A well-built Copilot template can still fail at launch. Not because it’s wrong, but because people keep using the old file saved on their desktop, don’t know the brand kit exists, or try Copilot once with a vague prompt, get a generic result and decide it doesn’t work.
Rollout is its own piece of work. This post covers the sequence, the messages, and the few habits worth teaching — along with how to use the first weeks of real use to improve the template.
Before launch: get the system settled
Employees will judge Copilot by their first few decks. Make sure those first decks come from a settled system:
- Brand managers designated, with at least one backup. Only brand managers can create and edit official Brand Kits, and IT has to designate them — Microsoft notes permissions can take up to 24 hours.
- One official Brand Kit per real brand, holding the finished template and approved assets.
- Strict brand adherence decided, after testing shows your layouts cover what people actually make.
- Note instructions and any Brand Skills tested against your prompt set. Microsoft’s guidance for Brand Skills specifically recommends governance review, testing before broad publication, and communicating availability to users.
- Old templates retired, or at least clearly labeled as retired, on SharePoint and in shared drives.
That last point is the most often skipped. If the old template is still the first result in a search, people will use it.
Sequence: pilot, then widen
We’d launch in three stages.
1. A pilot group. Ten to thirty people who make presentations often, across a few departments, including at least one skeptic. Give them the template and kit two to four weeks before everyone else. Ask them to use it for real work, not demos.
2. Fix what the pilot finds. Collect their generated decks — with permission — and review them. Sort problems by cause: template structure, examples, rules, prompts. Fix the system, rerun your test prompts, and confirm the fixes held.
3. Launch to everyone, with the pilot group as local champions who can answer questions in their departments.
What to tell employees
Keep the message short and practical. People need to know four things.
Where to start. “Select the [Company] brand in the Copilot pane before you prompt” or “Start from the template in [location].” One instruction, repeated everywhere.
What Copilot is good for. First drafts of everyday decks: updates, briefings, summaries, internal proposals. Turning a document into a structured deck. Being honest about scope builds trust faster than overselling.
What it isn’t for. Decks where the stakes are highest — board, investor, major customer, keynote — still go through the usual design support. Copilot can draft them; experienced people should shape them.
That review is their job. Microsoft’s own guidance is that Copilot output “should be human-reviewed and edited accordingly.” Say so plainly. Employees remain responsible for what they send.
The two habits worth teaching
Most prompt training covers too much. If employees remember only two habits, make them these:
Habit 1: Tell Copilot who it’s for and what you need
A topic isn’t a brief. “Q3 results” produces a generic deck. “Q3 results for regional leads, who need to approve moving budget to the northern region — about eight slides, recommendation first” produces a useful one. Audience, purpose, length, priority: four short phrases.
Habit 2: Fix the outline before generating slides
Copilot proposes an outline before creating slides, and Microsoft’s workflow lets you refine it. This is where the argument gets decided, and it takes seconds to change. Changing the same thing after slides exist takes much longer.
Everything else — follow-up refinements, specific edits — people learn quickly by using the tool.
What not to do
- Don’t train people to fix the template’s problems by hand. If employees routinely recolor charts or resize text after generation, the template needs fixing, not the people.
- Don’t publish a long prompt library. A few examples help. Dozens get ignored, and they go stale when Microsoft changes the product.
- Don’t launch with the settings still changing. Turning strict brand adherence on and off after launch confuses people about what the tool does.
- Don’t measure success by adoption alone. Usage tells you people tried it. It doesn’t tell you whether the decks were good or how much fixing they needed.
Learn from real use
The first six to eight weeks after launch are the richest source of information you’ll ever have about how the template behaves. Use it:
- Sample real generated decks each week, with permission, and review them the way you reviewed the pilot’s.
- Track fix-up effort: for a sample of decks, how much editing was needed before they were used — none, light cleanup, meaningful correction, or a rebuild.
- Tally failure types to see whether problems cluster in the template, the rules, or prompts.
- Feed fixes back into the system, rerun the test prompts, and tell employees when something improves.
Then settle into the quarterly routine the template owner will run from here on.
Common questions
Should we train everyone on prompting before launch?
A short introduction helps, but long prompt training tends to fade and go out of date. The two habits that matter most are briefing Copilot on audience and purpose, and fixing the outline before generating slides. Most other skills come quickly with use.
Should strict brand adherence be on at launch?
Only if testing shows your layouts cover what people actually make. Switching it on and off after launch confuses people about what the tool does, so decide before rollout.
What do we do with old templates?
Retire them clearly. Remove them from shared locations or label them as retired. If the old file is still the first search result, people will keep using it.
Key takeaways
- Settle the system before launch: brand managers, one official kit, deliberate settings, tested rules, and old templates retired.
- Pilot with real users, fix what they find, then launch with them as champions.
- Tell employees where to start, what Copilot is for, what it isn’t for, and that review is their job.
- Teach two habits: brief Copilot on audience and purpose, and fix the outline before generating.
- Use the first weeks of real use to improve the system, not just to count adoption.
Launching a Copilot template this quarter? Have your template scored first.
Related reading
- From prompt to deck: what happens when Copilot creates a presentation
- Who owns the template after launch
- For IT: the settings are the easy half
Sources
- Microsoft Support, Create a new presentation with Copilot in PowerPoint (June 2026)
- Microsoft Support, Create and manage official Brand kits (July 2026)
- Microsoft 365 Message Center, MC1473166 — Brand Skill support (16 Sep 2026)