# What is a cohort?

# Cohorts

The term “cohort” sounds fancy and a bit daunting, but it really just refers to a group of people that share the same characteristics. It originally referred to part of the Roman army: a Cohort was 1/10th of a Legion (technically, if you destroyed a Cohort, you decimated the Legion.)

In other words, a cohort is a group of people within a larger population.

Cohorts help us see how people behave under similar circumstances. It feels pretty obvious that we might want to organize people into groups and then look for differences between those groups, but few projects take such a deliberate approach. You should consider being one of the ones that does.

TIP

Understand your users better with cohort analysis tools on Userfront.

# Cohort uses

If you've ever heard of Baby Boomers, Millennials, or Gen Z, you're already familiar with cohorts. You probably even identify with one of these cohorts yourself.

We use high-level cohorts to talk about changes in society: how growing up with the Internet differs from TV-only, or how social media has impacted teens. Cohorts are useful for this kind of qualitative reasoning, but they really shine when we get specific.

Used in medicine, cohorts have helped us understand maternity, childhood development, general health, medicine, smoking, cancer, and virtually every important topic you'd want to know about.

Cohorts tell us about graduation rates, family values, the economy, laws, and how to run a business or grow a project. Any time there is variation in how a group of people experience the world, we can learn from cohorts.

Applied to more practical areas, cohorts tell us about our users, customers, or potential customers, and what we might do to engage them and provide value.

# Cohorts of users

Before defining our cohorts, it's a good idea to ask: “what change do we want to measure?” Maybe we have a new service with ongoing product changes. Maybe we are changing how or where we talk about a project, or perhaps different users are joining at different times of the year, and that might be meaningful. If we're changing how we interact with users, that's important too. For each type of change, we want to measure different things:

# Product changes

We can measure the effect of changing a product in two ways: by tracking all users over time, with each cohort being a week or month (“users who signed up in January”), or we can track the same users before and after the product change. If there are lots of minor changes to the product, measuring over time is probably a good solution; if there is one major change to the product, like a major new feature, it probably makes more sense to look at before and after.

# Messaging changes

Sometimes we change how or where we talk about our projects, and in these cases we can use cohorts to tell how perception affects usage. For example, if we offer our project as a cheaper version of a competitor, that can attract different users than if we offer it as something that has a certain benefit (even if the competitor has that too). In the first case, our potential customers might be more price-sensitive but also more educated about the market, while in the second case, potential customers might want to solve a problem but not know who we are.

# Timing changes

Sometimes the same person will behave differently at different times: due to seasonality, like holiday shopping; by time of the month, like after a payday; or even throughout the day, like social media usage. If we suspect there might be some difference in how we can serve these different cases, then it often makes sense to include each person in multiple “cohorts”, and instead group based on usage patterns.

# Support changes

People change their behavior a lot based on whether or not they understand something. This means that our onboarding, user education, and support can have a big impact on a cohort. But the opposite is also true: if our materials are good enough or if the product “makes sense”, then extra support can be an unnecessary cost or can frustrate people who just want to get started on their own.

Last Updated: 4/24/2020, 8:53:18 PM