The six signals
| Signal | Weight | What it measures |
|---|---|---|
| Demand | 22% | Search trend, community pain frequency, and money already being spent on the problem. |
| Moat | 20% | Defensibility from a data loop, workflow lock-in, regulation, or distribution. |
| Monetisation | 18% | Contract value against willingness to pay and the shape of retention. |
| Feasibility | 16% | How small the first shippable version can be, and how much technical risk it carries. |
| Timing | 14% | Whether something changed recently that makes this possible or urgent now. |
| Open field | 10% | How much room is left — high means incumbents have not covered this properly. |
Open field is deliberately the inverse of crowding: a high score means incumbents have not covered the workflow properly. It carries the smallest weight because an empty market is often empty for a reason.
Why the parts always add up
The composite is calculated as AOS = Σ(signal × weight) / Σ(weight) and nothing else. There is no hidden adjustment layer, no editorial override, and no secondary curve applied after the fact. If you add up the six numbers shown on any idea page with the weights above, you will get the score printed in the ring.
This constraint matters more than it looks. The moment a composite stops matching its visible components, a transparent score becomes a black box with extra steps.
How signals are calibrated
Raw signal values are derived from base rates attached to each axis of an idea — the industry, the buyer, the job, the technical mechanism, the market wedge, and the business model. Left raw, those values cluster tightly around their axis means, which would make almost every idea look the same.
Each signal is therefore mapped onto its own percentile curve across the whole corpus, spreading it over a 21–97 range with a mild S-curve that keeps a realistic centre mass while still populating both tails. A signal score is best read as how this idea compares to the other 117,264, not as an absolute measurement of the world.
Tiers
| Tier | AOS | Ideas | Share |
|---|---|---|---|
| S | 78+ | 1,252 | 1.1% |
| A | 70–77 | 10,884 | 9.3% |
| B | 59–69 | 48,842 | 41.7% |
| C | under 59 | 56,286 | 48.0% |
Thresholds are anchored to percentiles rather than chosen by feel, because the useful question is not “is 71 a good score” but “how does this rank against the alternatives”. The median idea scores 59; the highest in the corpus currently scores 87.
Badges
Badges such as Underserved or Weekend build are computed from the score components, never written by hand. A badge is a readable shorthand for a threshold that has already been met — it cannot claim something the numbers do not support.
What this score does not tell you
- Whether you specifically should build it. Founder-market fit dominates outcomes and is not knowable from an idea.
- Whether the market figures are current. They are order-of-magnitude base rates for comparison, not audited research.
- Whether anyone will buy. Only conversations with buyers answer that, which is why every dossier ends with a step that involves talking to eight of them.
- Whether it is already being built. A high open-field score means we found no obvious incumbent, not that none exists.
- How hard the last 20% is. Every dossier names a hardest part, and it is usually the honest answer to this question.
Reproducibility
The corpus is generated deterministically. Every value that looks random is a hash of the idea’s own structural fingerprint, so regenerating from the same taxonomy produces byte-identical output. Two people reading the same idea page a month apart see the same score, and any change to a score is traceable to a change in the published taxonomy rather than to drift.
- 117,264
- Ideas scored
- 61
- Industries
- 12,000
- Full dossiers
- 40
- Jobs mapped
- 14
- Mechanisms
- 12
- Business models
Questions
How is the AltF Opportunity Score calculated?
What does a good opportunity score look like?
Can I change the weights?
Are the scores predictions?
Where do the market figures come from?
Citing this page
AltF Ideas (2026). How AltF Ideas scores startup ideas. AltFTool. Retrieved from https://www.altftool.com/ideas/learn/scoring-methodology
Corpus snapshot: 117,264 ideas across 61 industries, AOS range 31–87, median 59. Collections currently span 12 computed shortlists.