Managed IT, cyber security and cloud services for accountants, solicitors, surveyors and other professional firms across the East Midlands. 32 staff. Customers we most want to win: professional firms with 20 to 100 staff.
Verdict: Promising but not proven: the difference is within the margin of error; keep recording. Proposals with a case study were won 22 points more often, but with 16 and 27 proposals decided the margin of error is 24 points, so none of it can be counted yet. Nor is it a controlled test: salespeople may add a case study when a deal already looks good. From now on, send one with every relevant proposal and record it; in a year the answer will be much clearer.
| # | Customer and result | Evidence | Fit | Story | Willing | Score |
|---|---|---|---|---|---|---|
| 1 | Thistleby & Moss Chartered Accountants Accountancy · Unplanned downtime from 31 hours a year to 2; support tickets down 38% | Our records | High | Strong | Yes | 94 |
| 2 | Quenby Legal LLP Legal · Passed Cyber Essentials Plus first time; phishing test clicks from 14% to 3% | Customer's own figures | High | Strong | Maybe | 92 |
| 3 | Hartington Financial Planning Financial advice · Regulator's review with no IT findings | Customer's own figures | High | Strong | Not asked | 90 |
| 4 | Pelham Court Property Management Property · Cyber insurance premium down 12% after the review | Customer's own figures | Medium | Strong | Yes | 88 |
| 5 | Corbridge & Kirk Solicitors Legal · Satisfaction 4.8 out of 5 | Our records | High | Weak | Yes | 74 |
| 6 | Pellingham Accountants Accountancy · Moved over in three weeks with no data lost | Our records | High | Some | Yes | 74 |
| 7 | Ravelstone Surveyors Surveying · Reports issued about two days sooner | Customer's estimate | High | Some | Yes | 67 |
Shaded = selected. Scored out of 100: results you can prove 30, fit with the customers you want 25, story 20, willing to be named 15, recent 10. Left out because they said no: Holmewood Precision Engineering.
Thistleby & Moss ran its practice software on a nine-year-old server in a cupboard. In the busiest weeks before the January tax deadline it failed twice, and staff lost most of a working day each time. Remote working was slow, and every new starter meant a visit from an engineer.
Kestrelgate moved the practice's email, files and practice software to Microsoft 365 and a hosted server over one weekend, set up secure remote access, and took over day-to-day support with a fixed monthly fee.
“We used to dread January. This year nobody mentioned the IT, which is the highest praise I can give it.”Practice manager, Thistleby & Moss (name withheld in this example)
Every result above is from our records, was checked by the customer and approved in writing on 24 Sep 2026; the final text and the quotation were approved on 28 Sep 2026.
| Case study | Owner | Stage | Approval due | Claims publishable |
|---|---|---|---|---|
| Thistleby & Moss Chartered Accountants | Marketing manager | Approved | 26 Sep 2026 | 3 of 3 |
| Quenby Legal LLP | Sales director | Interview booked | 20 Nov 2026 | 0 of 2 |
| Hartington Financial Planning | Sales director | Not started | 18 Dec 2026 | 0 of 1 |
| Case study | Claim | Source | Checked | Approved | Publishable |
|---|---|---|---|---|---|
| Thistleby & Moss Chartered Accountants | Unplanned downtime cut from 31 hours a year to 2 | Our records | Yes | 24 Sep 2026 | Yes |
| Thistleby & Moss Chartered Accountants | IT support tickets down 38% in the first year | Our records | Yes | 24 Sep 2026 | Yes |
| Thistleby & Moss Chartered Accountants | 45 staff moved to Microsoft 365 over one weekend | Our records | Yes | 24 Sep 2026 | Yes |
| Quenby Legal LLP | Passed Cyber Essentials Plus at the first attempt | Customer's own figures | No | – | Not yet |
| Quenby Legal LLP | Staff clicking test phishing emails fell from 14% to 3% | Customer's own figures | No | – | Not yet |
| Hartington Financial Planning | Regulator's review with no IT findings | Customer's own figures | No | – | Not yet |
Fictional, continuing the example: Kestrelgate IT Services and every customer and figure here are invented, and customers' people are named by role. The counts were tallied from the story, not by rerunning the workbook; seven of the agent's log rows are on the Agent sheet of the worked-example workbook.
Kestrelgate chose the guided route. At the final session the case study agent was set up in an AI assistant the business already uses, with its IT's agreement: connected read-only to the CRM (completed projects, proposals and their attachments) and the job and helpdesk systems, and writing only to the workbook and one shared drafts document. Before it ran, Kestrelgate updated its privacy notice to cover what it records about customers' results and contacts. The two agent settings are on Inputs: a newly spotted customer is put forward when it scores at least 60 (B27), and a published case study is flagged for refreshing after 24 months (B28). The run details are on the Agent sheet: the sales director checks every candidate it puts forward, and the managing director approves publishing. Every request was sent by a person, every claim checked and approved in writing by the customer, and nothing was published without the named customer's written approval recorded on Production. Nothing left the business, and no decision changed, without one of its own people approving it. Its first run was on Monday 26 October; at the check-in on Wednesday 25 November, 30 days later, Geoff read the log with the managing director, and his upkeep of the set-up ended there.
| Date | What the agent prepared or flagged | Approved or decided by | What happened |
|---|---|---|---|
| 26 Oct 2026 Prepared | Hartington Financial Planning Selected but not yet asked: drafted the request, saying what we would claim, that they approve every word in writing first, and that an anonymous version is fine. | Sales director | Sent his own version on 27 October; Hartington agreed to be named on 3 November. |
| 2 Nov 2026 Flag | Quenby Legal LLP Draft due, two claims not yet checked by the customer, approval due 20 November. | Sales director | Draft and claims sent on 4 November; all approved in writing on 18 November. |
| 9 Nov 2026 Prepared | Ravelstone Surveyors Firm figures found: report turnaround from 6.5 to 4.6 working days in the job system. Score 67 to 79. | Sales director | Evidence updated on Candidates; fifth, so it waits for the next round. |
| 16 Nov 2026 Flag | Proposal: accountancy practice, 40 staff About to go without a case study, though Thistleby & Moss fits. | Sales director | Attached before sending. |
| 23 Nov 2026 Prepared | Proposals log Five proposals sent since 12 October logged from the CRM, with or without a case study. | Sales director | Checked against the CRM. |
| 30 Nov 2026 Prepared | Quenby Legal LLP Ready to publish, with the customer’s written approval of every claim, the text, the quotation and the name recorded. | Managing director | Said go; published on 2 December. |
| 7 Dec 2026 Flag | Hartington Financial Planning Approval due 18 December, but its one claim isn’t checked: the review letter hasn’t been seen. | Sales director | Asked for the letter; approval moved to 15 January. |
The win-rate comparison is still promising but not proven: five more proposals don't narrow a margin of error of 24 points, and it is read again each quarter. What the agent added is that the routine held: the second case study reached the customer's approval on time, a fitting case study went with a proposal that would have gone without, and the log of proposals keeps growing without anyone re-keying it. Reading the weekly note took the sales director about 10 minutes (estimate). The months after add nothing to the figures above.
Candidates are scored on proof, fit with the customers you want, story, willingness and recency, with the points on Inputs; the top 3 are selected. A claim is publishable only if it comes from the customer's own figures or your records, the customer has checked it, and they have approved it in writing. Win rates compare proposals sent with and without a relevant case study; they are not a controlled test, because a salesperson may include a case study when a deal already looks good. The uplift counted is the difference less its two-sided margin of error at 90% confidence, applied to the 30 proposals a year that go out without one at £9,000 of gross profit each. The cost is your team's time and design costs, not the customer's time.