The “not yet” drawer is a useful piece of product infrastructure. It holds ideas that are interesting but depend on something you have not established.
Putting an idea there is a decision to investigate the missing premise before spending more time on the interface.
The clone with an invisible engine
Be cautious when the reference product depends on proprietary data, a large contributor network, licensed media or a distribution channel you do not have. Rebuilding its screens does not reproduce the business. First identify a narrower useful job and a legitimate source of the required inputs.
The subscription without recurring value
An occasional one-off task may be a good utility and a poor recurring offer. Test the natural return frequency and willingness to pay. Do not force a subscription into the design solely to make a revenue target easier to write.
The unlimited metered service
If each use costs money upstream, an unlimited low-price plan needs careful bounds and measured usage. Model the expensive customer and the provider outage before publishing the promise.
The sensitive expert replacement
A product that claims to replace qualified judgment in a high-stakes domain requires expertise, evaluation and appropriate safeguards. A convincing demo is not enough. A narrower administrative workflow may be a more suitable starting point for a small generalist team.
The idea with no reachable buyer
A problem can be real while the audience is expensive or difficult to reach. Find a credible distribution path and run a small offer test. A broad market-size slide does not tell you how the first customer will hear about you.
For each deferred idea, write the evidence that would move it out of the drawer: an authorized data agreement, a successful pilot, a measured cost profile or a repeatable acquisition path.
The objective is not to become cautious about everything. It is to spend your build energy where the next uncertainty can be resolved by building, rather than where a polished prototype will merely hide it.