The Feature Factory With a Sales-Promise Backlog
At a 900-person B2B SaaS company, nine product teams deliver about 28 new features per quarter. Product management is organized by sales region, and the top three enterprise deals can insert must-have custom requests into the roadmap with two weeks' notice. Usage data shows 19 of the last 40 features had less than 5% customer adoption after six months. The customer success team maintains a 300-item request list; the average age of open requests is 14 months. Engineering utilization is 92%, and the CTO reports rising defect escape rates. The new chief product officer believes the company is optimizing output rather than business value. The VP of sales argues that custom features are the reason the company wins enterprise deals and that any reduction will hit the current quarter's €30 million target. The CEO wants a value-based roadmap but has not defined value beyond revenue. The CPO asks you to help design a six-month transition that does not derail sales.
1. Diagnose how the operating model turns requests into output rather than value. Identify three specific value leaks.
Read the worked answer
Sales regions own product management, and the top three deals can insert custom requests with two weeks' notice. That makes the roadmap a commitment device for sales, not a hypothesis about value. Product teams have no protected discovery capacity; utilization at 92% means there is no slack to learn. The 300-item request list with a 14-month average age is a queue that hides prioritization: oldest requests are not necessarily most valuable. Adoption data shows 19 of 40 features below 5% after six months, a direct signal of output without validation. Rising defect escape rates suggest speed is being bought with rework.
Three leaks: first, build-and-abandon features that consume engineering but produce little adoption; second, queue aging and sales escalation, which creates rework and damages trust; third, custom-feature maintenance and defect escapes, which reduce capacity for core value. The root cause is not that sales is bad; it is the absence of a decision model that separates committed enterprise obligations from discovery bets. I would quantify the maintenance load and the adoption curve before proposing changes.
2. Design a six-month transition that protects the €30 million quarter. Include decision rules for custom enterprise requests.
Read the worked answer
Months 1 to 2: create a product council with sales, product, engineering, and customer success. Classify requests into three lanes: contractual or regulatory commitments, high-value configured extensions, and discovery bets. Reserve explicit capacity-for example 20% for committed custom work and 10% for discovery-rather than letting sales insert into 100% of capacity. Publish a service-level expectation: standard requests reviewed weekly; deal-critical requests need a named sponsor and a trade-off, meaning something comes out.
Months 2 to 4: run discovery on the top three enterprise opportunities and the top three adoption failures. Define outcomes and leading indicators. Give sales a value playbook: what problems the product solves, what is configurable, and what is custom. Pilot with two top accounts.
Months 4 to 6: review adoption and revenue expansion; kill two low-adoption features; reallocate capacity. Do not promise a value-based roadmap without changing the insertion right. Protect the quarter by ring-fencing existing commitments, not by refusing all custom work.
3. What metrics should the CPO put in front of the CEO to show value is improving, and what should be deliberately excluded?
Read the worked answer
Use a small balanced set: customer adoption and time-to-value for new capabilities; retention and expansion revenue in targeted segments; outcome attainment against stated hypotheses; request cycle time and age of open requests; change-failure and defect escape rate; and capacity allocation across committed, custom, and discovery work. These show whether the system is producing valued outcomes and whether the queue is becoming healthier.
Exclude feature count, story points, velocity comparisons, and utilization as value measures. They measure output or busyness and will push teams back to the feature factory. Be careful with revenue attribution: sales will claim custom features drive the €30 million target, so use cohort or win-loss evidence, not correlation. Present trends with baselines and confidence ranges. If adoption is below 5% for many features, the CEO should see that as a portfolio signal, not a team failure. The test is whether the metric changes a decision.