would kind of, uh, see new pockets of use.

And as we added new features, we kind of unlocked new use cases and it was sort of a slow role of, uh, just, you know, add this calendar view. Suddenly we see people using it for marketing use cases, or, you know, kind of add like this con bond and people are using it for more kind of, uh, project management or managing processes and pipelines.

 I feel like the MVP, uh, for our product was not like any one thing because each, you know, it's a horizontal product, so you're kind of like constantly unlocking new use cases. so we sort of had like this just kinda slow roll of, of building new features, unlocking use cases.

 And eventually we get to a spot where we felt comfortable putting it into beta and trying to get some more customers. And while that was just kind of our, our gut and seeing that it's resonating with our earlier customers. And honestly, just kind of feeling ready from a standpoint of, can we handle the, uh, potential traffic and do we have the features so people won't lose data and it's not gonna be a big headache for us to recover it.

So I think we were, um, you know, even when we were in like the off stages constantly kind of trying to find new customers and, and grow, but, uh, kind of timing the, the launches and the going from beta to public launch more about the readiness of kind of taking on that extra traffic more than anything.

 Okay. So you went from alpha to beta to public launch, and I assume alpha was invite only was, was beta invite only, also.

 Yeah, we, I think we actually had like a alpha that was invite only, but you could refer other people. So I think that helped us kind of get some organic, um, growth there. Not, not like a ton, but we'd invite somebody to get super excited about it. They'd add somebody else to their base and then maybe that person would build something else.

So we sort of saw, um, just some traction from that. And then beta, we did publicly, we, I think we launched on hacker news. It was like maybe 20 14, 20 15. And, uh, just kind of kept it as beta just to, to sort of, you know, two things. I, I think one was to give us an opportunity to have like a bigger kind of public press launch later on.

And the other was just kind of like put the beta tag on in, in case, you know, things didn't work right. Or , it kind of seemed somewhat half baked. So, uh, yeah, we had a alpha then beta then, then public launch. But you know, uh, there was quite a bit of time between each of those, but we, we definitely tried to acquire customers and learn from them as mu much as possible between each, each of those milestones.

 yeah. How long was the alpha and how long was the be.

 The alpha is probably, I mean, it kind of, it kind depends on when you define the start of it. Cause I think we probably didn't even call it alpha initially. It's just like we used it ourselves. We gave some, some customers, but I would say maybe like probably a year or two, I think maybe even like two or three years, it's quite a while.

Um, and I think we, you know, like it was just, we had to build so many features to kind of get to something useful and we had, like I said, a commitment to the long term and pretty good runway where we were in the mode of. Let's build and de-risk the core challenges, and it seemed like launching the, the product publicly wouldn't really get us more data there, or like give us, you know, accelerate the path to that.

So we, we definitely like took our time with those early, you know, alpha beta periods, because we felt like we were getting the customer learning. We felt like we were very focused and could build efficiently. maybe we waited a bit too long, I would say, uh, I think we probably could have launched earlier and that's always the case and kind of cut, cut some features.

But, uh, for the most part, I, I think we, we got the learning and, launched and we were ready and I think, you know, in hindsight that worked out, but there's probably a place we could have gone faster.

 during that, that beta period, you know, I'm always curious about horizontal products. Like, how did you explain what air table was to people? Like when, when you posted the beta on hacker news, you were like, air table is a, what.

 I think we, we sort of mixed a, a few things and, and it was super hard. Like it was hard to describe the product. And, you know, I think that's one of the hardest parts of horizon on a product, but we, we had like a couple messages. One was like, this is a way for you to kind of build software, which very early on.

 We, we were mostly just a database. We hadn't built things like automations and a lot of different views and interfaces and so on. Um, but that's kind of how we frame the mission. And then separately, we'd say like, oh, it's kind of like a spreadsheet database hybrid, or it's like a spreadsheet with power of database.

So I think the way our customers would describe us would be it's like a easy to use database or it's a spreadsheet plus plus, but I think we always try to kind of like describe the broader mission and vision and would kind of mix those two together. Uh, but yeah, it was tough. over the years we've gotten more and more, uh, sort of focused on how we describe the, the problems we solve and sort of the use cases we're good for.

 but early on, it was very much database spreadsheet and I think it was, you know, a slightly more sophisticated or early adopter audience, uh, that, that kind of resonated with. So, so it kind of worked out, but it's, it's always been a struggle for us.

 And at what point, over the course of the alpha over the course of the beta, like, was there a moment that you remember that hit you where you were like, Hey, I think we have product market fit or like something's starting to work here.

 Yeah. I think there's a lot of different milestones. I kind of think of, uh, in terms of product market fit. initially just kind of seeing one person getting value outta the product, like our friend aji, who's using us for video production. Um, and then you sort of see a team using us and getting value out of it.

So we had a nonprofit, there was like 10 or 15 people. that started using us early on for managing all their applicants. Um, and then like, I think maybe where I really felt like we had product market fit, like is when we had our first customer where, you know, the, the customer was WeWork and, um, we kind of landed with them and it grew kind of virally within the company and, and went more or less wall to wall.

 actually that's probably like one of the most, you know, kind of visceral memories I have. Of air table and thinking like, oh my God, we've actually built something that might make it. Is I, I like did this customer tour to New York after having like, mostly interacted with customers over, you know, uh, the sport channel and that type of thing over zoom or, or, um, you know, uh, whatever the video chatting apps were using back at the end, probably zoom, um, And so I went to New York and I, I visited WeWork and I looked around and everybody's computer monitor had air table open.

 And I was like, oh my God, this is like, actually a thing. People use it. It became so much more real at that point. But really, I think that's kind of when we started to see product market fit is both, you know, when we saw a few companies and where we expanded while the wall, but also certain industries and functions.

So for example, media, we were being used for video production, a lot, or, uh, marketing teams and a lot of companies we were being used and kind of seeing repeatable, um, you know, cases of us landing within those industries and functions and then expanding within a company. And so that's very much our, the growth engine that we, we, uh, kind of live by today.

So I think that that was kind of like the first case where it really felt like product market fit. And at that point, you really double down those customers become a hundred percent customer obsessed on them. And, and that's kind of the, the path we've been on for the past three, four years.

 amazing. So it's like you had these individual user use cases from video, you know, a nonprofit, and then all of a sudden you go viral inside a company and like hundreds of people are using it inside the company. And then did, did that sort of snowball, like did, did a lot of companies start

 Yeah, definitely it snowballed. I think it, it kind of happens in a lot of different S-curves and so, you know, we started seeing. Uh, media companies where we kind of spread word of mouth, or, you know, the head of production would go from Netflix to, uh, this is just a random example, like some other company and they kind of bring the tool with them.

So you kind of see like industries or functions where you get word of mouth adoption. Um, and then like, yeah, you kind of see more and more of those. And then you'd kind of start seeing more expansion within companies. So I think, you know, it, it wasn't any one point, I think I was a single inflection point.

It kind of built over time and you just had these growth curves, like layer on top of each other, but, you know, obviously it starts very slow, but then at a certain point, maybe for us, like 2018 or so just kind of starts to really inflect. So I think it's, you know, a combination of us building more capabilities that unlocked more use cases and having the maturity to expand within companies and kinda getting these pockets of word of mouth virality within these different industries and functions that kind of all layered together to, to cause, um, you know, pretty solid growth in, in, uh, the past five years.

 at any point, did you, did you guys internally do anything to, to capitalize or engineer sort of drive that word of mouth yourself? Or was it the product doing all that.

 It was mostly the product. I think the nature of the product is that somebody builds something in their table, whether it's a product, you know, roadmap tracker, or like a marketing calendar. And they share that with people and their team and their department to do critical work. And once, you know, people start doing work in it, it generates useful data.

So you, you suddenly have your product roadmap in it, or you have your marketing calendar. Other people want to consume that. So people invite 'em to, to the base that shows that. So kind of get like what we described, um, internally as like golden data sets, like things that are really important, critical data sets for the company.

And once you have that, you get a lot of, kind of, uh, viral adoption on top of that. And other teams kind of building workflows off of. So definitely like, you know, there's a lot we are doing in the, the product to make, make it better for enterprises and help support them in kind of building these, these, uh, golden data sets, which are incredibly valuable and then letting other teams kind of build on top of them in a safe way.

Uh, but yeah, I would say mostly in product. I mean, there's some stuff we did with like, you know, referral, uh, bonuses and, you know, uh, just, just some like stuff like that that maybe helped a little bit, but probably not too much. I think it was just mostly by nature of the product being, uh, a very collaborative product by its nature that I think helped us the most.

 Were there points in, in the product evolution where you sort of felt like, Hey, we actually just have to make this UI simpler or, or we have to evolve this UI into something that's just more natural for people.

 Yeah, I think there's a few examples of that. one is that I think initially kinda like I was saying before, a lot of people just felt like Airtable was a spreadsheet, just a glorified spreadsheet. And I think, especially in the early days we had mostly just kind of, um, text and number, you know, cells, I would say.

And so we slowly kind of started to make the product look more visual. So kind of adding fields that were very obviously like image attachments or fields that were colored select options like that all kind of helped convey like. Hey, this is different than a spreadsheet. These are, uh, columns that have a specific type that's very visual, you know, and I think things like adding different views.

So not only can you see your records into grid, but you can see a calendar and you can transform them. You can view them on a, uh, a timeline or like a gallery that all kind of helped convey that mental model of like you're dealing with records, not a, uh, you know, not a spreadsheet. So, um, I think, yeah, just making it super visual was one thing that really helped.

Photo of Andrew Ofstad, wearing glasses in a dark blue shirt against a gray background

62fd724a621d224af9df5cb2

Airtable’s co-founder Andrew Ofstad shares Airtable’s early journey to product-market fit — from prototypes and launch, to go-to-market strategy.

Airtable’s early journey to product-market fit

Airtable’s co-founder Andrew Ofstad dives into Airtable’s early days, from prototypes and launch, to go-to-market strategy.

Airtable’s path to product-market fit — co-founder Andrew Ofstad on building horizontal products

Read the article inspired by this episode

Airtable's Path to Product-Market Fit — Lessons for Building Horizontal Products