GoCardless was founded in 2010 by Matt Robinson, Hiroki Takeuchi and Tom Blomfield. Today, the recurring payments platform processes $25 billion in transactions a year and counts TripAdvisor and the Guardian among its 50,000 business customers around the world. Including a Series G round in 2022, GoCardless has raised $529 million in venture capital.
On scaling and product-market fit:
When we were struggling to move forward in our early days, Paul Graham of Y Combinator gave us some brilliant advice. He said, “Stop building these features that are going to take 18 months to complete. Instead, spend a week cold-calling your users. Tell them that you’ve already got all these features, and see what happens.” That week became the most depressing week of all of our lives. Of the 500 people we called, maybe five were interested in using our service.
We got to the point where we understood that there were hundreds and hundreds of different use cases--but even just trying to serve one of them was going to be incredibly difficult. So it pushed us to think: “How could we make it work so that we can serve these 100 verticals at the same time?” And that’s why we focused on the API. And why we went and focused on acquiring partners.
Really go deep. Try and figure out how that business works. What is their role in that business? Put yourself in their shoes: What decision would I make if I was them? Would I use this product? And why? And what would it do for me?
We didn’t start out with a problem. We tripped over one when we found someone who had a problem, and listened to them. Afterwards, we went and looked for 10 more people that looked like that. We assumed, if these 10 have a problem, there must be 100 other people that have that same problem. And then we went off to look for those.
On founder sales:
Because I understood the product and worked closely with the product team, we were able to come up with creative solutions that nobody else would have. For example, we had a company who needed to bulk change 40,000 people over by a particular day. It was essentially impossible to do but we managed to find a hack for how we could do that. So it was definitely a maverick sales organization.
I think we were in our fifth office when we hired our first real sales person. I was probably getting through the work of a handful of salespeople. But more importantly, I was able to distill it into feedback that I gave back to the product organization.
We knew we wanted Xero integration, but it took three or four years before we got them to integrate GoCardless. But then it got so much traction, our product eventually became a meaningful partner for them. We realised that it's only when you start to impact a partner's average customer experience or their P&L that they really start to care about you - and rightly so. That was one of the things we got wrong in the early days: you need to understand what your customers’ top three priorities are. You need to be one of them--ideally, the top one. Back then, we didn’t, and we weren’t.
On demos / “Wow effect”:
All of our sales were very tech-powered. We were able to give exceptional demos because we had brilliant demo accounts that we could use thanks to our product team. It sounds like nothing now but we would put the business’s logo on there and tailor the demo account to how they would use it.
At your demo, you don’t need to show the customer the whole product. Our demos were probably quite annoying for customers because of the half hour we booked I would ask them questions for the first 29 minutes. I was trying to find your killer problem. And once I found it, that’s the only feature I’m going to show you.
I know there’s five or six killer problems that we can solve, because those are the reasons why other people use our solution. I use my question structure to find out if you have one of those problems. As soon as I find one problem, I keep going to see if there’s another one. And once I know all of your problems, I do the demo. I’ve had demos where the person tells me, “Everything’s great! It’s just that each week, I have to manually upload 1,000 payments.” I just show them the bulk import CSV and say: “Oh, you should look at this feature.”
On the sales process:
The beginning of the sales process is when you have the biggest chance to build up all the ammunition for the entire sales cycle. Our time to sell could be 6 to 12 months, but if I don’t start to gather my ammunition in the first call, it’s going to be that much harder to get it later.
On segmentation / sales process:
Part of the difficulty in the old days was that we were trying to do too much. We were trying to serve 10 different verticals. We were trying to do what we would with 100 million in revenue: inbound sales, outbound sales, SMEs, corporate enterprise partnerships, all of these sales channels. That’s right, but that’s not how you get there. We were spread very thinly across all of these channels. When we saw something working, we capitalized on it, but it was quite an inefficient way to get there.
The metric I look at the board level the most, and I think is most useful, is the cohort chart. The guys refer to it as “Matt’s chart”. Every annual cohort is lined up, and it shows what they were worth after n months. It shows you if there’s a mismatch in what you pour in and what you get out, with a certain delay because of the sales cycle. We had a couple of years where I was convinced the machine wasn't scaling efficiently. And I could see it in that chart. We had increased our spend on sales and marketing by multiple X, and yet our cohort value only went up about 60%. That was clearly not working.
For SaaS, when your margins are 70, 80, 90%, it makes sense just to pour money into the machine and just acquire growth at all costs provided you don't build a huge fixed cost base. Cost of sales is not a big priority. Because time in market is your friend. As long as you’ve got good CAC versus LTV and a sensible payback period, then you’ve got a scalable engine. Sure, CAC’s not Total Cost of Sales. But the reality is: when the margins are good, and the lifetime is long, that stuff largely disappears.