The Microsoft 365 Copilot Business Case Your Leadership Team Will Actually Approve

I sit in a lot of leadership rooms. The slide that kills the deal is almost always the same one: a big number for "productivity gains" with no idea where it comes from.

Your CFO has seen that slide before. They know what it is.

I should say up front that leadership rooms are my favourite place to work, and not for flattering reasons. It is maximum impact. Senior leadership is the only group that can require an entire organisation to learn something, and the only group that can decide that real time and real budget go into it. When leadership understands this deeply, the organisation moves. When it does not, everything stays at pilot level.

What I like about the rooms I have been in lately is that nobody is asking whether any more. They are asking how.

Start with the number nobody wants to say out loud

85% of employees worldwide have access to an AI tool. 25% actually use one. And roughly 95% of enterprise pilots never move the bottom line. Those numbers come from the IBM CEO Study 2026, McKinsey, and MIT's Project NANDA, which is to say from people who study technology adoption and do not sell you the technology.

Then there is the BCG survey from July 2026, which asked 152 CEOs at companies above $500M in revenue. 88% report a benefit somewhere. 26% turned it into anything resembling a broad business change. And 14% can tell you what it did to the P&L.

Read that last one again. 14%.

So when you walk into the room, understand what your leadership team has already absorbed from the market: this stuff usually does not land. Your job is not to convince them it is exciting. Your job is to show them why yours will be different.

Name one process. Not a department.

"Improve productivity in finance" is not a business case. It is a wish.

"The monthly budget control pack takes 63 slides and 2 people 3 days to assemble, and we want that down to half a day" is a business case. You can measure it, you can fail at it, and everyone in the room immediately understands what you are asking for.

Pick the process that is boring, repetitive, document-heavy, and happens on a schedule. Monthly reporting. Contract review. Tender responses. Customer complaint analysis. Those are the ones where Microsoft's Copilot is genuinely strong today, because it sits on top of the files and the mail you already have.

I wrote up a set of these in more detail on the Copilot use cases page, and you are welcome to lift one that matches your organisation.

Budget in 3 lines, not 1

This is the part that gets skipped, and it is the part that decides whether the whole thing works.

Line 1: licences. Predictable, and usually the smallest number in the whole proposal. A few tens of dollars per user per month.

Line 2: implementation. Training, the person who runs this internally, the champions.

Line 3: the hours. The actual learning and practice time of the people who will use it, plus the time managers spend checking output in the first few months.

Line 3 is the one nobody writes down. It is also the one that determines whether lines 1 and 2 were worth anything. An organisation that approves only line 1 has approved an expense, not a project.

And the reverse maths is easy: a finance person who saves 2 hours a week pays back the licence several times over. If you are not funding Copilot for your finance team, you are leaving money on the floor.

Say who owns it. By name.

Here is the single strongest number I know in this whole space. An AI initiative with senior leadership backing is 3.8 times more likely to hit its goals. Of every factor studied, that was the biggest gap between success and failure. The research comes from Prosci, KPMG's Global AI Pulse 2026, and Gallup, all of whom work on organisational change rather than selling AI licences.

And next to it: only 24% of managers say the CEO is accountable for AI results, and only 22% of employees have heard from leadership how it will apply to them.

So put a name on the slide. Not a committee. The CEO and the C-suite are the owners of this project. Being the owner means giving teams time to learn, time to keep learning, and time to practise. It means giving your champions time to get good, time to mentor, and time to teach other people.

Projects owned by a committee are the ones I watch end quietly around month 8, with nobody ever declaring them over.

Ask for a mandate, not enthusiasm

The second thing to get out of the room is a decision about language.

Leadership has to require it, genuinely require it, of everyone below them. Use Copilot on every task, or at least try. Using AI is not embarrassing any more, it is part of the job. Employees start using this when their manager asks them about it in a meeting, not when a licence appears in their account.

And in the same breath, the second requirement: no copy and paste. Copy and paste is how you get publicly embarrassed. Every output is a draft, and what makes it worth sending is what you pour into it: your judgment, your experience, your creative thinking, your strategy, and a lot of human brain.

Those two requirements travel together. Demanding usage without demanding thinking produces an organisation that ships raw output, and that is worse than where you started.

And ask for one more thing while you are in the room: a standing slot. This has to come up in management meetings and team meetings as a recurring item with real time in the managerial routine, not a burst of interest once a quarter. A topic that never appears on the agenda tells everyone it is not actually important, whatever the launch email said.

What not to promise

Do not promise a percentage. You do not know it yet, and the moment you say "30% productivity gain" you have handed your CFO a stick to hit you with in 2 quarters.

Do not promise headcount reduction. Even if you believe it, saying it out loud guarantees your rollout meets quiet resistance from every person who has to make it work.

Do not promise it will be consistent. Copilot is excellent at analysis, investigation, and open questions. It is not the right tool for a report that has to come out identical on every run, and Microsoft says so themselves. Every number that leaves the process gets checked by a person.

What you can promise: a named process, a measurement date, and an owner.

Put the measurement in before you need it

Decide, in the same meeting, what you will measure and what a KPI actually looks like here.

Measure 2 different things. Usage, meaning how many people work with it in a given week 3 months after the training, with nobody reminding them. And business impact, meaning what happened to the process you named at the start.

Usage without impact is motion. Impact without usage data is a guess.

Set both before you approve anything, because otherwise the review in 6 months turns into an argument about feelings. And remember that 14% figure. Most organisations never get here at all, which is exactly why getting here is an advantage.

Questions leadership will ask you

How do we know this is not just another pilot that dies?
Because you named an owner from the executive team, you allocated learning hours in writing, and you set a measurement date. Those 3 things are what separates the pilots that survive from the ones that go quiet. I wrote about the failure pattern in detail in why Copilot rollouts stall.

Why Copilot and not something else?
Because it sits inside the files, mail, and meetings you already have, under your existing agreement, and your data stays in your tenant. That is a different conversation from a general chatbot, and for most enterprises it is the deciding factor.

How long before we see something real?
3 to 6 months for a measurable change in how work gets done, assuming training and follow-up. Anyone promising you results in 2 weeks is selling you a demo.

Do we need everyone, or start small?
Start with the group whose work is document heavy and whose output is visible to leadership. Finance and operations are usually the fastest payback. Expanding from a group that already works beats a company-wide launch that nobody owns.

One more thing

The thing that changes a room is not the slide. It is when the CEO opens their own laptop and tries the thing in front of everyone.

Your team watches what leadership does with this, whether or not anybody says a word about it. That is the real business case, and it costs nothing.

If you want help putting this in front of your leadership team, that is a large part of what I do. There is more on the Copilot training for executives page.

Every live online format and price sits on one page: live online Copilot workshops and lectures.

Eyal Marcus is an AI consultant, trainer and keynote speaker who helps organisations actually adopt AI, working in English and Hebrew with enterprises in Israel and across Europe. He has been working with AI since early 2022, almost a year before ChatGPT launched, publishes the weekly "Don't Panic" AI newsletter and hosts the "Hands On AI" podcast. He has delivered 140 sessions in 55 organisations, including Israel Electric Company, Clalit, Amdocs, Cellebrite, Keter, the Jewish Agency, UMI Group and Amisragas.