Your most-used feature isn't the one that keeps customers

Illustrative example. Rankloop is a fictional company, and the figures are illustrative, not measured results.

Every quarter, Rankloop's product team reviews feature adoption. The review sets the roadmap, the messaging and what goes into each plan. It rests on three questions.

The product analytics tool answers the use question in detail, down to the path each user takes. The other two need what happens before and after the product. Where did users come from, which plan did they pay for, and are they still customers months later? In Funnelsight, the team put each feature next to those outcomes, from the sign-up source to the renewal.

What usage says, and what revenue says

The flagship feature looks like a success, until revenue enters the picture. Rankloop built its pitch around the site audit, in its ads and in onboarding, and 92% of trials run it. On use alone, it works. Conversion tells another story: trials that use the audit convert barely more often than those that don't, at 18% against 15%. Digging further, it gets worse. The third of trials that followed it closely, marking five or more fixes as done, converted at 13%. Those that only looked converted at 21%. Most recommendations were generic, and seven in ten needed a developer the team didn't have. The users who tried hardest had little to show for it after two weeks, and saw no reason to pay.

The import matters, for one moment. Teams that bring their ranking history from their old tool convert more than twice as often. For them, an effortless migration is often why they started a trial at all. Once the history is in, the feature has done its job, and it does nothing for how long they stay. It is essential for one segment, the teams switching tools, and it is used once.

Revenue comes from two features Rankloop never put first. Rank tracking helps both to sell and to keep customers. The monthly report keeps them longest: customers who share it stay 50 months on average, against 15 for those who don't. At $150 a month, that is about $7,500 per customer instead of $2,300. Agencies send it to their own clients every month. But the report needs a month of data, and the trial lasts 14 days, so almost no trial user sees it.

What changed

• The audit gives fewer, better recommendations. Each one now says what it is likely to change and who can do it, a marketer or a developer. Generic items no longer appear.
• Onboarding follows what each customer came for. Agencies are walked to a first client report, built from the ranking history they import. In-house teams are walked to rank tracking, with a weekly summary for their manager.
• The roadmap follows revenue, not usage. Next quarter goes to reports and rank tracking, the two features linked to customers who stay.

The team also plans to test a harder question. Does a full audit belong in a free trial, when users can fix their main issues in two weeks and leave? In this illustrative scenario, trial conversion rose from 18% to 21% the following quarter. Lifetime takes longer to read, so the team will check it in six months.

What to copy

• Write down the job each feature was chosen for, and check it against conversion and lifetime, not usage.
• Separate the features that win the sale from those that keep the customer. They are rarely the same.
• When the feature that keeps customers needs more time than the trial, show it another way: with the user's own history, sample data, a guided tour or a demo.