In the last 6 months my life changed. I joined Founders Inc back in November and since then have worked on 4 projects: Poss.Ai, MagicStory, Scenario, and Leap. I’ve gotten to work with some of the best people in the software industry and wanted to start writing to capture the mental model I’ve observed of how building software works. The core framework — build, measure, learn. Software is about loops. 💡 You build the minimum loop you want to get someone through. You measure if it is working
7/16/2023 • 4 min read
In the last 6 months my life changed.
I’ve gotten to work with some of the best people in the software industry and wanted to start writing to capture the mental model I’ve observed of how building software works.
The core framework — build, measure, learn. Software is about loops.
💡 You build the minimum loop you want to get someone through. You measure if it is working. You extract some insight from that measurement and use that to refine your loop.
Step 1 — define your loop
With Poss.Ai the loop was signup → upload pics → generate pfps → share them on your socials
With MagicStory the loop was input email → fill out story form → get emailed story within 24 hours → read story to your kid → come back to make another
With Scenario.gg the loop was sign up → create a generator → generate images → use images in your game
With Leap the loop is sign up → try generating in the playground → get good results → add api / sdk in your app → launch app on top of us
Step 2 — measure your loop: define your KPI / northstar and track it
Since we launched with a usage-based pricing model at Leap, our KPI is number of weekly active non-dashboard API keys — a proxy for how many apps are built on Leap.
We stare at this number every day. We try to at least double it every week. If we don’t hit it, we know something isn’t working. If we do hit it, we figure out what caused it to go up, and then set a more ambitious goal.
Step 3 — get enough traffic coming in to have statistically significant data
Great. You have your KPI defined and visible. Now what? You need a consistent growth engine that will bring you enough people flowing through your loop. If done well, this is like a constant stream of water that flows into a mill.
The constant stream is your growth engine. The watermill is your loop. No matter how well your mill is converting water, the stream needs to continuously keep pouring water.
In practice, this means you get a statistically significant number of people to try your loop each week. For example: during week 1 of Magicstory, I launched without a backend ready and have a queue of 1,000 stories. I manually processed 100. That means I had a sample size of 100 people in my loop.
That’s a start, but not really statistically significant. I really needed to get more like 500 or 1,000 through the loop to start learning something meaningful.
Step 4 — finding insights from the data
Once you start measuring your loop, you will have clear evidence about where the problems are.
What to track:
You want to quantify it all with statistically significant data, and you want to create a view that shows this breakdown for each weekly cohort of signups so that you can attribute changes you make. (Ideally multiple views that show each cohort’s performance after 3 days, 7 days, 30 days, etc).
💡 you are building a machine → you should know how much it costs to acquire your KPI / northstar metric (CAC), and eventually the LTV of that KPI.
Step 5 — building a growth engine
“First time founders focus on product, 2nd time founders focus on distribution”
I used to think paid ads were for chumps. I was wrong. I was the chump bc I didn’t have enough people using my product to have statistically significant data about anything. In my early days as a founder, I was building a lot, but not able to measure or learn bc of this.
Paid ads are great fuel to a fire, and need to be part of a growth engine. In our case, our growth engine is driven by content + viral influencer. We make content about what we ship, we get eyes on that content through devrels or paid ads. Long term we make SEO plays. Short term we partner with established brands to build our credibility.
The goal? Drive down your CPA and CAC while bringing in high quality / high volume traffic.
How? The same way as with your product. Build, measure, learn. You should be measuring the attribution of every campaign you run. You need to figure out what your channel will be. Start with spray and pray and then double down where you have traction.
Defining your A+
Now the framework is in place — but where do you focus? Product? Growth? Visibility?
You need to define your A+ — if you have no loop then you need to build your loop. If you have a loop but no measurement then you need to measure it. If you have a loop + measurement but no insights you need more data and should focus on growth.
This is a daily question to ask yourself. What is my A+ problem to solve. Ignore everything else.
What winning looks like
Winning = explosive growth in your KPI. We shoot for at least 2x growth week over week, with a proper build, measure, learn cycle and you are on your way to a well-oiled machine of a business.
This week at Leap we 35x’d our KPI — active API keys went from 62 to 2,100. Now what?
The goalposts shift — the northstar is no longer just active API keys, but actual legit apps that are doing high volume API calls every week. We are still at 0/10 right now.
I’ll see you when we get there.