Why Your Product Roadmap Looks Healthy but Users Keep Churning

Your engineering velocity is up. Quarterly OKRs are green. The roadmap is full and the launch calendar is busy. And yet retention is sliding, support tickets keep rhyming with each other, and the sales team keeps hearing about the same gap from prospects. Something is off, and the roadmap review is not showing it.

The problem is almost never that the team is shipping the wrong number of things. It is that the portfolio is the wrong mix. To see this clearly, stop thinking about features as features. Start thinking about features as medicines.

Every feature is one of five medicines

A useful diagnostic only works if the categories are mutually exclusive and behaviorally distinct. These five are:

• Painkillers. Solve an urgent, sharp pain the user is feeling right now. The user notices the day you ship it. Removal causes immediate complaint.
• Vitamins. Improve long-term health. Nice to have. Removal would not be noticed for weeks. Most enterprise software is ninety percent Vitamins.
• Antibiotics. Treat systemic rot — a class of recurring complaints, a category of bug, a workflow that quietly drives churn. Hard to staff because nobody owns the disease.
• Vaccines. Prevent future problems. Investment today, payoff much later. Routinely deprioritized in favor of the urgent.
• Diagnostics. Reveal what is actually happening — telemetry, dashboards, qualitative loops. Without these you are prescribing in the dark.

The four prescribing patterns

When you classify a roadmap by these five medicines, four archetypes emerge. Each one explains a different version of the roadmap-looks-healthy-but-users-leave problem:

1. Placebo. The roadmap is busy and announcements are frequent, but no medicine is actually treating a real condition. Activity that looks like medicine. The user feels nothing change.
2. Symptomatic. You ship Painkillers fast every time a complaint spikes. The same complaints keep coming back, slightly mutated, because the Antibiotic that would treat the underlying disease is never prioritized.
3. Therapeutic. A balanced portfolio. Painkillers and Vitamins, with at least one active Antibiotic and a real Diagnostic loop. Most healthy product orgs land here.
4. Precision. Every feature is prescribed against a specific signal from Diagnostics, with a clear hypothesis and a kill-switch. This is what separates great products from good ones.

The single most diagnostic question

Pick your last six shipped features. For each one, write down: which medicine was this, and what specific user signal would tell us it worked? If more than half are Vitamins or you cannot name the signal, you are not in Therapeutic. You are in Placebo or Symptomatic, regardless of what the launch deck says.

Most teams find this exercise uncomfortable. That discomfort is the value. The PRX framework formalizes it across twenty-five questions and five dimensions, but the underlying move is the same: stop counting features, start naming medicines.