Methodology
How we calculate our headline statistics.
Last reviewed: 27 May 2026
Numbers shape decisions, so we put ours where you can check them. This page explains the basis for every headline statistic on the Skim website, including the source and the limits of each figure. Where a number is internal to Skim, we say so. Where the underlying benchmark is small or evolving, we say that too — better an honest small number than an inflated large one.
£3bn+ in public sector contracts won
Cumulative value of UK and EU public sector contracts won by the senior bid team and advisors behind Skim across their careers, measured in nominal GBP at the time of award. Includes wins delivered as in-house bid leads, as external consultants under Shipley and adjacent firms, and as Skim advisors. Compiled from internal records of contract awards and framework appointments. Does not include sub-contracted value or speculative pipeline.
14 vs 2 — contract recall against Claude Opus 4.5
Internal test, April 2026. Single-company case study.
- The task: retrieve every UK and EU public sector contract awarded to one specific named supplier over the previous twenty years.
- The ground truth: 14 contracts. Compiled by hand from Find a Tender Service, Contracts Finder, and TED award notices.
- Skim’s answer: 14. Complete match against the hand-compiled list.
- Claude Opus 4.5 (no tools, no web search), same prompt: 2 contracts.
This is a single-case benchmark on a recall task that depends on the model having ingested twenty years of award data. Skim has; general-purpose LLMs have not. The benchmark does not generalise to every question type, every company, or every model — and we will publish a broader benchmark suite when it is ready to defend in public. The reason we publish the single case now is that this is the gap our customers most often want verified before buying. The number on our website is the number we actually ran.
3 minutes vs 40 minutes — bid/no-bid decision time
Customer-reported figures. Three minutes is the median time Skim customers report from uploading a tender pack to receiving a bid/no-bid decision with full reasoning. Forty minutes is the median self-reported pre-Skim time for the same task, captured in onboarding interviews. Both numbers are self-report and will be replaced with platform-instrumented measurement when the in-product event stream is hardened.
10× faster retrieval of tender files and supporting documents
Customer-reported. Compares the time to locate and surface a relevant past response, case study, or accreditation from a customer’s bid library using Skim search versus the customer’s previous workflow (typically folder-based file navigation or generic search). The 10× figure is conservative across the customers who have reported numbers; individual results vary by library size.
30 minutes per week on opportunity research
Customer-reported. Median time Skim customers spend reviewing the agent’s shortlist of new opportunities each week. The pre-Skim baseline of roughly one hour per day comes from the same onboarding interviews. Self-reported; expect platform-instrumented numbers in a future update.
847 UK contracting authorities
The 847 UK contracting authorities that, between them, account for approximately 80% of SME-addressable public sector spend. Compiled from published award notices on Find a Tender Service and Contracts Finder covering the four years to April 2026, filtered to contracts under the SME threshold (typically < £10m), grouped by procuring authority, and ranked by award volume. The remaining 46,000+ figure is the total population of UK contracting authorities known to Skim, including parish councils, NHS trusts, academy chains, and arms-length bodies.
10,000+ public-sector portals
Count of distinct UK and EU public-sector procurement portals Skim reads, including Find a Tender Service, Contracts Finder, TED (Tenders Electronic Daily), CCS frameworks, NHS frameworks (NHS SBS, NOE CPC, etc.), YPO, G-Cloud, regional consortia (e.g. ESPO, CBC), and the national equivalents across EU member states. Includes both primary procurement portals and significant aggregator/intermediary platforms. Refreshed quarterly.
12,000+ prime/sub partnership relationships
Count of distinct prime-contractor / subcontractor pairings identified from UK and EU public sector consortium award notices over the twenty-year period to April 2026. A relationship counts once per unique pairing, regardless of how many contracts it appears on. Source: aggregated UK and EU award notices published on Find a Tender Service, TED, and successor registries.
20 years of UK and EU award data
Skim’s award notice corpus runs from April 2006 to the most recent published notice. Sources: OJEU/TED (1 April 2006 onwards), Contracts Finder (March 2015 onwards), Find a Tender Service (January 2021 onwards), and the regional/sector portals referenced above. Coverage is most complete from 2010 onwards as digital publishing standardised across EU member states.
70%+ of SME-addressable spend won via formal procurement
Industry estimate of the share of UK SME-addressable public sector revenue that flows through formal tender processes (as opposed to referral, framework call-off without competition, or direct award). Derived from UK government published procurement statistics, National Audit Office reporting, and Crown Commercial Service published data. Broadly consistent with figures in UK government procurement policy statements. The figure is most useful as a reminder of how much pipeline sits outside the network of relationships an SME can build alone.
Tony Birch — credentials
Tony Birch is an advisor to Skim. Forty years’ experience in UK and EU public sector bidding. Developed the APMP (Association of Proposal Management Professionals) certification programme. Former Chairman, Shipley UK. Has personally led or contributed to over £20bn in UK and EU public sector contract wins across his career, including consortium work for primes and direct wins as a bid lead. Skim’s methodology is built on his playbook.
How to flag an inaccuracy
If you believe a number on this site is wrong, misleading, or out of date, email justin@justskim.ai and we will investigate. Where we update a figure, we record the change in this page’s revision history.