Scalable Growth

Why create a wave when you can ride one?

When we began StartPlaying we had one scalable growth channel.

For us, the channel was paid ads. Both Google and Facebook ads worked. When we were small, the CAC was low, the market wasn't saturated, and IOS updates hadn't yet supported privacy over lower CPCs.

So, ads were a brilliant way to scale.

The problem is, over time, they were not. Things change, and nothing lasts forever. We had to find a new scalable growth channel. And it took us four years to get there.

After ads, we were at a loss. We didn’t know what we should pursue. There are infinite little growth channels, and we were past the point where little channels would matter. We needed something scalable. So we went back to the basics, and asked ourselves, "How do consumer products grow?"

There are three ways. First is through paid ads. You put money in one side, and get paying customers out the other. For us, this worked up until a point. But as we grew bigger, it stopped functioning. We needed to find another channel.

Second is word of mouth (WOM), the most important channel. If you've achieved high product velocity and created a valuable product, WOM will always be great.

We noticed that more than half of our conversions came from WOM, but our WOM was a little different from most.

We have two users in our marketplace: Game Masters and players. Game Masters (GMs) host games, and players paid GMs to host games. Or so we thought.

Originally, we thought players would come to StartPlaying (SPG) in order to book GMs to run games for their groups. A player would bring a group of four or five friends, and book a GM. Presto salvo, business model made.

It turns out that was not the case.

When we built our product to serve groups rather than individual players, we did not have product-market fit. Sure, we could get our friends to join, and GMs were signing up, but very few players made bookings.

That is, until we made slotted games, which took our product to the next level.

In a slotted game, a player comes to the site because they don't have a group. Yes, in a sense, they’re looking for a GM and an excellent story. But in a far deeper sense, they're looking for a group to play with. They come alone. They come because they don't have friends playing at the same place, at the same time, or game they want to play.

If these players definitionally don't have folks to play with, then they don’t make good referrals. And this is what we found. Players historically didn’t tell their friends about SPG.

So why does our platform see such high WOM? Game Masters spread the word.

GMs sign up to the platform because we enable them to do the impossible. We provide legitimacy and a stable set of players so that they can fulfill their dream career, a career where they get to be creative, work from home and set their own schedule - the best gig economy job on the planet.

The third way consumer startups grow, is with search engine optimization (SEO). You hear a lot of bad things about SEO. It's slow, and you're relying on an external platform like Google. Some of these are true. But most are product dependent. We found we could quickly and easily affect our SEO.

SEO was our second-biggest channel, aside from WOM. Our users were already searching for us or products like us. Roughly speaking, their searches bucketed into two categories.

Some queries were to play Dungeons & Dragons, our biggest game on the site. These were highly competitive queries. And fortunately because of our domain authority, we were already ranking for them. Other queries were for very specific games, what Google would call long-tailed queries. And parts of our site were already ranking for them too.

Across the board, we were doing an okay job at SEO without ever having tried. These are the best opportunities.

When someone joins SPG, one of their required readings is The Mom Test. This tiny book, barely more than a hundred pages in simple prose, talks about how to discover real problems. It reads, "If you torture the data long enough, it will tell you anything." At SPG we say: “If you torture the users long enough, they’ll also tell you anything.”

It's very easy to invent problems. And I saw this before at Google. (I even wrote an entire essay about this called Hacking Culture.) We wanted to avoid this problem at SPG. When we try to understand our users, we try not to ask leading questions. We try not to probe on specific feature ideas or problems that we invent in our mind.

Instead, we try to understand what directly frustrates users - what is top of their mind. And often user behavior is the best tell. If your users are already doing it, then there's good evidence that they want to do it more.

Our users were already using Google in order to find us. (In fact, some of our users would use Google to find specific games on SPG!) From this alone, we knew that SEO would be a win.

Let’s distill two principles so far: 1) If you’ve already achieved product-market fit (roughly $1M ARR) you need to find large channels for sustainable, scalable growth. Dicking around with small activations won’t work anymore. 2) Don't build castles in your head. Talk to your users and observe their behavior, most of all. Why create a wave when you can ride one?

Our initial experiment in SEO was, in hindsight, very rash. We had terrible landing pages, so we built new ones. Then, in a very bold move, we pointed each of our old pages (the ones Google already liked) to our new pages.

And guess what? Google ignored us. Not only did it ignore us, it promptly downgraded all old pages and ignored the new ones. It was terrible. In one fell swoop, we had decreased our SEO by 20%, and it was the first test that we did. What's more, it was irreparable. We couldn't just immediately make a code change and go back. It might take months or years for Google to give us the SEO value that we had just lost.

So, did we give up?

Of course not. We kept tinkering. We learned a lesson, we measured, and we held ourselves accountable.

Instead of forcing Google to new pages, we made our new pages rank better independently. Instead of directing Google, we softly nudged it. And we did so in a very MVP way.

While we knew that SEO could be a powerful tool, we didn't know whether it would be scalable or achieve results fast enough. So everything we built was scrappy. We made shitty code, launched quickly, and measured ferociously.

And here's the next two principles: 1) Always do a full product rep. Your first foray into many fields will fail. And if you only throw spaghetti against the wall, you'll never know what will really stick. Was it too low quality or the wrong direction entirely, you’ll never know. We focused the entire company on building and improving SEO. And we continued until we built a great product. 2) When experimenting, build MVP. An MVP does not mean a low quality experience. It means focus on the happy path, the thing that 90% of users see.

Prior to doing SEO, we were on a bad run. We had launched a series of experiments and tests that didn't move the needle. So when we launched SEO and got a big win, the team was exhilarated. We saw business metrics move. And it felt like we hit on gold.

The fundamental reason was simple. We weren't creating something new. We were riding a wave. We noticed a behavior in our data. And we didn't try to change the behavior. We enabled it.

What we did next, was equally important. We could keep doing MVP launches, but they would occupy core team members (pilots), whose activity should be spent elsewhere.

So at this point, we decided to scale.

And to speed up, we needed to slow down. We put the brakes on doing any future SEO work and went back to the engineering drawing board. We knew what worked before, and we needed to build the framework and the scaffolding to quickly ship and iterate on new SEO projects.

We launched small. And once we understood the core concept could work, we had to scale. We spent the next couple of weeks engineering to make sure that we could scale SEO, not from zero to one, but from one to n.

Next we made sure that the non-technical side, the operational side, was scalable.

This is simpler than it sounds. To ensure that the technical side scales, you need reproducible code, steps that any computer could follow. And you need servers to scale up the code depending on demand.

On the operations side, it's the same thing. You build playbooks that any human can understand. And depending on how fast you want to scale SEO, you can hire more or fewer contractors, thus scaling up your servers.

This turns a tactic which requires a pilot into a strategy that you can scale out, monitor and measure.

You start with experimenting. This is SPG’s strength. It's being creative. It's finding little-known channels. It's doing Reddit ads, influencer marketing, handing out dice bags at conventions, running D&D charities, partnering with publishers, and doing email blasts.

All these are creative, fun and excellent ways to take the business from zero to one, but not from one to n.

There's a key reason for this. These creative experiments require a pilot.

There's a dichotomy that's rather useful when thinking about startups and hiring. You can hire two types. You can either hire pilots or you can hire rocket fuel. Your startup is a rocket ship that needs both.

Pilots are people that can lead projects. They're smart, creative, multidisciplinary. Folks with range and good leadership. They could be founders themselves.

But pilots are hard to find and align. Even more so, you can never be sure of them until you've put them in the cockpit.

Rocket fuel is something else. Rocket fuel takes what a pilot is doing and amplifies it. Without rocket fuel, a pilot is just an idiot in a large hunk of metal. But together, a pilot and a team filled with rocket fuel can go far.

If you have projects with straightforward, easily measurable objectives, you need rocket fuel. Projects that are dynamically changing, that need rapid experimentation, need pilots.

But if you have to pay pilots to continuously experiment on new channels without ever being able to scale them, you'll never effectively transform your cash into new users.

Instead, you want growth channels in straight lines that you can put cash or contractors behind to scale your business. (Obviously, this becomes easier the more product value you provide, which is why Product Velocity is ultimately the most important thing you can do.)

So how do you achieve a scalable growth channel? You experiment with many growth channels. You hold yourself accountable. You measure as concretely as you can. And then ones that pass the Bradford Hill criteria (in the positive direction), you invest into. You try to automate them, such that you could print out the instructions in a binder, hand it over to someone, and they could go out and get you new users.

And finally, you constantly measure your efficacy.

We failed here. We thought we found a growth channel in ads, and we didn't need to measure its efficacy anymore. But we, like all startups, suffered from the law of shitty click-throughs.

Eventually, people get bored. Or you capture all of your potential users in that channel, or they move on to different platforms.

Nothing stays the same forever.

We spent four years wandering the growth desert, hopping from small activation oasis to oasis, but never truly finding the promised land of scalable growth.

But during that time, we've learned much. We not only learned how to find and verify scalable growth, but we learned what we were looking for. My hope is that in some way this piece provides a map, albeit fragmentary, of how we found our way out of the desert and how you can avoid it altogether.

So experiment. Find highly effective growth channels. Build playbooks to scale them out. Delegate and continuously measure their efficacy. And achieve scalable growth.