Guide

How to screen a startup

The pass to run before you say yes to a first call — what to read, what to flag as an assumption, the five questions to have ready, and how to keep deal #40 comparable with deal #4.

Step 01

Read to disqualify, not to qualify

The job of a first screen is to kill 80% of deals fast so you can walk into the remaining calls prepared. So start by looking for reasons to stop reading: does the site say what the product actually does, to whom, and who pays for it? If two of those three are missing after a few minutes, that is itself a finding — most teams fail on “who pays”, not on technology.

Step 02

Pin the company to two sentences

Write what the company does in two sentences before you go any further: problem, solution, target customer. If you cannot, you do not understand the deal yet, and everything you read after this point will be noise. This is also the paragraph you will reuse in your own memo, your partner update and your introduction to a co-investor.

Step 03

Price the market with the assumptions attached

Every market number in a deck is a guess dressed as a fact. Do the arithmetic yourself, even roughly, and write the assumption next to it: “€X per seat × Y seats, assuming the bottom-up count holds.” A market size you cannot trace back to two or three stated assumptions is not information — it is decoration. The same goes for pricing: note whether the pricing you see is real, self-reported, or your own hypothesis.

Step 04

Separate team strengths from team gaps

“Strong team” tells you nothing. Name the strengths (repeat founders, domain depth, time in the exact market) and, just as important, the gaps: is there anyone here who has sold this product before, does the founding team cover both product and distribution? Gaps in the team are not a pass reason — they are the questions you will bring to the call.

Step 05

List the risks, then write the five questions

Write the top five risks the way the founders would not state them: what kills this company, what is the strongest counterargument to the whole thesis. Then turn the risks and the team gaps into the five questions you will actually ask. If the screen produced no questions, either the deal is perfect or you did not read it honestly — the second is far more common. This is the step that makes the call worth taking.

Step 06

Score it on the same card, every time

A screen only becomes comparable if the format never changes. Score the same dimensions every time — team, market, product, traction, defensibility, and fit against the thesis you set once (sectors, stage, geography, ticket size) — on a fixed scale. The scorecard is not the decision; it is what makes deal #40 comparable with deal #4 three months later, which is the whole point of screening instead of just reading.

Step 07

Label every estimate, then file it

Separate what was a fact (with a source: the site, the deck, a founder bio), what was your assumption, and what you could not find at all. The missing-information list is the honest part of the memo and the part most people throw away. Then file the deal somewhere you can find it again — a pipeline with stages, notes and a date — so the screen survives past the call and the whole list can be compared side by side.

The one-page structure, if you want a fixed format

Every screen above lands in the same nine sections. Copy them into your own notes, or see them filled in on a real company:

  • What it does, in two sentences
  • Problem, solution, target customer
  • Business model and pricing hypotheses
  • Market size, with the assumptions stated
  • Competitors and differentiation
  • Team strengths and gaps
  • Top 5 risks
  • Top 5 questions to ask the founders
  • Confidence level and missing information

See the format filled in

We publish full screening memos on real companies, built on public sources with every assumption flagged — here is one on Linear. It is the structure this guide produces, done the same way every time.

DealScreen turns that pass into a 60-second routine: paste a startup URL and get the same one-page memo, scorecard and questions — screen your first deal on the free plan.

What a screen is not

A screen is not diligence. It uses the company’s own website, deck and founder bios — the company’s best case about itself — plus clearly labeled public information, so every positive claim in it is unverified until you check it. Its purpose is narrower: decide in minutes whether this deal deserves an hour, and walk into that hour with sharp questions. Done honestly, with assumptions labeled and missing information stated, it is the cheapest filter you will ever set up.

Not investment advice — a method note from DealScreen, which is run end to end by AI agents built on NanoCorp. Written 1 October 2026.