Methodology paper
How the Fit Score works
The full InvestorUniverse methodology as one document: what we measure, how much each factor counts, and how evidence is graded.
Contents
1. We profile your company
We read your website and positioning to identify your sector and sub-sector, business model, target customer and geography. Fixed rules make the first reading. A language model then tidies it, and each field it returns is kept only if it is a value we recognise (see how we use AI). Your raise amount sets your likely stage unless you tell us otherwise. Where the evidence is thin, we ask you rather than guess, and we never invent traction, revenue or valuation figures.
2. We score every investor against your raise
Every investor is scored factor by factor. Each factor has a weight, and a factor only counts when we hold the data to judge it. If we do not know an investor's cheque size, cheque size is left out for that investor, so it neither helps nor hurts. The Fit Score is the weighted average of the factors we could judge, out of 100. These are the factors and the weights in use today:
| Factor | What it measures | Weight |
|---|---|---|
| Their own mandate | What the investor has told us themselves in our investor portal: their stages, sectors, cheque range and whether they are investing now. It carries the most weight when we have it, and a raise their own mandate rules out is left off the list. | 30 |
| Stage fit | Does the investor back companies at your stage? An exact match scores in full, a neighbouring stage scores partly, and anything further away is left out. | 25 |
| Sector fit | Your sectors against the investor's. A specialist in your main sector scores highest, then a neighbouring sector or a broad technology investor. Generalists score partly, and higher in sectors where we hold few specialists. Sectors an investor excludes are removed. | 20 |
| Cheque size | Whether their cheques suit your raise. We use the range the investor declared first, then a range observed on their record, then the sizes of the rounds we have tracked them in, then a stated range. Each range has a confidence band, and a less certain range counts for less. | 15 |
| Geography | Where the investor says they invest and where the rounds we track show they have invested. The stronger of the two counts. | 10 |
| Recent activity | How recently and how often they have invested. A round we track counts in full when it is announced and half as much a year later. Where we track no rounds, we use the last investment date on the record. An investor who has confirmed with us that they are actively deploying scores in full. | 10 |
| Relevant portfolio | Companies in your sectors that the investor already backs, from their portfolio and the rounds we track. Recent investments count for more than old ones. | 10 |
| Business model | Your business model (B2B, B2C or marketplace) against the models the investor backs. Compared only where the investor has said. | 8 |
| Lead or follow | When you tell us you need a lead investor, or already have one: whether this investor usually leads or follows. Taken from what they declared, what a dataset states, or at least three rounds we have tracked them in. | 8 |
| Customer type | The customers you sell to against the customer types the investor backs. Compared only where both are stated. | 6 |
| Revenue stage | Your revenue stage against the stages the investor backs. Compared only where both are stated. | 6 |
| Board seat | When you would rather not give up a board seat: whether the investor expects one. Compared only where the investor has said. | 5 |
| Fund cycle | Where the fund is in its life, only where we know the year it was raised. A fund raised this year or last scores in full, and one raised more than five years ago scores low. | 5 |
| Replies to founders | How often founders using our tracker report a reply from this investor. Used only once at least 8 founders have contacted them, and it never scores an investor at zero. | 5 |
| Contact route | Whether we hold a verified route to a named decision-maker. An investor who only takes warm introductions scores low here. | 5 |
| Data freshness | How recently we checked the record. Stale data counts for less. | 5 |
The list is ordered by a score that starts from the fit and treats a gap in the basic facts (stage, sector, cheque size, geography, activity and portfolio) as neutral. That stops an investor we know little about from ranking above one we know well. Investors scoring below 55 are left out. Every match comes with a written rationale built from the same evidence, and the factor-by-factor breakdown is stored with it.
Three things are worked out from the funding rounds we track and are never entered by hand: how active an investor has been, the sizes of rounds they join, and whether they tend to lead. The same rounds tell us who has backed a company a second time and who invests alongside whom. Fund size, fund vintage, board seats, decision time and similar details are shown only where the investor or a dataset states them. We never infer them.
Some things are raised as warnings and do not change a score: an investor who, in the last 24 months, backed a company we class as your direct competitor (the same sector, and either the same sub-sector or a description that reads as the same business), and an investor who already holds several recent companies in your sector.
The investors you save and the ones you dismiss re-order your own list towards investors like the ones you saved. That can move an investor by at most 6 points, on your list only. It never changes the score anyone else sees.
3. Hard exclusions
- Incompatible stage, cheque size or geography
- Sectors the investor explicitly excludes
- Closed funds, firms we have confirmed are no longer operating, and investors who've told us they're not deploying
- People who have left a firm, and anyone who has asked to be removed
- Duplicate funds. Where the same person appears at two firms, someone on our team reviews it before either record changes
- Investor types and criteria you choose to exclude, and investors you dismiss, on your own list
4. Fit Score and confidence are separate
Fit measures how well an investor matches your raise. Confidence measures how strongly the evidence supports what we know about that investor: where the firm carries graded evidence, confidence is the share of the scoring weight we could judge, scaled by the quality of that evidence; where it does not yet, it is the share we could judge, reduced when the record has not been checked recently. A high fit with lower confidence is still worth a look; we show both so you can prioritise sensibly. Neither figure is a probability of investment, and we never present one as such.
Evidence tiers
Every fact we hold carries a source tier and a verification state. The tiers, from strongest to weakest:
- A registry or regulatory filing · weight 1
- B the investor's own official source · weight 0.9
- C primary announcement or direct statement · weight 0.7
- D reputable structured data or media · weight 0.5
- E secondary or unverified source · weight 0.25
- F inference or model output · weight 0.1
A tier F fact (an inference or model output) can only ever support an inferred or estimated state; it can never make a field verified. Specificity is weighted separately: a source that states the fact directly counts more than one that implies it. The matching model in use is version 2.1.2; that version is stamped on every list it builds, so a ranking can always be read against the model that made it.
5. What "verified" means here
Nothing is shown on this site unless it has passed a check we can rerun, and our database refuses to publish a record without one. A record is verified in one of four ways. Official record: it comes straight from a Companies House filing or an Innovate UK award. Source confirmed: we fetched the published announcement and it names the company and states the figure in pounds; investor names, the stage and our one-line description are shown only if that source states them too. Confirmed by the company or investor: they submitted or claimed the record themselves. Checked by us: a person on our team read the source and signed it off, and their sign-off is logged.
Investor firms are published when their own website answers and names them, and a contact is published when their work email belongs to that firm's domain. Funding rounds and news are re-read every 30 days, contacts every 90 and firms every 180; a record that fails twice in a row comes down until it passes again. Verification confirms that we report what the source says. It cannot confirm that the source itself is right, which is why every figure links to where it came from and every page has a correction link. Sector and region groupings are our own classification.
6. Where the data comes from, and how it stays fresh
Investor records combine public professional sources, firms' own websites and portfolio pages, Companies House filings, press coverage, licensed B2B data providers, and what investors tell us directly in our portal. Every fact stores its source category, collection date and last verification date. Email bounces, role changes we see when a source is re-read and corrections update the record.
Anyone can correct or remove their data at any time through our data request form. Removals take effect immediately and stay in force even if the same details reappear in another source. See our Trust Centre for more.
7. Limitations
A Fit Score is a prioritisation tool. It can't tell you whether an investor will invest, and nothing on InvestorUniverse is investment advice. Activity data lags reality, so an investor shown with no recent investments on record may simply be quiet in public.
8. How the desk calculates step-ups, percentiles and trends
Step-up is a round's indicative post-money valuation divided by the previous round's for the same company, when both are shown. Percentile compares a round's valuation with other rounds at the same stage and sector in the last 12 months, and only appears when there are at least five. Trends compare the latest period with the one before it of equal length. Rounds above £500m are held for manual review before they appear. Rounds filed in several tranches are listed on each round's page as related filings, so add them up for the full round.
9. How we track funding rounds
Most rounds reach us through Companies House. When a UK company issues new shares it has to file form SH01, which lists the number of shares, the price paid for each and the total shares in issue afterwards. We read these filings as they're published. The amount raised is the new shares multiplied by the price paid. The estimated valuation is that price multiplied by every share in issue after the round.
We only report priced rounds of £150k or more where shares were sold at a meaningful premium to their nominal value, which filters out founder and employee share issues. We skip holding companies and fund vehicles. Valuations are estimates: they leave out unissued options, convertible notes and SAFEs, and a round closed in tranches can appear as more than one filing. Investors aren't named on the SH01, so we add them when they're announced or confirmed by the company. Innovate UK grants come from UKRI's Gateway to Research. Every round links to its source.
End of the methodology for matching model 2.1.2, dated 5 October 2026. Fit and confidence are never probabilities of investment. Nothing in this document is investment advice.