Evidence and judgment. Never predictions.

Every assessment follows one written template: a verdict, a market reality check, feature scrutiny, a pre-mortem, fundability, gaps, assumptions and open inputs, verified research, a 90-day plan and a 24-month roadmap with gates. Research is done and reviewed by a person. AI tools may help search and draft, but nothing is published without human review, and no assessment is an automated scan.

The 38 assessments here contain 495 research pages and 1,225 cited links.

Primary sources first: pricing pages, product documentation, regulators and official announcements, then app-marketplace listings, review sites and public founder interviews. Each claim links to the page it came from. Sources that terms of service forbid us to collect are read, not scraped. Where research is not yet verified, the page says so and lists no sources.

• Verified fact: supported by a current source we checked.
• Customer evidence: what buyers said or did, with the date and context.
• Estimate: a model input or a derived range, shown with its math.
• Judgment: an analyst's conclusion from the evidence. Odds and verdicts are judgment.
• Unknown: something that matters but could not be verified. Unknowns are listed, not hidden.

Each idea is judged twice: as described (the idea as first pitched) and narrowed (the wedge that could work). Verdicts are expressed as Continue, Narrow, Pivot or Stop, always in words, never by colour alone. 37 of 38 ideas are Stop as first described; most then narrow to a smaller, specific version.

Each assessment estimates the odds of reaching $1k, $3k, $5k, $10k, $20k and $50k MRR by month 24, at about 15 hours a week and full-time, assuming the founder conditions are met. There is no verified base rate for small software businesses, so these are reasoned estimates, not statistical probabilities, and never a guarantee. Read “≈ 15 in 100” as: in 100 similar attempts, about 15 get there.

Between 15 and 40 hours a week, odds are interpolated linearly between the two estimates. Each missing founder condition multiplies the odds by the factor the assessment states; compounding several missing conditions is an assumption.

Wherever ideas are ranked, the order is the assessment's judgment odds of reaching the target you chose, at the hours you chose, by month 24. Ties are broken alphabetically. “Months at plan pace” uses each plan's own price and churn, and the number of new customers a month that hits its month-24 checkpoint. Nothing is ranked by payment, popularity or a hidden score.

The five answers never leave your browser; they live only in the page link. Ideas are ordered by four rules, in this order: ideas in the areas you picked come first; then ideas needing fewer founder strengths you don't have yet (building, selling, industry access, derived as described under catalog filters); then ideas sold the way you prefer; then the assessment's odds of your income goal by month 24 at your hours. There are no hidden weights.

Odds are estimated at ~15 hrs/week and full-time only. For 25 hours we interpolate between the two; below 15 hours we show the 15-hour figure and say it is probably too high. “One to skip” is an idea in your areas that its assessment says to stop, or else the one with the lowest odds.

Every research finding carries the analyst's own confidence label. We turn its leading words into one of four tags, and never guess:

• Sourced: “high” or “verified”, checked against a source we read.
• Inferred: “medium”, an estimate, or a vendor's own claim.
• Unproven: “low” or “unverified”.
• Not rated: the research gave no confidence for that finding.

Every assessment shows when it was last reviewed. Reviews expire after 90 days; the page shows the date the next review is due.

For context next to our odds we show a public base rate: an analysis of 10,150 products listing verified revenue on TrustMRR (pulled 4 Sept 2026) found a median of about $30 a month, with about 5% above $8k. That group is self-selected and skewed to indie products, so it is context, not an input. Our odds are judgment for one narrowed idea with its founder conditions met, which is why they sit well above the all-comers rate. When a condition is missing, the what-if model lowers the odds, and the base rate is a reminder of where unprepared attempts land.

• Customer type: the buyer recorded for each idea.
• Founder skill: from the founder-condition titles: access, partner or trade wording → Industry access; selling or calls → Sales; technical, API or platform-app wording → Technical.
• Distribution path: an app-marketplace platform → Platform marketplace; an agency buyer → Agency partners; otherwise Direct outreach & content.
• Evidence maturity: cited sources: 40+ Strong, 20–39 Moderate, 1–19 Thin, 0 Unverified.
• Time: handled by the hours you choose (about 15 a week, or full-time), because each assessment estimates both.
• Capital requirement: not shown yet. It isn't recorded consistently, and we won't guess it.

Narrowing picks one buyer, one region or platform, and one urgent job, usually one with a deadline, a required record or a marketplace where buyers already look. The narrowed version gets its own odds, conditions and gates. If no narrowing is credible, the verdict stays Stop.

Every assessment shows its last-reviewed date. Assessments are revisited when material evidence changes, such as a competitor launch, a price change or a regulation date. If you find a factual error, tell us and we will correct the page and note the change. Corrections never quietly upgrade a verdict.

• Your skills, network, runway and persistence, beyond the conditions you tick.
• Private competitor numbers, unpublished roadmaps and future platform changes.
• Whether a specific buyer will pay you; only selling answers that.
• Product–market fit. No report can guarantee it. An assessment can only make the next decision cheaper and the evidence that should change it explicit.