Long-term differences

Some things about your business are true month after month: customers on annual billing churn less, customers in one region have more failed payments, customers who start on one plan upgrade more often. These long-term differences are the background your other insights play against.

A difference that holds steady isn't news, so Insights tells you about one when it appears, changes or ends.

What's compared

At every month-close GrowPanel compares groups of customers with everyone else over the 12 months before the close, on four measures:

The groups are:

• Where customers are: every level of geography, from continent to region to country.
• The plan they started on: the plan group of their first subscription.
• The billing interval they started on: monthly, quarterly or annual.
• How they pay: card, bank debit or another way.
• Whether they started with a discount.
• Whether they started with a trial.
• Two of these together, for example "customers in Western Europe who started on annual billing", but only when the combination does something its two parts don't already explain (see below).

Compared at the same age

Churn and expansion change a lot with a customer's age: new customers churn far more than customers who've stayed a year. A group that is simply newer than the rest, like everyone who joined after a price change, would look like it churns more even if it behaves exactly the same.

So GrowPanel compares like with like: brand-new customers with brand-new customers, customers in their first year with others in their first year, and long-standing customers with long-standing ones. The everyone else line is what the group would show if it behaved like everyone else at the same ages, which is why it can differ a little from one group to the next.

This is also why these numbers don't appear in any single report: no report compares a group with an age-matched everyone else. The group's own rate is its churn (or expansion, or payment rate) over the 12 months.

When a difference counts

A difference has to be clear, lasting and big enough to matter:

• It has to be very unlikely to be chance, and hold through most of the year, not just one busy quarter.
• It has to be a real difference, not a few percent, and involve enough customers to mean something. Small groups are left out until they have enough history.
• Once found, it's kept while it stays reasonably clear, so a difference doesn't flicker on and off. It ends only after it has clearly faded for several months in a row, and one that ended has to prove itself again before it's back.
• When the gap grows or shrinks a lot, that's told as a change, but not more often than once a quarter.

Two sides of the same fact are told once. For billing interval, payment method, discount and trial, only the smaller side is looked at. "Customers paying by bank debit churn 41% less" is told, not also "card payers churn more", which would be the same fact against a tiny "everyone else".

Geography is told at the level that explains it. If the UK explains a difference, you hear about the UK, not Europe. If a whole region shows it, you hear about the region, and a country that stands out even inside it is mentioned as a detail.

Two-dimension groups need to earn their place. If customers in Western Europe churn 17% less and customers paying by bank debit churn 55% less, customers in Western Europe paying by bank debit would be expected to churn about 63% less just from the two added up. The combination is only reported when it clearly does something beyond that. Otherwise it's just two known facts stacked.

How they're told

A difference becomes an insight when it moves:

• Appearing: "Customers who started on Team now churn 36% more than everyone else"
• Changing: "…now churn 16% more than everyone else (was 36% more)"
• Ending: "…no longer churn more than everyone else"
• Widening from a country to its whole region

Open one to see the group's rate and everyone else's over the last two years, each point being the 12 months up to that month, with when the difference was found, changed and ended. For churn, it also names the cancellation reason the group gives clearly more often than everyone else, when one stands out.

A difference about churn carries a money amount: roughly the extra cancellations a month, valued at your average customer. That's how it's ranked against your other insights in the Situation room.

Once a difference has ended, it stays on the Earlier timeline, so you can see what was true when.

Acting on a difference

A difference tells you who behaves differently. It rarely tells you why, and it almost never means that moving customers into the better group will give them its numbers.

Customers who choose annual billing churn less partly because they were more committed when they signed up. Customers on your top plan expand less because there's less to expand to. Pushing everyone to annual billing or to another plan changes the label, not the customer. That's why the suggestions on insights never suggest moving customers between plans, billing intervals or payment methods to lower churn, unless your own data shows that customers who moved actually did better afterwards.

Good uses of long-term differences:

• Set expectations per group. A region with 20% more churn isn't a crisis if it's always been that way.
• Decide where to look. A group that churns more and names one reason more often ("too expensive", "switched to another service") is a clear place for customer conversations.
• Target the fixable part. More failed payments in one region is often about local payment methods or card issuers, and that's fixable. More unrecovered payments points at dunning.
• Watch the trend. Insights tells you when a difference grows, shrinks or ends, which is often the first sign that something you changed is working.

Use Ask AI on any of these insights to explore what might be behind a difference.