Chartered building surveying, property management and lettings for landlords and commercial property owners across Sussex and Surrey. 45 staff on Microsoft 365 Business Standard. A six-week pilot with 8 people, and a rule for when to extend.
Phase 2 covers 28 people: licences £5,410 a year against time released worth about £71,111 a year (each role's estimate scaled by what its pilot users measured; across the pilot, 1.15 times the estimate). The time becomes money only if it goes on chargeable work, faster turnaround or work that is now outsourced.
Estimated from hours a week on email, meetings, documents, spreadsheets and finding information, at assumed shares of time saved and counted like the pilot. The pilot then scaled each estimate by what that role's pilot users measured. 6 people are in phase 3 and 11 not yet. Accounts fell from phase 2 to not yet because its one pilot user measured well under the estimate: a hint to retest with month-end tasks, not a verdict.
Copilot can find and summarise anything a user is allowed to open, so access is tidied first. Items holding anything up are in bold.
| Check | Needed | Status | Owner |
|---|---|---|---|
| Multi-factor sign-in on for every user Copilot makes a stolen password more damaging: it can summarise everything that account can open | Before switch-on | Done | IT provider |
| 'Anyone with the link' sharing reviewed and switched off by default Copilot can surface any file a user can reach, including through old open links | Before switch-on | Done | IT provider |
| Sites and Teams open to 'Everyone' reviewed A site open to all staff is open to every Copilot user | Before switch-on | Done | Operations manager |
| HR, payroll and board files restricted to the right people The most damaging thing Copilot could summarise for the wrong person | Before switch-on | Done | Finance director |
| Leavers' accounts and shared mailboxes reviewed Old accounts keep access nobody is watching | Before switch-on | Done | IT provider |
| Guest access to Teams reviewed Guests should not see internal files through shared channels | Before extending | In progress | Operations manager |
| Acceptable-use policy agreed and signed by pilot users Clear rules before anyone starts | Before switch-on | Done | Managing director |
| Client personal data: data protection assessment recorded Copilot will process client and tenant personal data | Before switch-on | Done | Managing director |
| Old files archived or deleted under the retention policy Out-of-date documents lead to out-of-date answers | Before extending | Not started | Operations manager |
| Sensitivity labels on confidential documents Helps Copilot and people treat confidential files properly | Good practice | Not started | IT provider |
| Phase 2 users have signed the acceptable-use policy Everyone with a licence agrees the rules first | Before extending | Not started | Operations manager |
| Pilot users' baseline time recorded Without a baseline the payback can't be measured | Before switch-on | Done | Operations manager |
| Pilot user | Days used, last 3 weeks | Minutes saved a week, last 3 weeks | Counted hours a month | Value ÷ licence |
|---|---|---|---|---|
| Pilot 1 Directors | 100% | 210 | 6.7 | 31.0× |
| Pilot 2 Chartered surveyors | 93% | 217 | 6.9 | 20.5× |
| Pilot 3 Chartered surveyors | 100% | 187 | 5.9 | 17.6× |
| Pilot 4 Chartered surveyors | 27% | 13 | 0.4 | 1.3× |
| Pilot 5 Property managers | 93% | 167 | 5.3 | 10.5× |
| Pilot 6 Property managers | 100% | 200 | 6.3 | 12.6× |
| Pilot 7 Administrators | 93% | 123 | 3.9 | 5.1× |
| Pilot 8 Accounts | 47% | 33 | 1.1 | 1.8× |
Use and time saved are both measured over the last three weeks, after the learning curve. 2 pilot users used Copilot on fewer days than the rule (Pilot 4 (Chartered surveyors), Pilot 8 (Accounts)): ask why before phase 2. Low use usually means the training or the tasks don't fit the role.
| When | Action | Owner |
|---|---|---|
| November 2026 | Finish the three 'Before extending' items: guest access to Teams, archiving old files, and the policy signed by phase 2 users | Operations manager |
| November 2026 | Talk privately to the two low-use pilot users; adapt the training for site-report work; retest Accounts on month-end tasks before deciding on that role | Operations manager |
| November 2026 | Decide what the released time goes on: more chargeable survey hours and faster report turnaround | Managing director |
| December 2026 | Licence the phase 2 roles (21 more people) before 31 December, if the offer still applies; check the price first | Managing director |
| January 2027 | Role-based training for phase 2, using the pilot's prompt list; weekly log for six weeks | Operations manager |
| March 2027 | Measure phase 2 against the same rule; decide phase 3 (lettings negotiators) | Managing director |
The licence price is sourced (Microsoft UK business plans page, checked 12 October 2026: Copilot Business add-on £16.10 per user per month on annual commitment; £13.80 for the first year on new commitments made 1 July to 31 December 2026). Roles are ranked on assumed shares of time saved; the pilot measures the real saving, and each role's estimate is then scaled by what its pilot users measured (or by the pilot as a whole for roles not in it). With one or two pilot users in a role, that is a hint, not proof. Pilot users report their own minutes saved, so only 50% of it is counted. Value is time released at each role's cost per hour; it becomes cash only if the time is used for something else.
After the pilot review, Ashdown Taylor chose the guided route. In the final working session the agent was set up in an AI assistant the business already uses, connected read-only to the weekly log responses and the Microsoft 365 Copilot usage report (its IT provider confirmed this before it committed). On the first Monday of each month, and every Monday while a six-week log is running, it puts the latest log into a new dated copy of the plan, runs the checks and sends the managing director a short note. It uses the rules on Inputs: use on at least 70% of working days, value at least twice the licence, no more than 1 day a week between the log and the usage report, and 60 days' warning before the offer ends on 31 December. Nothing left Ashdown Taylor, and no decision changed, without one of its own people approving it. These events continue the fictional example.
| Date | What the agent found | What was decided |
|---|---|---|
| 2 Nov 2026 Flag | The first-year offer ends on 31 December, 59 days away: inside the 60 days' warning the managing director set. | The managing director asked the IT provider for a dated quotation for 21 licences, and to confirm the offer applies to licences added to the existing subscription. Phase 2 stayed conditional on the three readiness items. |
| 7 Dec 2026 Flag | The operations manager had marked the last three readiness items Done, with evidence. Extend to phase 2? moved from "Yes, once the 3 items are done" to "Yes: the pilot meets the extend rule". | The managing director licensed the 21 phase 2 users from 14 December on the provider's quotation (£13.80 a user a month for the first year, £16.10 after), and booked the training for January. |
| 18 Jan 2027 Flag | One phase 2 user had logged 6 days used in a week, so Checks showed CHECK. The note said so, named the check and gave no figures. | The user had counted a Saturday. The operations manager corrected it with them, and the note followed the same morning. |
| 8 Feb 2027 Flag | Two phase 2 users, both chartered surveyors, were newly below the 70% use rule over the last three weeks: 7 and 8 days of 15. | The operations manager talked to both: most of their week is site reports, which the January training didn't cover. A site-report session was added on 12 February, using the pilot's prompt list. |
| 15 Feb 2027 Flag | One user's log said 5 days used in a week; the usage report showed 2, more than the 1 day allowed. | The user had filled in the form for the previous week, and corrected it. |
| 22 Feb 2027 No flag | The phase 2 log finished: use on 84% of working days over the last three weeks and counted value 10.6 times the licence. Both users flagged in February were back above 70%. | Nothing needed. The figures went to the March review, as planned, to decide phase 3. |
At the 30-day check-in on 9 November, Geoff and the managing director read the first log together. No rule needed changing.
What the agent added. The method is unchanged: same spreadsheet, same checks, same rules. The agent made sure the phase 2 log was measured every week against the rule set before the pilot, so low use was raised in week 4 instead of at the March review, and the offer deadline was raised two months ahead. The figures above are fictional log readings; the value case above is unchanged and no new £ figure is claimed for the agent.
Fictional and labelled: these events continue the worked example. The same six events are the sample rows in the Agent log of the worked-example spreadsheet. The agent is part of the guided and done-for-you routes; the self-implement kit does not include it.