Copilot readiness
Oversharing Copilot could surface, with one score
Microsoft 365 Copilot shows people anything they already have access to. This report finds content shared more widely than it should be, gives one score, and ranks the fixes by how much content each exposes. Nothing is changed.
Reads only. Nothing in SharePoint is changed.
What does it tell you?
- Top fixes: the oversharing to fix first, biggest first, each with the tool that fixes it.
- Score: how the score was reached, check by check, and which checks could not run.
- Public sites and teams, grants to Everyone and Everyone except external users, sites left in organisation-wide search, and apps that can read every site.
- If you ask for the library scan: Anyone and organisation-wide links, and how many files in each library carry no sensitivity label.
How it runs
- Sign in once. Reports › Copilot readiness. Sign in as a SharePoint administrator so every site is read.
- Library scan, if you want it. Also scan every library: sharing links, unlabelled files and lists with their own permissions. Slow: every file is read.
- Run report. The run screen lists each thing it reads as it goes: Microsoft's reports, the directory, and each site. Stop offers Finish, which reads nothing new and builds the report from what was read, or Stop now.
- Read the result. The Done screen gives the headline and anything that could not be read. The workbook goes into History; Save report keeps a copy where you choose. Its About sheet names every source and what came of it.
What do you need?
- The SharePoint administrator role.
- On App authentication, a site you are not an admin of is opened by making you a site collection admin for the run and removing that again at the end (the site-admin rule). Cert authentication needs no grant.
What will it not do?
- The checks follow Microsoft's own Copilot deployment guidance on oversharing (public groups, Everyone grants, broad links, restricted content discovery, sensitivity labels, app access). The weights are ShareMaster's.
- A check that could not run is left out of the score, and the score says how many checks it is based on.
- Which files Microsoft classified as sensitive is only shown inside Microsoft Purview, so the report counts every unlabelled file instead.
- On User authentication, public groups and apps need Microsoft Graph and are left out, and the report says so.
- A read that did not happen is shown as not read, never as an empty result: a site the report could not open keeps its row, with why, and nothing is concluded from it.