Skip to content

Copilot Adoption User Guide

Jump to bottom
sambetts edited this page Aug 23, 2026 · 4 revisions

Copilot Adoption – User Guide

How to use the Copilot Adoption area of the admin site to decide who should keep a Microsoft 365 Copilot licence, who should get one, and where to spend your enablement effort.

This is the practical guide. If you want to know why the tool is designed the way it is, or you are adding a feature to it, read Copilot Adoption Tool instead — that page is the specification.

Contents

What this tool is for

Two questions, and deliberately no others:

  1. Who is paying for a Copilot licence they are not using?
  2. Who would benefit from a licence they do not have?

Everything on the page serves one of those, or justifies a number that does.

It is not a general Copilot usage report. If you want prompt volumes, key phrases, sentiment or agent chatter for their own sake, that lives in Reports → Copilot.

It does not rank named individuals against each other. Every per-user list exists to judge whether a seat is well placed, which is why the default sort is least-engaged first.

Before you start

The tool needs data. Check Health if anything looks empty.

Import What you lose without it
User metadata (licences, department, job title, manager, account status) Everything. Without licence assignments there is no licensed population and no denominator
Copilot audit log (Audit.General → Copilot) Per-user engagement, unlicensed Copilot Chat use, per-app breakdown, agents, and the weekly trends. This is the important one
Microsoft Copilot usage report A fallback for per-user activity if the audit import is unavailable. Covers licensed users only
Microsoft 365 usage reports (Teams / Outlook / SharePoint / OneDrive) The ability to find heavy Microsoft 365 users who have never had a Copilot seat

If an import is missing the page still works — it tells you what it could not calculate rather than showing a silently wrong number.

Expect a delay. Copilot events reach the Activity API roughly 2–3 days after they happen, and Microsoft's usage reports are published a few days in arrears. "Yesterday" is never complete.

The first run of each period is slow. The analysis reads a lot of audit history and is then cached, so the first load after choosing a new period takes noticeably longer than the second.

Your first fifteen minutes

  1. Open Copilot Adoption. Leave the period on Last 28 days — it matches Microsoft's own reporting month, so your numbers will line up with theirs.
  2. Read the red and amber headline cards. Reclaimable seats is the money. Habitual users is whether the programme is working.
  3. Look at the two gauges. If adoption is high but habit is low, everybody opened Copilot once and stopped — a very common and very expensive pattern.
  4. Scroll to the adoption funnel and find the biggest single drop. That is where your effort goes.
  5. Open the Enablement plan card. It tells you how many people need each kind of intervention.
  6. Go to Licensed users, sort by Least engaged first, and export the CSV. That is your reclaim conversation.
  7. Go to Licence opportunities, tick Recommended only, and export. That is your purchase or reallocation case.

Anywhere a number surprises you, click the small ⓘ next to it. It tells you what the number means, how it was calculated, the formula, and which import it came from.

The tabs

Overview

The executive view. In reading order:

  • Headline cards — seats, adoption, habit, reclaimable seats, engagement, unlicensed use, recommended candidates.
  • Where you stand — the adoption and habit rates as gauges against a judgement scale.
  • Adoption funnel — licensed → ever used → active → habitual → champion. The narrowing is the point.
  • Habit formation — of the people who use it, how many have a real habit.
  • Engagement mix and Where Copilot is used — the population split, and which surfaces carry the traffic.
  • Enablement plan — how many people need each action.
  • The shape of adoption — your typical user against your own champions.
  • Weekly trends — active users, and interaction volume including unlicensed.
  • Usage concentration — whether adoption is broad or carried by a handful of people.
  • Licensed and unlicensed side by side — the reallocation view.
  • What Copilot is working on — whether it is grounding in your content or just chatting.
  • Frequency vs intensity, adoption by department/country, where the unmet demand is.

Licensed users

Everyone holding a Copilot seat, with their engagement score, band, and a recommended action tag.

Filter by band, department, Cowork use, or Disabled accounts only — a disabled account still holding a Copilot seat is the easiest reclaim there is. Sorting and filtering happen server-side, and the CSV export is exactly what is on screen.

Unlicensed usage

People using Copilot Chat with no seat, treated as a population in their own right. Microsoft's own reporting cannot see these people at all. Anyone in the Frequent or Daily tile has built a Copilot habit with no seat, no training and no prompting — the strongest licence case you have.

Agents

Your Copilot agent estate, with a Keep / Review / Retire / New verdict and a stated reason for each.

An agent appears only once it has been used — the audit log records agents that ran, not agents that exist. New agents (first seen recently) are exempt from review on purpose: a brand-new agent with two users has not failed, it has not started.

Licence opportunities

Unlicensed users ranked by how strong the case for a seat is, each with a written justification you can paste into a request. Unlike the licensed list, the justification genuinely differs per person.

How this is calculated

Every formula, with worked examples and the full list of SKUs that were and were not counted as Copilot seats. Send sceptics here.

Three things people actually use it for

1. A licence reclaim review before renewal

  1. Set the period to Last 90 days — a longer window is fairer before taking a seat away.
  2. Note Reclaimable seats on the Overview.
  3. Licensed users → filter band Never used → export. These have produced nothing; confirm the seat is needed.
  4. Filter band Dormant → export. These tried Copilot and stopped — worth a conversation before you reclaim.
  5. Tick Disabled accounts only → export. Reclaim these immediately.
  6. Cross-check the Licensed and unlicensed side by side table. A department with idle seats and heavy unlicensed use is a reallocation, not a purchase.

2. Planning an enablement programme

  1. Adoption funnel → find the biggest drop. That tells you the kind of problem.
  2. Enablement plan → the size of each job.
  3. The shape of adoption → whether your typical user differs from your champions on frequency, depth or breadth. If they match on frequency but not breadth, more "use Copilot more often" training is wasted effort — they already do, just in one place.
  4. Adoption by department, worst first, is your running order.
  5. Frequency vs intensity → frequent but shallow departments need richer scenarios; deep but occasional ones need a reason to come back tomorrow.
  6. Usage concentration → if the top decile does most of it, your programme is propped up by enthusiasts and will collapse when they change team.

3. Proving a programme worked

  1. Before you start, download the Excel report and keep it.
  2. Run your programme.
  3. Download it again with the same period length.
  4. Compare. The Report sheet in each file records the period and every threshold used, so you can prove the two are comparable.

Check the thresholds match. The tuning is adjustable. If someone changed a band boundary between the two runs, "adoption went up" might just mean the bar moved. Both files record what they used.

Reading the charts

Chart Read it as Common misreading
Gauges The rate against a judgement scale The bands are deliberately stepped, not a gradient — 41% and 43% are the same verdict
Funnel Each stage is a subset of the one above. The % is conversion from the stage immediately above, not from the top A healthy 90% following a terrible 40% is still a good step
Habit tiles Of the people who use Copilot, how often Percentages are of active users, not of all seats. Never-used seats are in the reclaim figure
Engagement mix The whole licensed population in six exclusive bands —
Treemap Area = share of interactions Counts interactions, not people. One heavy user can dominate a surface
Frequency vs intensity Position, with your own medians as the dividing lines The lines are your medians, not industry targets. Bubble area is seats
Radar The shape of engagement, typical user vs your champions The gap between the outlines is the finding, not their size
Concentration How much of all activity each cohort accounts for Only active users are ranked
Stacked area Total volume and its composition Only the bottom band sits on a flat baseline, so only it reads precisely

The Excel report

The Excel report button exports the whole thing: twelve sheets with live, editable Excel charts bound to the cells — not pictures. You can re-plot, re-theme, and paste them into a board deck.

Sheet Use
Report What this file is: period, imports used, every threshold, and any warnings. Read this first when comparing two files
Headline figures, Adoption funnel, Engagement The executive numbers
Weekly trend Six months of history regardless of the period you chose
Departments and apps Every segment breakdown
Agents, Unlicensed usage The two populations Microsoft cannot show you
Enablement plan The size of each job
Licensed users, Licence opportunities The full lists, auto-filtered
How this is calculated Every formula, and the SKU classification

Glossary

Term Meaning
Licensed users People holding at least one licence classified as a Microsoft 365 Copilot seat. The true seat count
Analysed / scored users The licensed users this run actually scored. The same as above unless a very large tenant hit the analysis row cap — every rate is of this number, and the page says so when they differ
Active At least one Copilot interaction in the period. A deliberately low bar
Habitual Engagement of 50+ — Copilot is part of the working week. This is the figure that tracks value
Dormant Used Copilot before this period, but not once inside it. Needs a conversation
Never used No Copilot activity at all in the history window. Needs onboarding, or the seat back
Reclaimable seats Dormant + never used
Engagement score 0–100 from frequency (50%), depth (30%) and breadth (20%), each capped
Bands Champion 75+, Established 50+, Developing 25+, Trialling below that
Habit buckets Infrequent 1–5, Moderate 6–10, Frequent 11–19, Daily 20+ active days a month
Breadth How many different Copilot surfaces (Teams, Word, Outlook, Chat…) someone uses
Business case score 0–100 for an unlicensed user. Already using Copilot Chat is worth the most, because it is evidence rather than inference
Cowork Microsoft 365 Copilot Cowork. Only shown if actually detected

Warnings you might see

Warning What it means What to do
"No licence information has been imported" The user metadata import has not run Enable it; the tool cannot work without it
"Neither the Copilot audit import nor Microsoft's usage report has data" Every user will look unused Check Health before acting on anything here
"The Copilot audit import has no data … figures come from Microsoft's report" Falling back. That report covers Microsoft's window, not yours, and excludes unlicensed use Treat unlicensed figures as unavailable, not as zero
"This tenant has 'concealed user information' enabled" Microsoft's per-user report returns hashed identities Audit-derived figures are unaffected — this is why the audit import matters
"only N users could be analysed" A very large tenant hit the row cap Rates describe those N users, not the whole tenant. Do not quote them as tenant-wide
"Could not load …" One query failed; the rest of the report is fine The named chart is missing. Check Health, then retry

FAQ

Why is my adoption rate 100% but habit rate near zero? Everyone opened Copilot at least once and almost nobody kept using it. This is the most common and most expensive pattern, and it is exactly what the adoption rate on its own conceals.

Why do the numbers differ from the Microsoft 365 admin centre? Different sources and windows. This tool primarily uses the Copilot audit log, matching your chosen period exactly and including unlicensed users; Microsoft's dashboard uses its own aggregation window and licensed users only. The How this is calculated tab names the source for every figure.

Why is a Copilot-branded licence not counted as a seat? Microsoft ships Copilot-branded products that are not a Microsoft 365 Copilot seat — Copilot Studio, Copilot for Sales. Counting them would inflate the licensed population and make adoption look worse than it is. Every SKU found is listed under How this is calculated, including the excluded ones.

Someone is on the "never used" list but says they use Copilot daily. Check the audit import in Health, and whether they hold a seat at all — heavy Copilot Chat use without a seat appears under Unlicensed usage, not in the licensed list.

Can I change the thresholds? They are defined in the analysis options rather than hard-coded, and every screen quotes the values actually used. Changing them is a code change today — but if you do, remember that two Excel snapshots are only comparable when both used the same values, which is why each file records them.

Why does the agent list not show an agent I know exists? Agents appear only once they have been used. An agent built but never run is invisible here — and to everyone else.

Related pages

Overview

Getting started

Installation

Operations

Upgrade notes

Solutions

Reference

Project

Clone this wiki locally