The Deckwise Method

The Deckwise Method is a framework for evaluating startups on three separate questions: is it true, is it a good company, and is it a good fit. Every claim is checked against public evidence before it counts toward a score.

Most startup evaluation collapses three different questions into one gut call. A deck looks polished, the founder is convincing, the market sounds huge, and a committee says yes or no on a blur of impressions. Then a number turns out to be inflated, or the one you passed on raises from someone else six months later.

The Deckwise Method keeps the three questions apart, answers each with evidence, and only then combines them into one comparable score and a shortlist you can defend line by line.

The three questions

1. The Facts — "Is it true?"

Before anything else: do the claims in the dossier hold up? Traction numbers, market size, the team's track record, patents, funding history, corporate standing.

Every material claim is pulled out of the deck and checked. Official registers first, then sourced web search for the rest. Each claim gets one of four verdicts:

• Verified — multiple public sources agree.
• Qualified — sources partly agree, or the claim is true with caveats.
• Contradicted — the public record disagrees with the claim.
• To confirm — no public source can settle it. It is shown to you, never used to inflate or deflate the score.

Verified claims lift the reliability of a dossier. Contradicted ones sink it. This is the layer nobody else runs, and it is the reason a Deckwise score means something different from a polished-deck score.

2. The Company — "Is this a good company, objectively?"

Independent of whether it fits you. Traction, management, market, product and moat, financials, team, forecast, risk.

This layer mixes hard data (registers, patents, headcount) with sourced signals. When private companies keep the real numbers hidden, indirect signals stand in: hiring velocity, headcount growth, web traffic, analyst reads. Real momentum and genuine defensibility lift the score. Thin traction or an unclear moat pull it down.

3. The Fit — "Is it a good company for us?"

A great company can still be the wrong one for your fund or program. The Fit layer scores against your thesis, your stage and geography focus, your values, and your explicit anti-thesis.

You enter your thesis once. Every candidate is matched against what the evidence actually shows, not against what the deck claims. The anti-thesis check catches the ones that look strong but cross a line you drew on purpose.

The verification backbone

The Facts layer is not a vibe. It runs against real public sources, official registers first because they are authoritative, then sourced web search for what registers cannot answer.

Registers and authorities used include SIRENE, INPI and BODACC in France, SEC EDGAR, USPTO and Secretary of State records in the United States, Companies House in the UK, and GLEIF across the EU.

One rule keeps it honest: a register that only matches a company name proves an entity by that name exists, not that a specific claim is true. Deckwise does not over-read a name match. A claim is only marked Verified when the evidence actually supports the claim, not just the existence of the company.

From three scores to one decision

The three questions produce one score on the same basis for every candidate, so a stack of 300 dossiers becomes comparable instead of a pile of impressions. Candidates land in ranked tiers.

Two safeguards matter here:

• The wildcard. A candidate that misses the top tier overall but spikes on a single dimension is flagged, so a standout in one area never gets buried by an average total.
• Red flags. A hard negative signal pulls the overall score down on purpose, so a contradiction or a risk finding cannot hide behind a strong average.

The weighting between the three questions is set by sensible defaults and you can adjust it per program type. An early-stage accelerator and a growth-stage fund do not weigh the same things, and the method bends to that.

The human decides

Deckwise is a co-pilot, not a judge. It surfaces the evidence, flags what does not hold up, and hands you a ranked shortlist with the reasoning attached. The decision stays with your committee.

That is how juries and investment committees already work, and it is the right posture when you are evaluating people. Under the EU AI Act, evaluating candidates is high-risk and calls for human oversight. Deckwise is built for that: it makes your judgment faster and more defensible, it does not replace it.

How it shows up in the product

The same three questions run across all three Deckwise pillars:

• Source finds startups worth looking at.
• Select scores and ranks a batch of applications on the same basis.
• Verify goes deep on the few that matter, claim by claim, into a sourced memo your committee can read.

The score you see in the app is calculated with the method on this page. The content and the product tell the same story on purpose.

Learn the full framework in the Deckwise Playbook.