Your boss says, "Let's get everyone using Copilot," and somehow you've become the person who has to make it happen. You start looking for training, only to find that the sponsor is picturing a faster month-end close, the analysts want help untangling formulas, and IT is still checking which features people can use. You're all talking about the same class, but you may be planning three different things.
As the trainer, this is where I want to slow down and ask what should be different when everyone goes back to work. If the class goes beautifully but the monthly report is still a copy-paste marathon two weeks later, what did we actually accomplish? A better starting point is to pick one piece of work, agree on what improving it would look like, and find out what people need to get there.
To help you have that conversation, I've turned questions from my training correspondence and workflow reviews into two short, fillable PDFs. Download whichever one you need and save a copy to fill in together.
1 PAGE
For the first conversation
Use the sponsor worksheet in a 30-60 minute meeting to agree on what you're trying to accomplish.
2 PAGES
For the work that follows
Use the rollout plan to work through the details with the trainer, IT and the people doing the work.
Get everyone talking about the same project
I'd start by getting the sponsor and somebody who does the Excel work into the same conversation. Bring in IT or the trainer where their input would help, and give yourselves 30-60 minutes with an example you can open and talk through. It's much easier to agree on a useful class when everyone can point to the work they're hoping to improve.
Let's stick with that monthly report. Maybe "easier" means spending less time matching department names, or getting fewer corrections back from the reviewer. It might also mean having somebody else who can prepare the thing when the usual spreadsheet person is out. That last one can be a pretty compelling business case if your reporting process currently depends on one person never taking a vacation.
As you talk it through, you'll probably find a few different kinds of work hiding in the request. Someone needs help understanding a lookup, the team needs to agree on what happens to unmatched departments, and IT needs to check access to a feature. That's why the worksheet asks about people, process and tools together. We can plan the training while also giving those other questions to the people who can answer them.
How I'd use the half hour
- 5 minutes: Agree on the result you want.
- 10 minutes: Talk through the people, process and tools.
- 8 minutes: Set the scope and check available resources.
- 7 minutes: Make decisions and assign next steps.
With an hour, add twenty minutes to walk through the actual work and ten to resolve the questions it raises. Leave with a first deliverable, a few things you've agreed can wait, and a person and date attached to each open question.
Here's the one-page worksheet to keep that conversation on track. A few useful notes in each box are plenty. If an answer is still "we need to ask IT," write down who will ask so it doesn't become everybody's job and nobody's job at the same time.
Tell me what people already do in Excel
Once we know which piece of work we're taking on, I want to know who's going to do it. "They build PivotTables but usually ask someone else to fix the formulas" gives me a lot to work with when planning a class. "They're intermediate" leaves me guessing. I'd ask about Copilot separately, too, because someone can be very comfortable in Excel and still be opening the Copilot pane for the first time.
I get into those distinctions in my post on beginner, intermediate and advanced Copilot in Excel. A couple of examples of what people already do will help us choose an exercise the group can follow, with something extra for anyone who's ready to go further.
From there, we can be specific about what someone should be able to do afterward. For our monthly report, I'd like to see them prepare a regional summary, check it against the source and explain how returns were handled, without me walking them through every click. That tells me what to build into the exercise, and it gives you something to look for when people try the work on their own.
This is where teaching Excel alongside Copilot becomes so useful. If somebody understands which records belong in a total, they have a way to check the answer Copilot gives them. We can build both skills into the same exercise. I shared some thoughts on that balance here:
Bring the workbook and explain the weird parts
To build that exercise, I'll need a small fictional or approved example of the work, ideally with a finished result we can compare against. The workbook will tell me some of the story. You'll need to fill in the parts that live in people's heads: which period matters, which records belong in the report and what happens to the odd rows.
Looking back through my workflow review notes, I find questions about table names, consistent column names, where the template lives, and what to do when a match is missing or uncertain. These are easy details to leave out when you've been doing the report for years. They're also exactly the details a trainer needs when turning your process into an exercise.
Take the department names in our report. Should "Sales – East" and "Eastern Sales" be treated as the same department? If the answer is "usually," I'd like to hear about the exceptions before we ask Copilot to tidy everything up. We might have it suggest matches and leave the uncertain ones for review, then practice deciding which suggestions to accept. Now the checking is part of the exercise from the beginning.
Agree on a few checks while you have the example in front of you. The total should reconcile to the source, every input row should be accounted for, and unresolved matches should be easy to find. Say what may change, too: perhaps the source stays untouched and the reviewed output goes on a new sheet. Microsoft's Copilot in Excel FAQ calls for reviewing and verifying the output. This is where we work out what that means for your report.
Have IT try the exercise before the class
Will it work on their laptop?
With an exercise in mind, we can give IT something concrete to try. Have someone use a participant's account, device and file location to open the workbook, use the required feature, make the change and save the result. "We have Copilot" becomes a much more useful answer when somebody has actually worked through those steps. It also makes the first ten minutes of class much more pleasant for everybody.
Microsoft's FAQ covers requirements such as app versions, licensing, network access and privacy settings. If a feature is in preview, IT can also check Microsoft's Frontier guidance. Test any custom skill or agent work separately if it's on the agenda. Being able to open a chat window doesn't tell us whether that part of the exercise will work.
Keep a note of who tried it, when and what happened. If something is unavailable, we have time to work out another exercise, pair a demonstration with a design activity, or adjust the agenda. IT can make the access decisions, and I can build the session around what your people will actually have in front of them.
Work out who is doing what
By this point, you have the outline of a class: a piece of work, the people learning it, an example with some rules, and a plan for getting the tools to cooperate. The two-page rollout plan gives you somewhere to put that together. Use the first page to prepare a small pilot and the second to decide who will follow up, what you'll compare and when you'll review the results.
Work through the logistics while you're there. If people will follow along on laptops, somebody needs to get them the files. If there are two sessions, make sure everyone knows whether they're repeats for different time zones or whether the second builds on the first. Agree on recording and internal reuse ahead of time, too.
I'd also put some practice time on the calendar while you still have everyone's attention. Otherwise, there's a decent chance the next time somebody opens the workbook will be when you ask how the training went. Start with a small pilot, give people a chance to try the work, and use what you learn before bringing in the next team and its particular collection of spreadsheet quirks.
Give yourself something to compare afterward
Remember the faster monthly report the sponsor was picturing? Before the class, let's find out what that report takes today. Have someone complete a typical version using their current approved method, including any AI they already use. Keep the output and note the time, the checks it passes on the first review and how much help the person needs. The rollout plan has room for that small before-and-after comparison; keep longer logs with the project files and repeat the comparison if results vary.
Be a little fussy about what goes into the time figure. Include preparing and building the report, reviewing it and fixing it, across everyone involved. If Copilot gets a draft out quickly but somebody spends the afternoon repairing it, that afternoon belongs in the comparison. Keep waiting time separate so you can see the difference between active effort and elapsed time.
Here's a made-up example of the time you might record across everyone involved:
| Work involved | Before (minutes) | Later (minutes) |
|---|---|---|
| Build the report | 45 | 25 |
| Review it | 15 | 12 |
| Make corrections | 10 | 8 |
| Total effort | 70 | 45 |
We'd have 25 minutes less effort for the same accepted result, which is worth discussing. Just make sure the tasks are comparable in size and difficulty, and give people a fresh example rather than the exact workbook we practiced in class. A small comparison helps us decide what to try next; it won't prove that training caused the difference.
I'd also want to see whether people can explain their result, catch an error and work with less help. A good session rating is nice to receive, of course, but the reconciled report tells us something the feedback form can't. Microsoft's employee AI enablement guidance discusses baselines and value measurement, and its Copilot reporting guidance can help with broader usage questions.
Set the review date around when people will have a chance to use what they've learned. A weekly task might give you something to discuss in two to four weeks; a monthly report needs a reporting cycle. My 30-day follow-up plan for Excel teams goes further into that part of the work.
Bring what you have
When the sponsor asks how the training went, you'll have a report to look at together, some checks to discuss and a better idea of what people can now do on their own. You may also have found an access problem, an unclear business rule or a step that needs more practice. Those are useful things to bring into the next conversation.
For now, bring the workbook and the questions you have. The sponsor worksheet will help you get that first conversation going, and the rollout plan is there when you're ready to work through the details. If you'd like me involved, take a look at my Copilot training for Excel teams. I'd be glad to help turn "let's get everyone using Copilot" into a class that makes sense for the work your team actually does.