The Not-Built List
What I decided not to build, what I built instead, and why. Each one links to the evidence.
-
An AI feature I had already validated
Built insteadReducing friction in the main campaign-creation flow, and only the features that supported adoption.
WhyAI would have parsed the input data into the system, so users would check and correct instead of filling everything in. That would have cut the time and raised the value of the tool. I validated it with a prototype and a time measurement. But users were still struggling with the main flow, and adoption and new clients came first.
-
A score for every grant application
Built insteadFlags with the reasoning and a link to the source. The reviewer decides.
WhyThe managers did not trust the score and did not know what to do with it. They sign the application, and they would not act on a number they could not defend.
-
Writing the applications too
Built insteadChecking only.
WhyIt looked easier to demo and sell. I tested it with mockups and user interviews, and the pain turned out to be that applications got rejected, not that they were hard to write.
-
One score for the game
Built insteadTwo scores: Launch and Judgment. A win needs both.
WhyAn AI tester won 12 of 12 runs by taking shortcuts. A single score was rewarding the opposite of what the game is meant to teach.
-
Hints that gave away the answer
Built insteadContext: the problems ahead, never which card solves which.
WhyThe tester who scored 100 had learned to pick the right cards. Matching a technique to a problem is the actual job.
-
Quoting Reddit verbatim
Built insteadParaphrased voices, and the raw data stays private. Quotes from published studies stay verbatim, with a source.
WhyA verbatim comment can be searched and traced back to a person who never agreed to be in a portfolio piece.
-
A “strong call” for launching everywhere at once
Built insteadWhere patients carry the risk, launching to every clinic at once scores “partly” at best, even with evals.
WhyPlaying the clinic story myself, I launched to all 40 clinics on day one and the game rated it a “strong call”. A call cannot be strong if another plan in the scene is more responsible.