Tools I actually think with.
Four small, real tools instead of one static page — pick a tab.
Drag ideas from any field onto the matrix. High-impact, low-effort ideas ship first.
Place a few ideas on the matrix to see a suggested shipping order.
RICE = (Reach × Impact × Confidence) ÷ Effort. Move the sliders for any feature and watch the score update.
Start with 1,000 users at Awareness. Adjust each stage's conversion rate and watch the funnel — and the overall conversion — update live.
Five habits from computational biology that translate directly into product thinking.
RiboWin only calls an ORF "translated" if two independent replicates agree.
Require a metric to move and qualitative feedback to agree before rolling a change out fully.
Ribosome offsets are calibrated per-sample, not hardcoded — every dataset gets its own baseline.
A single fixed fraud-score cutoff undershoots some users and overshoots others — calibrate per cohort instead.
Of thousands of genes, only ~100 actually predicted survival outcome.
Instrument broadly, but ruthlessly prioritize roadmap around the handful of signals that actually predict retention or revenue.
Zooms out from single genes to whole biological pathways changing together.
Ten unrelated-looking feature requests are often one underlying job the user is trying to get done.
ChIP-seq peaks below a confidence score are discarded before any gene gets annotated.
Small-sample A/B results and single-user complaints both need a bar to clear before they drive a decision.