Usage Depth

For years the only question anyone asked about a customer was whether they were logging in. Daily logins meant healthy, two quiet weeks meant at risk. It missed something large. A customer who signs in every morning, opens one screen and leaves looked exactly the same as a customer who signs in twice a week but runs everything from drafting to approval inside the product. Both showed up green.

The gap surfaced at renewal. Accounts with three straight months of strong login numbers kept giving notice, and the explanation was always some version of the same sentence: they had used it every day, but only as a notepad, so switching to something else cost them nothing. The product had been opened. It had never been woven into how the team worked, and a login count cannot tell those two states apart.

Usage depth exists to fill that blank. It counts how far into the product a customer gets rather than how often they arrive. A customer using one feature can find a replacement for it in an afternoon. A customer who has chained several features together would have to change the way the team works in order to change tools.

Fix a list of core features, then count how many of them an account has touched in the last 30 days. Pagedeck settled on twelve core features, so an account that used four of them has a usage depth of 4, or 33 percent.

Three decisions come before the first calculation, and each one moves the number.

1. What counts as a core feature. A longer list mechanically lowers every score, so it can't be every screen that exists. Pagedeck kept only the features tied to the reason customers pay: co-editing, comments, approval requests, templates, permission settings. Changing a profile photo is not on the list.
2. How much use counts as use. If a single click qualifies, every account spikes to its ceiling the week after onboarding training and collapses by half the following month. Pagedeck counts a feature only when the account used it on three or more separate days.
3. Who the unit is. The contract is with a company, so the count belongs to the account, not to individual people. Counting per person turns a fifty-seat customer where everyone uses one different feature into an average of 1, which erases the fact that the company as a whole is using all twelve.

The first month of a contract is not read at face value. Until time to first value — the stretch between signing and the day the customer sees a real result — has passed, a low number is simply what a new account looks like. Pagedeck starts reading the metric at day 60.

There is no threshold that means safe. List lengths differ between products, and six features out of twelve is not the same state as six out of twenty. The average Pagedeck account uses three of twelve, and that number is not a problem on its own.

Three readings turn the raw count into something you can act on.

1. Read the distribution, not the average. An average of 3 covers both a world where most accounts sit at three and a world where half the accounts use two or fewer while a handful use eight. Pagedeck tracks the share of accounts at two or below separately and reworks onboarding when that share passes 40 percent.
2. Read the same account over time. An account that went from 2 in month two to 5 in month six is healthy. An account that went from 4 down to 2 over the same stretch is not, and the drop usually means the internal champion changed jobs or a neighboring team moved to a different tool.
3. Read it against what the customer wanted. A customer who bought the product to cut document review time and is using templates and approval requests is fine at a depth of 2. A customer sitting at 6 on features unrelated to their goal is not reassuring.

The reason to watch this metric at all is that it moves earlier than the revenue ones. Gross revenue retention (GRR) — the share of existing customer revenue you keep — only settles once renewals close, while a rise in shallow accounts shows up two or three quarters ahead of it. Net revenue retention (NRR), which adds expansion into the same calculation, lags in the same way.

"Logins on this account look fine, but usage depth is 2 out of 12. Renewal is three months out and we need to do something."

An account that reads green on activity is actually cheap for the customer to walk away from.

"Let's not switch on every feature to lift the depth. Attach the three that map to what this customer is trying to do."

Raising the number and helping the customer get a result are two different projects.

"Median depth for last quarter's new customers went from 2 to 4. The onboarding change worked."

Here the metric is grading the onboarding process rather than any individual account.

At Pagedeck, Sofia sorts her accounts by usage depth every quarter and works from the bottom of the list up. For anything at two or below she starts by asking what the team is trying to get done this quarter, and if a feature tied to that goal is sitting unused she books a short session about that one feature. The goal is not to get customers to twelve — it is to remove reasons to switch — so she attaches at most two features at a time. Accounts above eight get the opposite conversation, about adding seats.

The most common misreading is that more features used means a better customer. A bigger number looks like deeper commitment, so the assumption is easy. But customers buy outcomes, not feature counts. An account with six unrelated features switched on mostly gains more settings to maintain, which pushes up customer effort — how much work the customer had to do to get what they came for. Leave it alone and you end up with a high-depth account that files a lot of support tickets and still hesitates at renewal.

Making the number itself a target has the same shape. Put "raise average depth from 3 to 5 this quarter" on a team's goals and the fastest route is not attaching useful features — it's getting customers to click once on features they've never tried. A round of in-product prompts and checklists lifts the average within a month, and three months later those features have all gone quiet again and the number is back where it started. Meanwhile the renewal forecast was built on the inflated figure.