👈 Back to all MicroConf 2017 talks
- John Collison, co-founder of Stripe. Interview hosted by Mike Taber
- Biggest initial challenge in building / developing Stripe?
- We had no idea how to build Stripe
- Anyone who had used the original merchant providers knew what a pain it was
- “How hard can it be to build something better?”
- Very hard, it turns out
- Spent a long time searching around for the right partners. Learning how to speak the language of those financial partners
- One of the first partnerships – Wells Fargo – we were referred to. So we had to try and have that credibility
- So useful to sit down and talk to customers
- First commit – 2009
- Took a long time to get the first 100 customers
- Hard to know what customers want when you don’t have any yet
- How long did it take to get the first customer?
- 3 months, when we were working on it part time
- January 2010
- Was helpful – Stripe’s not a 1-shot product. You integrate it and then keep using it for a long time. Your business runs on it. So once we had a customer, we had to keep going
- Early days – how much of your time was working on the technical stack vs going out and talking to people / marketing? Word of mouth was big for Stripe, it seems. What did you do in the early days to drive that?
- I’m not sure how generalizable any of Stripe’s story is
- But we did a lot of customer development in the early days
- Messaging – telling the story early
- Word of mouth isn’t some fundamental truth of the universe
- How well / how much you tell your story are big inputs into the word of mouth equation
- There was PayPal. And a bunch of banks. Who weren’t used to speaking to the online / developer community
- We made it clear we understood SaaS. Developers. Etc
- The code on the homepage that you can copy/paste to the terminal was purely to make clear that we understood devs. We understood that the API was important
- Injecting the logged in user’s token into the code snippets. Was that intentional from the beginning? How did that come about?
- We’ve since retrofitted a strategy on top of that. But it was accidental at the time. At the beginning you just do whatever feels right. You take the things that work and pour more gas in them
- So those very detailed error messages, and real tokens in the code snippets
- We did them because they were useful, but they turned out to be hugely valuable. People would tweet “hey look how great this error message is”
- So now it’s more intentional
- Do you have a process in place now for the documentation. How early did you start being intentional about those things? The API keys in the docs, the error messages, the little things that all make Stripe so “right”
- A lot of it was just part of the product from the beginning.
- Stripe now has ~200 engineers, and most work directly or indirectly on the product. Plus many other employees. You have to formalize those things eventually
- See what works well. Double down on those things
- Stripe wasn’t an overnight success
- Year after year, you just look at what’s working, what isn’t
- What 2 things would you do differently given what you know now?
- There’s no big giant thing that we definitely would’ve changed
- Spent years banging our head against the wall to things that were pretty simple in retrospect
- Usefulness of the dashboard. Would’ve done that much earlier
- Maybe would’ve had a cutoff that you couldn’t still be using the API versions from years ago
- The dashboard. Given the size of the customer base, how difficult was it to change over? Particularly where you have people still using the API version from 2011, which is amazing. New features, making sure things still work, customer expectations…
- That was something people had with the PayPal API. Stuff would randomly change, and break. A fundamental of an API is it should keep working.
- When you make a request to the old API, there are a series of translations that happen to translate to the latest API.
- Is there a low point that sticks out in your mind – anything you weren’t sure you could get through?
- Yeah. PR never tells the whole story. It’s been useful to have both Patrick and I, to dampen out the swings of high highs and low lows.
- Early days – some pretty bad outages
- “We once had a very serious outage on a weekday in December. I wouldn’t recommend that as a life experience.”
- Pretty scary
- In the early days no one would notice (had no customers)
- But as we grew… that got pretty scary
- Early days – an employee leaving was scary too. Are we going to pull through this? That’s hard
- What’s your big focus in terms of acquisition now? Are you really focused on acquiring new users? What’s the strategy? Or still word of mouth
- Word of mouth is definitely a big part of it. We’re in a well-defined space where people have to buy SOMETHING – you can’t just not take payments.
- We’ve always worked actively on PR. You can use that to help drive word of mouth. This week – events. Acquisition of Indie Hackers. Relaunched Atlas.
- If you get that right, have something interesting to talk about (or can make it interesting)
- Atlas launch was on the front page of the New York Times. Would not have believed that 5 years ago
- You literally can’t buy that (or it’d be pretty expensive)
- Always working on the funnel
- Think about how we bring people into that
- “Content marketing” as a term starts to lose its interest as a term. But we’ve always done it on an ad-hoc basis. Capture The Flag was big in the early days. We want to do more stuff like that
- It’s hard to do very creative and interesting things at scale. We’ve not figured out a systematic way, it’s still pretty ad-hoc
- Where do you see the mobile checkout experience going over the next few years?
- Super interesting. We think Apple Pay and Android Pay are going to win. They’re big enough they can keep iterating til they get it right. The conversion rates on those is higher.
- On the consumer side – look at Wish. Conversion-optimizing machines. Their funnels are crazy good. Hoover people in from Facebook.
- What I’m excited by – new products being possible BECAUSE mobile checkouts are easy enough.
- E.g. paywalls in journalism. I’ve always hit back. But as the effort to pay disappears, I suspect there’ll be more uptake there (buying a day pass, say)
- It seems from the way you talk, you and your partner are a very unified front. What’s the decision making process behind the scenes, how do you split it
- For it to work, you need a few things:
- Shared long term vision in where things are going. Worldview. You have to be pretty aligned there, and that’s pretty hard at times. Broadly what Stripe is, where it’s going
- Agreement of what matters and what doesn’t. Many things Patrick and I disagree with on a daily basis, but you pick your battle about what’s important
- We have more overlap between tech + business than is typical (although we have the area each of us owns)
- We’re brothers, we’ve been resolving conflicts for a while!
- For it to work, you need a few things:
- Any plans re cryptocurrency?
- Strategically, Stripe should be the way a business accepts payment from its customers. We’ll support whatever our customers need. We launched ACH. We’re beta testing a number of the European mechanisms. Pragmatically, if and when a cryptocurrency gets significant customer share, we’ll support it because our customers need it. But we don’t push it for its own sake
- There are interesting things happening in the newer cryptocurrencies, but it’s not top of our list
