By revenue, your top sector is Financial services: 40% of your revenue from 32% of your deals (48 wins).

Your data

Based on all 150 won deals. Pattern search: 473 of 505 combinations could not be tested, because they had too few won deals or were too rare to judge.

Every figure

Computed in code from your won deals; a share is of the won deals where that field is recorded.

Where you win

Won deals by industry and company size. Based on 150 of 150 won deals with both industry and company size (100%). Stronger shading means more won deals; the highlighted cells are the target profile.

Target profiles

The target profile's median deal and time to close are compared with the rest of your won deals: those outside the profile. The check shuffles which won deals are in the profile 2,000 times and counts how often a gap at least as large turns up by chance. A gap is stated as real only when the chance of a false find is under 1% across all 2 comparisons together (under 0.5% each); otherwise the card says it could be chance. The typical deal and time to close in the tiles are over all your won deals, the profile included.

Patterns

We looked at 505 combinations of two or three company attributes in your won deals (for example a sector together with a company size). 456 had fewer than 5 won deals, too few to say anything about. 11 were no more common than you would expect from each part on its own. 17 were too rare to check fairly against chance (fewer than 5 won deals expected). That left 21 to check against chance: 4 stood out. A combination we could not check is not ruled out: untested is not the same as absent. UK area isn't recorded on any of the 56 won deals located in the UK, so it cannot be used in a combination. The check compares each combination with 500 shuffled copies of your data, and allows a 5% chance of a false find. Target profiles and Worth knowing use a stricter bar, a 1% chance of a false find, so a combination can stand out here without becoming a target profile. Method icp-patterns/3 · input d00ee651c260 · params e258205ddae0.

Qualification rubric

In plain terms: a criterion earns weight when your wins bunch on a few of its values, more than chance alone would give; one spread about evenly earns none. The working: Weights: C = 1 - H / ln K; raw = C x coverage when the wins sit in at most sqrt(K) of a fixed vocabulary's values or C beats an even spread over the observed values (500 seeded draws, alpha 0.05 split over the criteria), else 0; largest remainder to 100 (all 0 when no criterion qualifies: no rubric); geography weighted once (Country if no country holds 80% of wins with a country, else UK region). Method icp-rubric/7. Concentration (1 is a single value, 0 is spread evenly): Sector 0.27, Company size (employees) 0.2, Country 0.18, Tech stack 0.01, Lead source 0.

Written text

Every sentence in this report, the actions, quick win and risks included, is written in code from these figures. No AI model writes any of it.

Method

icp-profile/2 · industry-sectors/2 · regions-uk/1 · input c8e116161107 · params c954fc6fbe06 · icp-visual/3 a0aef9ee8439