How validation scores are calculated

How scoring works

Below is the formula, the inputs each check reads, and how confident you should be in each part of a report.

Validate an idea free

An example of how a full report is scored: the idea is read by an AI model in five passes, a fixed formula weights eight dimensions from what it returns, a penalty of 2 comes off, and the score of 62 out of 100 falls in the Pivot band, between 50 and 69.

What is a validation report?

StartupDeckAI turns a short idea description into a structured memo: validation score, competitor landscape notes, market sizing sketches (TAM / SAM / SOM), revenue scenarios, and risk prompts.

The output is a draft for founder diligence, not investment advice, legal guidance, or a guarantee of market success. Each layer has different reliability, combining rules-based scoring with LLM synthesis.

An example validation report filling in: a Pivot verdict with a score of 62 out of 100, then the market, the competitors, the risks and a pitch deck outline, each figure labelled with where it came from.

The Scoring Engine

Every score comes from a fixed formula in code. The weights are the same for every idea and are not adjusted to flatter one. What goes into the formula depends on which check you run.

Quick check

No AI model is involved. Rules read your industry, buyer and description, and score nine dimensions with these weights.

  1. Market potential 18%
  2. Demand signal 16%
  3. Monetization fit 14%
  4. Defensibility (moat) 12%
  5. Buyer clarity 10%
  6. Pricing power 10%
  7. Competition (inverse) 8%
  8. Distribution (inverse) 6%
  9. Execution (inverse) 6%

Full report

An AI model reads your idea in five passes, and a fixed formula scores eight dimensions from what it returns. The formula is ours; most of what it is given is the model’s judgement. Every figure in the report carries a label saying where it came from.

Five example figures from a report, each labelled with its source: found on the web, from you, calculated, a model estimate, or an industry default.

Penalties reduce the final score when the engine finds inconsistent inputs, missing figures, or unit economics that rest on industry benchmarks instead of your numbers.

  1. Structured LLM pass

    Extracts likely direct and indirect competitors from your idea text and industry.

  2. Industry templates

    When evidence is thin, known players for the selected sector may seed the list.

  3. Web research

    The places listed under "Where a report looks" are searched before the model is asked. A competitor is marked web-verified only when its name appears on a page that search found.

  4. Similarity ranking

    Embedding-style matching ranks competitors by semantic overlap with your idea.

Competitor Identification

Competitor names come from a multi-step research pipeline, not a single chat completion. We gather, check and rank the candidates.

You should verify every competitor name, website, and positioning statement before citing them to investors. We surface candidates, not certified market share data.

Where a report looks

Before the model is asked about your idea, the report searches these places. The list is printed from the code that runs the search, so a place that is not searched cannot appear here.

Every page found is listed in the report's Sources section with a link, under the place it came from. A place that returns nothing adds nothing: the report does not fill the gap with a figure. The model reads the company and product pages when it lists competitors. It does not read the threads and stories, which are there for you.

Confidence Levels

How far to trust each part of a report. While market and financial figures are model estimates, a full report is rated Medium at best.

Human Responsibility

Founders remain responsible for customer discovery, legal compliance, financial decisions, and investor communications. StartupDeckAI does not replace accountants, lawyers, or licensed financial advisers.

Start Validating

Contact Support

Explore validation tools

Tools & workflows

Free idea validator Validation tool overview Startup metrics dashboard AI chief of staff MVP roadmap generator Business model generator

Industry validators

ai ml validator marketplace validator proptech validator agtech validator

Comparisons

All tool comparisons Compare vs founderpal Compare vs pitchbob Compare vs hyper Compare vs cerenovus

Guides

ai vs traditional startup validation create investor pitch deck 2026 understanding tam sam som startup validation india guide

Trust & docs

About the team Product changelog Plans and trial limits