Don't wait for post-launch metrics to validate an idea. The essential evidence for whether to build something is gathered through direct, face-to-face conversations with users about their problems. This pre-build signal is far more reliable than any data collected after shipping.
The question itself reveals a systemic failure. The real problem isn't the feature, but a lack of upfront validation and product discipline before the feature was ever built. The focus should be on pre-build evidence, not post-build justification.
The single most important conversation is where strategy is set and bets are funded. If product managers are not in that room, they cannot influence strategy and are simply handed a roadmap to execute. This is a clear litmus test for the health of a product organization.
The cost of keeping a feature isn't just server time or bug fixes; it's the innovation you're sacrificing. By forcing the team to name the specific, valuable project they cannot pursue because of this maintenance burden, you make the opportunity cost real and immediate.
Evaluating a feature based on a single customer request is a trap. You must zoom out to the portfolio level to understand its strategic fit, opportunity cost, and financial implications. A feature that makes sense in a vacuum can be absurd in the context of the entire product strategy.
While frameworks can be useful, leaders who demand strict adherence to scoring systems for every decision may be compensating for a lack of genuine product intuition and experience. They substitute a mechanistic process for the difficult work of forming a strategic opinion.
Avoid the paralysis of a "kill list" by managing features through a defined lifecycle. Every feature should be in an active phase: 'Exploring' (testing), 'Exploiting' (scaling), or 'Sunsetting' (managed decline). A feature you can't place in a phase is in limbo and draining resources.
Raw engagement metrics are misleading. A feature with low overall usage might be indispensable for a small but crucial user segment. Understanding qualitative context is more important than just tracking quantitative numbers like daily active users or time-on-page.
Actively killing or investing in a feature has clear outcomes. The most damaging path is perpetual limbo where a feature is left "in pilot" or "gathering data" indefinitely. This passive indecision consumes ongoing maintenance effort and opportunity cost without resolution.