The numbers in this spreadsheet are sample dummy data to protect confidentiality. They do not represent Quizlet's actual data.

Once your company is aligned around a qualitative understanding of how the business grows, a quantitative growth model can be built to measure the key variables responsible for driving each core growth loop. At Quizlet, we built this model to help teams make better decisions and deliver more impact.

Anybody from Quizlet can come to this tab and play around with these different assumptions using their own copy. Other than myself and the PMs/analysts on the growth team who regularly operate this model, this tab is the only one that other employees typically need to use. It's meant to be simple!

All of the key drivers of our growth model are listed in blue. To give an example of how we might use this model, let's say we had a team focused on reducing student churn by 10%. If we plug -10.00% into this cell, the impact of that reduction on all our outputs through the rest of the year will be calculated in the Alternate 2020 Forecast and Delta vs. Baseline 2020 Forecast tables.

You can see how powerful this tool is: you can change any variable or combination of variables you want, and immediately see what impact it's expected to have through the rest of the year in the US.

Below, we have a Baseline 2020 Forecast and an Alternate 2020 Forecast. The Baseline is what we would expect to see in 2020 if our growth trajectory from 2019 didn't change. The Alternate shows what would change if we adjusted any of the blue variables on the lefthand side. The Delta table calculates the impact of any incremental change made.

Again, all numbers here have been faked and do not represent actual Quizlet data.

The numbers in the Alternate 2020 Forecast and Baseline 2020 Forecast tables were generated for illustrative purposes only. They do not represent real Quizlet data.

The core of the model is this Summary tab, which shows all the most important outputs we care about as a business for the whole calendar year. You can dive deeper into each of these buckets on the individual Acquisition, Retention, and Monetization tabs.

In Quizlet's actual working copy of this growth model, there are many more tabs that contain raw data pulled directly from SQL queries to populate all of this information. Because the model was built on our SQL database using an easy-to-update Python script, it could be dynamically updated at any time and adapted for different types of analysis and different countries.

In this tab, we break our acquisition assumptions down into the most important user types. This is a really important point: as you're building out your growth model, you need to treat different segments of users differently. However, you don't want to have different segments for every single different type of user, because that will make the model way too complicated. There is a fine line between making the model sophisticated and robust enough to be accurate and valuable, but not so complex that it becomes completely unusable.

My advice is to use the fewest possible cuts you can without sacrificing the directional accuracy of the model. When we were determining which user types and assumptions to include, it was obvious we needed to break out students and teachers. There are 7 different persona types within that breakdown in the US. We could have built the model with every single one of those personas broken out, but it would very quickly become too complicated to be useful. Additional cuts we made were U.S. vs. other countries, free users vs. subscribers, and content creators vs. passive users. This was more of a borderline case — we could have argued that maybe this wasn't a necessary distinction. But when we looked at our analytics, it turned out that users who create content (both students and teachers) tend to be far more engaged overall than users who don't. In more recent iterations of this model, we also split out web users vs. native iOS users vs. Android users.

Here, you can see very clearly how our key actions and qualitative growth loops get translated into this quantitative model. These key actions (from both students and teachers) flow back into both the Acquisition tab and the Monetization tab.

This is the most straightforward tab: we take the users from the Acquisition and Retention tabs and apply conversion rates and average bookings per conversion to our different revenue lines. That sums up to total revenue across all 3 of our business lines.

Phil Carter

At Quizlet, we built this quantitative growth model to measure the key variables responsible for driving each core growth loop in the U.S. It can be adapted for any country. 

U.S. Growth Model at Quizlet

U.S. Growth Model

Analysis

Summary

Acquisition

Retention

Sign up to see more