The renewal meeting needs one page, not a dashboard
Finance, IT and security want different answers from the same Copilot scan. What belongs on each one-pager, what to cut, and why the comparison matters more than the number.
There is a particular way a Copilot renewal goes wrong. Somebody opens a dashboard, shares their screen, and starts explaining what each tile means. Fifteen minutes later the room has learned a lot about the tool and nothing about the decision.
A dashboard is built for the person who owns the problem every day. A renewal meeting is full of people who look at it twice a year. Those are different artefacts, and trying to make one do both jobs is why the meeting runs long and ends without a decision.
Three readers, three questions
The same scan can answer all three. It cannot answer them on the same page.
Finance: what did we spend, and what can we stop spending?
One page. Current monthly and annual spend, the reclaimable portion expressed in money rather than seat counts, and the same two figures from last quarter so the trend is visible without being asked for. If an action was taken since the last review, what it was worth.
Leave out adoption percentages, app breakdowns and anything that invites a methodology debate. If a number needs a definition to be understood, it belongs in an appendix, not on this page.
IT: who needs help, and what do we do on Monday?
A prioritised list, not a chart. Which users are dormant, which are low-usage, which apps are flat, and the specific next action for each group — this cohort gets a reclaim notice, this cohort gets a peer session, this app needs the underlying content looking at before any more training is booked.
This is the only one of the three that benefits from detail, because it is the only one whose reader has to act on individual rows.
Security: what is running, and who owns it?
The agent inventory, filtered hard. Published agents with no owner. Agents grounded in sensitive content. Agents nobody has used since they were built. Three counts and three lists.
Security does not need the licence economics and will not thank you for them. What it needs is the answer to "if an auditor asked what AI is running in this tenant, could we answer?"
What to cut
The hardest edit is removing things that are genuinely interesting.
Cut anything with no decision attached to it. A chart that cannot change what anyone does is decoration, and decoration is what makes a pack feel like it is hiding something.
Cut raw counts where a derived number exists. "412 users" means nothing without a denominator; "a third of assigned seats" means something; "£3,100 a month" means something you can act on.
Cut the methodology from the page and keep it on hand. You want to be able to answer "how is dormant defined?" immediately and precisely — but having the definition on the slide invites the question before anyone has absorbed the number.
The comparison is the argument
A single scan gives you a position. Two give you a direction, and direction is what persuades.
Say the scan comes back at £3,100 a month of dormant Copilot licences. On its own that is a finding, and findings get noted. "We had £4,900 a month in July, we acted on it, and we are at £3,100 now" is a track record, and track records get budget. Same figure, entirely different meeting.
This is the practical reason to scan on a cadence rather than before each renewal. The scan a quarter before the meeting is the one that gives the meeting its evidence — because it is the one you had time to act on.
Make it repeatable or it will not repeat
Any reporting routine that depends on somebody remembering to assemble it in renewal week has already failed; it will be assembled late, from whatever the data looks like that day, with no comparison.
Two things fix that. Put the scan on a schedule so the snapshot history builds itself. And send the summary to the people who will be in the room on a regular digest, so that by the time the meeting happens nobody is seeing the numbers for the first time. Meetings that contain no surprises are short.
One scan, three outputs
Copilot Insights is built around that split: a CIO/CFO executive ROI summary, an IT reclamation and adoption pack, and a security view of agent sprawl and posture — all generated from the same read-only scan, exportable to CSV and PDF, with scheduled email ROI digests and a snapshot history so each report is a comparison rather than a standalone number.
Related: assigned is not adopted, setting a dormancy threshold you can defend, and finding the agents nobody remembers building.
