As we spun up the growth team at Drift, we were trying to figure out what the best way to run it was and the best way to make prioritization decisions. Our VP of Growth, Guillaume Cabane, had a couple growth leaders from some different companies join for a lunch-and-learn to see how other teams prioritized and did growth experimentation and whatnot. Darius Contractor presented his EVELYN Airtable framework, and we took his concept of prioritizing with the different columns and inputs and all that to help make priority decisions. And then we really made it our own by adding the spin of focusing it on unit economics. That ultimately became our process at Drift, and we built our whole Airtable system around it. This presentation was created to document our process and share it with other teams and companies who wanted to adopt it.

Before we implemented this, we were running into situations where we were prioritizing based on a kind of vague idea of something being the “best idea,” or just deferring to the loudest voice in the room. We really wanted to create a more objective means of approaching it because the whole point of growth teams and growth experimentation is that the data is supposed to lead the work that gets done. We knew that our core goal as a growth team was to drive additional revenue, so we just worked our way backward from that, and it led us to realize that using unit economics was the best way to understand upfront what the potential impact of a given piece of work was on our goals.

Prioritizing via unit economics ends up forcing the decision-making to be about the assumptions made around the experiment. So instead of debating the assumptions, we would be debating the data that led to those assumptions, because that is the starting point of the output. If someone says, “I believe that this could drive 100 more sign-ups a week,” then instead of debating whether it’s actually 100 or if it’s more like 50, we could instead focus on questions like: Where is that 100 coming from? What do we have that’s showing us that? And what other experiments have we run that tie into it? Maybe you could get there even if you’re not prioritizing on unit economics, but using those numbers really forces you to peel back all the data inputs in the first place. If you’re really rigorous with this and use it to track all the results of every single experiment, you can get to a place where you start to create a loop that corrects all the assumptions that go into the model in the first place.

This is really just about working backwards from revenue. So it's a matter of picking the things that you track as a company. You might only be tracking the number of signups and the number of customers, and you can use those as your two inputs and then you run all your experiments off of those. For example, if you have 100 customers paying $1M total, that gives you $10k per customer. If to get to those 100 customers, it took 1000 sign-ups, then you can work your way back to saying a sign-up is worth $100. It doesn’t require any complicated math or tons of records.

We had a weekly meeting where we would review everything that was input into the table that week and then debate the underlying assumptions for each one. In this clip, I walk through what those 45-minute meetings typically looked like, but the key takeaway here is that they served as a forcing function to work through whether things that were presented as high confidence really were, so that we always knew that the work we were moving forward with was around initiatives that would really deliver the value we expected it to.

Matt Bilotti

There are so many ways that prioritization can go wrong. This approach helped the team at Drift be sure that the work we were doing always had solid data behind it.

As we spun up the growth team at Drift, we were trying to figure out what the best way to run it was 

Before we implemented this, we were running into situations where we were prioritizing based on a kind of vague idea of something being the “best idea,”

Prioritizing via unit economics ends up forcing the decision-making to be about the assumptions made around the experiment

This is really just about working backwards from revenue

We had a weekly meeting where we would review everything that was input into the table that week and then debate the underlying assumptions for each one

Experiment Prioritization Guide at Drift

Sign up to see more