From decks to systems: how presentation work is changing

For most of the history of corporate presentations, the unit of work was the deck. You needed a presentation; someone made it. If you had a lot of presentations, you had a lot of someones โ€” an in-house team, an agency, a freelancer network, or all three.

That’s changing, and not in the simple way the headlines suggest. Presentation work isn’t disappearing. It’s moving: away from producing individual decks and toward designing the systems that produce them, and toward the judgment calls no system makes on its own. This post is our view of that shift, as a company that has designed corporate presentations for fifteen years and is changing how we work because of it.

Organizations now have their own engines

The core change is that most large organizations now own a presentation engine. With Microsoft 365 Copilot, employees can generate a structured, branded first draft from a prompt or a document. The engine is built into the tool they already use.

That engine has three properties worth noticing:

  • It learns from what you give it. Microsoft’s guidance describes the template’s sample slides as Copilot’s primary learning source, with placeholders and theme definitions shaping the rest. The engine’s quality depends on its inputs.
  • It can be governed. Since September 2026, brand managers can lock Copilot to approved layouts with strict brand adherence, put instructions in template speaker notes, and publish written rules as Brand Skills.
  • It needs review. Microsoft is clear that Copilot’s output should be reviewed by a person.

Each of those properties points at work that still needs doing โ€” just not the work of making every deck by hand.

Where the valuable work moves

It helps to think of presentation work as a sequence: understand what needs to be said, decide how to say it, make it, check it, and learn from what happened. For a long time, most outside help was bought as the whole sequence, delivered as a finished deck.

When an organization has its own engine, different parts of that sequence change value.

Making routine decks becomes cheaper and more internal. This is the part the engine does. It’s also the part that consumed most of the hours in traditional presentation work.

Designing the system becomes more valuable. The template, its example slides, the Brand Kit and the written rules decide the quality of everything the engine makes. That’s upstream work with enormous leverage โ€” and it requires exactly the skills presentation designers have: knowing how a company actually communicates, what good hierarchy and density look like, and how brand shows up beyond a color palette.

Judgment on the decks that matter stays valuable. Board presentations, investor materials, keynotes and major pitches are hard because of the argument, the audience and the stakes. An engine can draft them. The consequential decisions remain human.

Checking becomes more important. A generated deck can look finished while missing the point. Assurance โ€” independent review of whether a presentation says the right thing, the right way โ€” matters more when first drafts are cheap and polished.

Learning becomes continuous. The system needs tending as the organization and the platform change. Microsoft changed these settings twice in September 2026 alone.

What this means for buying presentation help

If you buy presentation design from outside, the shift suggests some new questions.

Where do you actually need help? Not “do you need someone to make this deck?” but “where in the process is your unresolved problem?” For some organizations it’s the system. For others it’s the decks that matter. For many it’s both.

Are you paying production rates for judgment work? When routine production moves in-house, the work you still send out tends to be more judgment-heavy. It’s worth making sure the way you buy matches the work you’re buying.

Who looks after your engine? If nobody owns the template and rules after launch, the engine degrades quietly. That’s a gap an outside partner can fill, on a regular cadence, or an internal owner can fill with clear responsibility.

How we’re changing

We’ll be direct about our own position. Much of our history is production: designing presentations, one at a time, for companies that needed them done well. That work isn’t going away โ€” the presentations that matter still need to be excellent, and we still make them.

But we’ve also started designing the systems that produce everything else: Copilot-ready templates, example slide sets, Brand Kit setups and written rules, tested against real prompts and looked after as the platform changes. We built our own system first, before offering one to anyone else, because we wanted to learn how Copilot actually behaves rather than repeat what we’d read.

What we won’t do is promise what a generative tool will produce. Nobody can. What we commit to is conformance with Microsoft’s published guidance, testing with real prompts against your own template, and correcting what drifts.

Common questions

Does this mean presentation agencies are becoming obsolete?
Not in our view. Routine production is moving in-house. System design, judgment on the decks that matter, independent review and ongoing care of the system are becoming more important. The mix of work changes more than the need for expertise.

What’s the first step for an organization making this shift?
Find out what your template currently teaches Copilot. Everything else, from example slides and rules to oversight, builds on that.

How should we think about budgets?
Separate routine production, high-stakes work and system work, and fund each for what it is. Paying production rates for system or judgment work tends to underinvest in exactly the work that now matters most.

Key takeaways

  • Organizations now own a presentation engine that learns from their template, can be governed, and needs review.
  • The valuable work moves: routine production gets cheaper and more internal; system design, judgment on high-stakes decks, assurance and ongoing learning become more valuable.
  • Buyers should ask where their unresolved problem is, whether they’re paying production rates for judgment, and who looks after their engine.
  • The presentations that matter still need to be excellent; the systems behind everything else now need design too.

Curious where your engine stands? Have your template scored.**

Related reading

Sources

Fix-up time: the one number that tells you whether Copilot is paying off

When organizations measure Copilot, they usually start with adoption: how many licensed people use it, how often, for what. Those numbers are easy to get and genuinely useful. They answer the question “are people trying it?”

They don’t answer the question leadership eventually asks: “is it saving us anything?” For presentations, that question has a surprisingly clean answer, and it isn’t adoption. It’s fix-up time โ€” how much work a generated deck needs before it’s actually usable.

Diagram of a five-level fix-up scale from 0, usable as generated, to 4, rebuild

Why fix-up time is the right measure

The value of Copilot for a presentation is the difference between how long the deck used to take and how long it takes now. “Now” has two parts: generating the draft, and fixing it.

Generating is fast and fairly consistent. Fixing is where the variation lives. A generated deck that’s ready to use has captured almost all the available saving. A generated deck that needs rebuilding may have saved nothing at all โ€” or cost time, if someone spent an hour trying to rescue it before starting again.

So fix-up time is the term that decides whether savings are real. It’s also the term most directly affected by things an organization controls: the template, its example slides, the written rules and the upkeep. That makes it both the best outcome measure and the best diagnostic.

A simple scale

Timing fix-up precisely is difficult and intrusive. A five-level scale is easier to apply consistently and nearly as informative:

Score Level Meaning
0 No intervention Usable as generated
1 Light Minor cleanup โ€” a typo, an alignment, a word
2 Skilled Meaningful correction โ€” restructuring slides, fixing emphasis, reworking a chart
3 Senior judgment Rethinking the argument or structure; needs an experienced person
4 Rebuild The generated deck provides little usable value

Scoring takes seconds per deck once reviewers are familiar with it. The average across a sample โ€” and the distribution across levels โ€” tells you far more than adoption numbers.

What to measure against

Fix-up scores mean most when compared. Three comparisons are especially useful:

Before and after a template change. Run a fixed set of realistic prompts against the template, score the output, make a change โ€” new examples, rebuilt placeholders, a Brand Skill โ€” and run the same prompts again. A falling average says the change helped.

Against Microsoft’s baseline. If you’re rebuilding a template, score output from a version that simply follows Microsoft’s published guidance โ€” representative sample slides, correct placeholders, theme-defined colors and fonts โ€” and then from your fuller rebuild. That shows what the extra design work adds above the baseline.

Over time. Score a sample of real generated decks each quarter. A rising average with no changes on your side is a sign of drift โ€” the organization’s needs or Microsoft’s platform have moved.

Reading the distribution

The average is useful; the shape is more useful.

Mostly 0s and 1s: the system is working. Savings are flowing through.

Clustering at 2: decks need skilled correction. Usually a template problem โ€” missing examples, weak hierarchy, density mismatches, placeholders sized wrong. Fixable, and worth fixing, because every point of improvement applies to every deck.

Clustering at 3: structure and argument are wrong. Look first at how people are prompting and whether they’re reviewing outlines before generating; then at whether the template has layouts and examples for the kinds of arguments people make.

Any meaningful share of 4s: find out which deck types they are. Often they’re decks that shouldn’t be generated in the first place โ€” high-stakes work that needs experienced people from the start โ€” or deck types the template simply doesn’t cover.

Pair the score with a reason

When a deck scores 2 or above, note why โ€” technical, brand, hierarchy, structure, density, visualization, intent, prompt, or a model limitation. The combination of score and reason tells you not only how much fixing was needed but where the fix belongs.

Practical tips

  • Score the same way every time. Agree what each level means with a few example decks before starting.
  • Use two reviewers occasionally to check you’re scoring consistently.
  • Sample, don’t census. A modest, varied sample each cycle is enough.
  • Keep the fixed prompt set stable. It’s the only way to compare runs fairly.
  • Report the distribution, not just the average. Leadership understands “most decks now need only light cleanup” better than “average 1.4.”

Why we like this measure

It’s honest. It doesn’t assume savings; it detects them. It doesn’t reward activity; it rewards usable output. And it points directly at the things that can be improved.

It’s also the measure we hold our own work to. If rebuilding a template and designing its examples doesn’t reduce what a person has to fix afterward, the work hasn’t earned its fee.

Common questions

Who should score decks?
People who know the brand and presentation quality, usually designers or the brand manager. Agree what each level means using a few example decks before you start.

Can employees score their own decks?
They can report how much editing they did, which is useful. Consistent scoring across teams works better with a small group of trained reviewers.

How does fix-up time relate to cost?
Multiply the reduction in fix-up time by the number of decks and a loaded hourly cost to estimate capacity freed. Count it as cash only where it reduces spending.

Key takeaways

  • Adoption shows people tried Copilot; fix-up time shows whether it saved anything.
  • Score generated decks on a five-level scale, from usable as generated to rebuild.
  • Compare before and after changes, against Microsoft’s baseline, and over time.
  • Read the distribution: clusters at 2 usually mean template work; at 3, prompting and structure; 4s, the wrong decks or missing coverage.
  • Pair each score with a reason to know where the fix belongs.

Want a fix-up baseline for your template? Have your template scored.**

Related reading

Sources

Reviewing generated decks: a faster workflow for brand teams

Brand review was designed for a world where presentations were scarce and expensive to make. A few important decks came through the brand team; everything else went out unreviewed and everyone accepted the risk.

Generated presentations break that model from both ends. There are far more decks, made far faster, and many of them look finished enough that nobody thinks to ask for a review. Trying to review them all is hopeless. Reviewing none of them is how brand problems scale.

This post describes a different workflow: review less, learn more, and fix causes rather than slides.

From inspection to diagnosis

Traditional brand review is inspection: a reviewer checks one deck, marks what’s wrong, and the deck’s owner fixes it. The deck gets better. Nothing else does.

With generated decks, most problems aren’t unique to one deck. They’re patterns, produced by the system โ€” the template, its example slides, its rules. The same missing layout produces the same forced content in every deck that needs it. The same hardcoded color produces the same off-brand charts everywhere.

So the useful kind of review is diagnosis: look at enough decks to see patterns, trace each pattern to its cause, fix the cause once, and confirm the fix worked.

Microsoft now helps with the inspection half. Its Brand Reviewer, part of Brand Kit, “identifies issues and suggests one click fixes addressing incorrect colors or fonts, misplaced or unapproved logos, off brand imagery or picture styles, layout, spacing, or alignment issues.” Encourage people to use it: it cleans up the deck in front of them. But it fixes the deck, not the reason the deck came out that way. If Brand Reviewer keeps correcting the same color on every chart, that’s a pattern pointing to the template โ€” exactly what the workflow below is for.

The workflow

1. Sample, don’t sweep

Each review cycle โ€” weekly during launch, monthly or quarterly once things are stable โ€” take a sample of real generated decks. With permission, and with sensitive content handled appropriately, collect decks from different teams and different kinds of work. The goal is variety, not volume.

Alongside real decks, rerun your fixed test prompt set: the same realistic prompts every time. Real decks show what people actually ask for; the test set shows whether the system’s behavior has changed.

2. Grade readiness quickly

For each deck, a single judgment first: how much work would it need before it could be used?

  • Ready โ€” usable as generated.
  • Minor corrections โ€” cleanup only.
  • Remediation โ€” meaningful design correction.
  • Reconceptualization โ€” the structure is wrong.

This takes seconds per deck and gives you the headline: is the system producing usable decks, and is that improving?

3. Tag the failures

For decks that need work, tag why, using a short set of categories:

  • Technical โ€” overflow, cropping, content in the wrong placeholder.
  • Brand โ€” wrong colors, fonts or imagery.
  • Hierarchy โ€” the important thing isn’t prominent.
  • Structure โ€” the wrong way to communicate the content.
  • Density โ€” too much or too little.
  • Visualization โ€” weak charts or tables.
  • Intent โ€” misses the point.
  • Prompt โ€” the request was too thin.
  • AI constraint โ€” a model limit no template change fixes.

4. Count and prioritize

Tally the tags. The biggest category is usually where the next fix belongs. Technical, structure and density failures usually point to the template and its examples; brand and visualization failures to the theme, styles and Brand Kit; prompt and intent failures to employee guidance and outline habits.

5. Fix the system, once

For the top pattern, find the smallest system change that would prevent it:

  • a new or revised example slide (Microsoft describes sample slides as Copilot’s primary learning source);
  • a resized or retyped placeholder;
  • a note instruction in the relevant template slide’s speaker notes;
  • a rule in a Brand Skill;
  • a correction to the theme or Brand Kit assets.

6. Verify

Rerun the fixed test prompts. Did the pattern shrink? Did anything else get worse? Keep the change only if it helped.

7. Report briefly

A few lines to whoever owns brand: readiness distribution this cycle versus last, the top failure pattern, what was changed, and whether it worked.

Where individual review still belongs

Sampling handles routine decks. It doesn’t replace individual review for:

  • external, high-stakes presentations โ€” customers, investors, boards, press;
  • regulated content โ€” anything with legal or compliance requirements;
  • new use cases โ€” the first decks of a new kind, before the system has been tuned for them.

Microsoft’s guidance is that Copilot output should be human-reviewed. For routine decks, that review belongs to the person sending the deck. For high-stakes decks, it should include people with the judgment the stakes require.

Why this is faster

The time saved doesn’t come from reviewing faster. It comes from reviewing once for many decks. Fixing a pattern in the template corrects it everywhere, including in decks the brand team will never see. Over successive cycles, the most common failures shrink, and review time moves toward the harder problems.

It also changes what the brand team is known for. Instead of being the bottleneck that sends decks back, it becomes the team that makes everyone’s generated decks better without being asked.

Common questions

How many decks should we sample?
Enough to see patterns across different teams and deck types, and few enough that the review actually happens. Variety matters more than volume. Keep the fixed test prompt set alongside, so you can compare cycles fairly.

How do we collect generated decks without invading privacy?
Ask for volunteers, agree what will be looked at and why, and exclude sensitive content. The purpose is to improve the template, not to evaluate individuals; say so plainly.

What if the biggest problem is prompts, not the template?
Then the fix is in guidance for employees, and possibly in a Brand Skill that sets sensible defaults, so that a thin prompt still produces a reasonable structure.

Key takeaways

  • Reviewing every generated deck is impossible; reviewing none lets problems scale.
  • Shift from inspection (fix this deck) to diagnosis (fix the cause).
  • Sample real decks and rerun a fixed prompt set; grade readiness, tag failures, count them.
  • Fix the largest pattern with the smallest system change, verify it, and report briefly.
  • Keep individual review for high-stakes, regulated and brand-new use cases.

Want a baseline to start reviewing against? Have your template scored.**

Related reading

Sources

Bad Copilot output? Find out whether it’s the template, the prompt or the model

When a generated slide comes out badly, people tend to reach for one of two explanations. The first is “Copilot isn’t good enough.” The second is “the user wrote a bad prompt.” Sometimes one of them is right. Often neither is, and the real cause โ€” a missing example slide, a placeholder sized wrong, a rule nobody wrote down โ€” goes unfixed and repeats in thousands of decks.

Diagnosis matters because each cause has a different fix, owned by a different person. This post gives a practical way to sort failures by cause.

Table of nine failure codes for generated slides, each with its symptom and where the fix usually belongs

The categories

We sort generated-slide failures into nine types. Each has a letter so they can be tallied quickly during review.

Code Failure Typical symptom
T Technical Text overflows, image cropped, content in the wrong placeholder
B Brand Wrong colors, fonts, imagery or visual language
H Hierarchy The important thing isn’t the most visible thing
S Structure Wrong way to communicate the content โ€” e.g. bullets where a comparison was needed
D Density Too much or too little for the layout
V Visualization Poor chart choice or poorly presented data
I Intent Deck doesn’t achieve what it was meant to
P Prompt The instruction was insufficient or unclear
A AI constraint A limit of the current model that no template change fixes

Where each usually comes from

Each failure type tends to point to a particular part of the system.

T โ€” Technical โ†’ layouts and placeholders. Microsoft’s guidance says Copilot uses placeholder type, position and size to decide what goes where. Overflow usually means a placeholder sized for idealized content; cropping usually means proportions or overlapping decoration; content in the wrong slot often means text boxes where placeholders should be. Owner: template builder.

B โ€” Brand โ†’ theme, styles or Brand Kit. Default colors on a new chart point to a theme that doesn’t hold the brand. Wrong images point to the Brand Kit or asset library. Wrong fonts point to theme fonts. Owner: template builder and brand manager.

H โ€” Hierarchy โ†’ example slides. Copilot learns emphasis from examples. If generated slides consistently bury the key point, the examples may not show strong hierarchy โ€” or may not exist for that slide type. Owner: designers.

S โ€” Structure โ†’ missing layouts or examples, sometimes rules. Content forced into the nearest available layout suggests the right layout doesn’t exist, or has no example showing its purpose. With strict brand adherence on, Copilot can’t create a new layout to compensate. Owner: template builder and designers.

D โ€” Density โ†’ examples and rules. If the template shows only one density per slide type, Copilot applies it to everything. Microsoft recommends examples at light, medium and heavy. Note instructions (“Keep the text minimal”) and Brand Skills can reinforce limits. Owner: designers and brand manager.

V โ€” Visualization โ†’ theme, styles and data examples. Chart colors come from theme accents; chart appearance from styles; chart choices and takeaway titles from data examples and rules. Owner: designers.

I โ€” Intent โ†’ prompt, source or person. A deck that is well-formed but misses the point usually lacked information about audience and purpose, or its outline wasn’t reviewed before slides were generated. This is rarely a template problem. Owner: the person creating the deck.

P โ€” Prompt โ†’ training and rules. If the same prompt weakness recurs across many people, it’s worth addressing in guidance for employees โ€” and sometimes in a Brand Skill that sets sensible defaults so a thin prompt still produces a reasonable structure. Owner: brand manager and whoever runs enablement.

A โ€” AI constraint โ†’ acceptance, review, or escalation. Some failures persist whatever you change. Microsoft is clear that Copilot’s output should be reviewed by a person. Recognizing a genuine model limit prevents endless template tinkering that won’t help. Owner: review process; revisit when Microsoft updates the product.

A diagnostic routine

For each failed slide:

  1. Name the symptom โ€” what exactly is wrong?
  2. Assign a code. If more than one applies, pick the most fundamental. A structure failure often causes density and hierarchy failures downstream.
  3. Check the prompt and source first. Rerun the same request with a clearer prompt. If that fixes it, it’s P or I, not the template.
  4. Check the example set. Is there a finished example for this kind of slide at this density? If not, that’s the likely cause of H, S, D or V.
  5. Check the layout. Is there a layout for this job, with correctly typed and sized placeholders? If not, it’s S or T.
  6. Check the theme and kit. For any B or V failure involving colors, fonts or images.
  7. Try the smallest system change that would prevent the failure, and rerun the full prompt set.
  8. If nothing helps, classify as A and make sure review catches it.

Count the failures

The real value comes from tallying. After reviewing a meaningful batch of output โ€” a full run of your test prompt set, or a sample of real generated decks โ€” count failures by code. The distribution tells you where to invest:

  • Mostly T, S and D? The template’s structure and examples need work. This is usually the highest-leverage fix, because it improves every deck.
  • Mostly B and V? Theme, styles and Brand Kit.
  • Mostly P and I? Employee guidance, outline review habits and sensible defaults in Brand Skills.
  • A significant share of A? Set expectations, strengthen review, and retest when Microsoft ships changes.

We don’t publish a “typical” distribution, because it depends heavily on the template and the organization โ€” and because the point is to learn yours. Count, fix the largest category, rerun, count again.

Key takeaways

  • Different causes need different fixes, owned by different people; diagnose before fixing.
  • Sort failures into nine types: technical, brand, hierarchy, structure, density, visualization, intent, prompt and AI constraint.
  • Check the prompt first, then examples, then layouts, then theme and kit.
  • Tally failures by type to see where to invest, and fix the system rather than the slide.
  • Recognize genuine model limits, and make sure review catches them.

Want to know what’s behind your generated decks’ problems? Have your template scored.**

Related reading

Sources

How to test a Copilot template before your company uses it

New templates have traditionally been tested the same way: a few people open it, make a couple of slides, and agree it looks good. That was a reasonable test when the template’s only users were people.

It is not a reasonable test for a template Copilot will learn from. Copilot generates presentations from prompts nobody on the template team will ever see, using source material nobody chose, for audiences nobody anticipated. The only way to know how the template behaves is to generate a lot of presentations deliberately and look at them systematically.

This post lays out a method. It is simple enough for a small team to run and rigorous enough to tell you whether a change actually helped.

Diagram of six steps for testing a Copilot template: build the prompt set, run the baseline, score, diagnose, change one thing and rerun, keep the record

Principle: the same prompts, every time

The most important rule is also the easiest to break. Use a fixed set of prompts, and reuse exactly the same set every time you test.

Without a fixed set, you can’t compare. Did the new example slides improve the output, or did you just happen to write easier prompts this time? Did Microsoft’s latest change make things worse, or were today’s prompts harder? A stable test set turns opinions into comparisons.

Step 1: Build the prompt set

Aim for somewhere around fifteen to twenty-five realistic scenarios โ€” enough to cover the variety of what your people make, few enough to review properly. Draw them from real work: recent presentation requests, common deck types, recurring meetings.

Deliberately vary the dimensions that stress a template:

Dimension Range to cover
Density light, medium, heavy
Intent inform, compare, recommend, explain, persuade
Content text, data, process, timeline, imagery, mixed
Audience internal team, leadership, executive, external
Source quality well structured, messy, incomplete
Source type prompt only, a document, existing slides
Stakes routine, elevated

For each prompt, record a few fields so the test is repeatable and reviewable:

  • ID and business scenario (“monthly operations update for regional leads”);
  • the exact prompt, word for word;
  • source material, if any, saved alongside;
  • intended outcome โ€” what a good deck would achieve;
  • expected behavior โ€” which layouts you’d expect, roughly how many slides, what density.

The goal isn’t to write perfect prompts. It’s to write representative ones. Include the vague, rushed prompts real people write, not only the carefully crafted ones.

Step 2: Run the baseline

Before changing anything, run every prompt against the current template and save every output. Don’t fix anything yet. This baseline is what every future result will be compared with.

If you’re rebuilding a template, run the baseline against the old one. If you’re building from scratch, run it against the first version that follows Microsoft’s published guidance. Microsoft’s recommendations โ€” representative sample slides at three densities, correct placeholders, theme-defined colors and fonts โ€” are the floor; your testing will show what’s needed above it.

Step 3: Score what comes out

Review each generated deck on four dimensions:

Technical behavior. Did Copilot choose appropriate layouts? Did content land in the right placeholders? Did it fit? Did images and charts behave?

Brand behavior. Correct colors, typography, imagery and visual language? Appropriate density? Would someone recognize it as your organization’s deck?

Communication behavior. Is the hierarchy right? Is the most important point obvious? Does the structure match the intent? Is it understandable without a presenter?

Readiness. Assign each deck one of four grades:

  1. Ready โ€” usable as generated.
  2. Minor corrections โ€” cleanup only.
  3. Remediation โ€” meaningful design correction needed.
  4. Reconceptualization โ€” the structure is wrong; start over.

The difference between grades 2 and 4 matters enormously. A slide that needs alignment fixes is a small problem. A slide where Copilot chose the wrong way to communicate the content is a different kind of problem, usually with a different cause.

Step 4: Diagnose, don’t just fix

When output is poor, the instinct is to fix the slide. Resist it. Ask what caused the failure:

  • the template’s architecture (layouts, placeholders);
  • a missing or misleading example slide;
  • a density mismatch;
  • a brand rule missing or unclear;
  • the prompt itself;
  • ambiguity in the source material;
  • a limitation of the model that no template change will fix.

Then ask: what’s the smallest change to the system that would prevent this failure from recurring? A new example, a resized placeholder, a note instruction, a rule in a Brand Skill. Fixing the system helps every future deck; fixing a slide helps one.

Step 5: Change one thing, rerun everything

Make the change. Rerun the entire prompt set, not just the prompt that exposed the problem. Then compare with the previous run:

  • Did the change fix the target problem?
  • Did it improve similar slides elsewhere?
  • Did it make anything worse?

That last question catches a common trap. A new example that fixes comparisons can start pulling unrelated content into the comparison layout. Only a full rerun reveals it.

Change one thing at a time where you can. If you change five things and the output improves, you don’t know which one helped โ€” or whether one of them hurt.

Step 6: Keep the record

Keep every run’s outputs and scores. Over time the record becomes valuable in its own right:

  • it shows which changes actually improved output, and which were theory;
  • it separates Microsoft’s rules (published guidance) from your own techniques (proven by your tests);
  • it gives you a baseline to compare against when Microsoft changes the product.

When to rerun

  • Before switching on strict brand adherence.
  • After any change to the template, examples, Brand Kit, note instructions or Brand Skills.
  • After significant Microsoft updates โ€” two arrived in September 2026 alone.
  • On a regular schedule, at least quarterly, because behavior can drift even when nothing on your side changed.

Key takeaways

  • Use a fixed, representative prompt set of roughly fifteen to twenty-five scenarios, and never change it casually.
  • Run a baseline before changing anything.
  • Score technical, brand and communication behavior, plus a four-level readiness grade.
  • Diagnose the cause of each failure, and fix the system rather than the slide.
  • Change one thing at a time, rerun everything, and keep the record.

Want your template tested this way? Have your template scored.**

Related reading

Sources

Reporting Copilot presentation savings to leadership

Six months after a Copilot rollout, someone senior will ask what it’s saving. For presentations, the easy answer is an adoption chart: more people using Copilot, more often. It’s also the answer most likely to disappoint, because adoption measures activity, not value โ€” and a leader who hears “usage is up again this quarter” will reasonably ask “and what did we get for it?”

This post describes a simple, honest way to report presentation value from Copilot: four measures, a one-page quarterly format, and some rules for handling numbers so the report holds up when someone checks.

The four measures

1. Fix-up time

How much work generated decks need before they’re usable. Score a sample each quarter on a five-level scale:

  • 0 โ€” usable as generated;
  • 1 โ€” light cleanup;
  • 2 โ€” skilled correction;
  • 3 โ€” senior judgment needed;
  • 4 โ€” rebuild.

Report the distribution and the average, compared with last quarter and with the baseline before the template was rebuilt. This is the measure that shows whether Copilot is actually saving time on presentations, because it captures the correction that eats into drafting savings.

2. Routine presentation spending

Spending on routine decks with agencies, contractors and overtime, compared with the same period before rollout. This is the measure most likely to be cash rather than capacity, and the one finance cares about most.

Be careful with attribution: spending can change for many reasons โ€” budgets, reorganizations, fewer projects. Report the change and name other factors you know about.

3. Where design time goes

The share of design team or agency time spent on routine decks versus high-stakes presentations and system work. If Copilot is working, the routine share should fall and the high-stakes share should hold or rise. This shows value that isn’t a cost reduction: better attention on the presentations that matter.

4. System health

A short status of the system that produces generated decks:

  • results of the quarterly re-test with the fixed prompt set;
  • changes made to the template, examples and rules;
  • Microsoft changes this quarter and their effect (September 2026 brought strict brand adherence, note instructions and Brand Skills);
  • any generated deck problems that reached external audiences.

Adoption: include it, don’t lead with it

Adoption still belongs in the report as context. If fix-up is improving but only a handful of people use Copilot for presentations, the total value is limited โ€” and that’s worth knowing. But it’s a supporting measure. Leading with it invites the “so what?” question.

A one-page quarterly format

Headline โ€” one sentence. “Routine decks now mostly need light cleanup; routine agency spend is down; two template changes this quarter.”

Fix-up โ€” distribution this quarter vs. last quarter vs. baseline, in a small chart or table.

Spending โ€” routine presentation spend this quarter vs. the same quarter before rollout, with known confounders named.

Design time โ€” share of design/agency time by category.

System โ€” re-test result, changes made, Microsoft changes, incidents.

Next quarter โ€” the top one or two system fixes planned, and why.

One page. Leaders read it. Longer reports get skimmed.

Rules for handling numbers honestly

Measure before claiming. Report what you measured. If something is an estimate, label it.

Separate capacity from cash. Saved employee time is capacity. Reduced agency invoices are cash. Say which is which.

Show the baseline. Every number needs a comparison: before rollout, before the rebuild, last quarter.

Name confounders. Reorganizations, budget changes, product launches. A number that ignores obvious confounders loses credibility when someone else points them out.

Cite external research with its conditions if you use it at all. The most-cited study โ€” Forrester’s Total Economic Impactโ„ข of Microsoft 365 Copilot, commissioned by Microsoft โ€” reported 34.2% content-creation time savings for a composite organization, and Forrester advises readers to use their own estimates. It can frame expectations; it shouldn’t stand in for your results.

Report bad news early. If fix-up rose because a Microsoft change altered behavior, say so, and say what you’re doing about it. A program that reports problems is trusted when it reports successes.

Why this approach works

It answers the question leadership is actually asking โ€” is this worth it? โ€” with evidence rather than activity. It separates cash from capacity, which finance respects. It shows the system is being looked after, which reduces the risk leadership worries about most: an embarrassing deck reaching a customer. And it builds a record that makes future decisions easier, whether that’s expanding licenses, investing further in the template, or changing how the organization buys presentation design.

Common questions

What if fix-up time goes up?
Report it, explain the likely cause, whether a Microsoft change, a new deck type or template drift, and say what you’re doing about it. Reporting problems early is what makes good news believable.

How do we separate Copilot’s effect from other changes?
You often can’t fully. Name the other factors you know about, compare against a baseline, and avoid claiming more precision than the data supports.

Should we report adoption at all?
Yes, as context. If only a few people use Copilot for presentations, even great fix-up results add up to limited value.

Key takeaways

  • Adoption measures activity; leadership wants value.
  • Report four measures: fix-up time, routine presentation spending, where design time goes, and system health.
  • Keep it to one page each quarter, with a one-sentence headline.
  • Measure before claiming, separate cash from capacity, show baselines, name confounders, and cite outside research with its conditions.
  • Report problems early; it makes the successes believable.

Need a baseline to report against? Have your template scored.**

Related reading

Sources

Building the business case for a Copilot-ready template

Your organization has bought Microsoft 365 Copilot licenses. Now you’re asking for more money โ€” to rebuild a template that, to most of the finance team, already exists and looks fine. That’s a harder conversation than the license was, because the value of the template is less visible than the value of the tool.

This post is a structure for that business case: what to include, what numbers to use, and how to make it credible to a skeptical reader. It’s built on one principle: use your own numbers, cite outside evidence with its conditions, and commit to measuring the result.

1. The problem, in one paragraph

Lead with the problem, not the solution. Something like:

We’ve licensed Copilot for [number] people. Copilot in PowerPoint learns from our template โ€” Microsoft describes the template’s sample slides as its primary learning source, and says that without enough of them Copilot “falls back to the Slide Master โ€” which may result in less optimal output.” Our current template was built for people, not for a model: [findings โ€” e.g. brand colors hardcoded rather than in the theme, layouts built from text boxes, no example slides]. As a result, generated decks need significant correction before use, which reduces the return on our licenses.

The findings should come from an actual review of your template against Microsoft’s published guidance. Specifics persuade; generalities don’t.

2. Why now

Finance teams reasonably ask why this can’t wait. Good reasons:

  • The licenses are already being paid for. Every month the template teaches poorly is a month of reduced return on an existing cost.
  • Microsoft now lets us lock Copilot to the template. Strict brand adherence, announced in September 2026, restricts Copilot to approved layouts. That’s a strong control โ€” and it makes every gap in the template apply to every locked deck.
  • Volume is growing. As more people generate more decks, the template’s flaws are multiplied more often.

3. The options

Show that you’ve considered alternatives:

Option Cost Outcome
Do nothing None up front Generated decks keep needing heavy correction; flaws multiply as usage grows
Train people to prompt better Training time Helps with argument and structure; doesn’t fix what the template teaches
Configure settings only Minimal Brand Kit and settings in place; output still limited by the template
Rebuild the template for Copilot Project cost + upkeep Addresses the cause; improves every future generated deck

The key point: training and configuration are useful and cheap, but neither changes what Copilot learns from. Only the template does.

4. The numbers

This is where business cases lose credibility, usually by borrowing someone else’s savings figure. Instead:

Estimate current presentation effort. Decks per year by type โ€” routine, important internal, high-stakes โ€” and hours per deck today, including designer and agency time.

Estimate fix-up time with the current template. Ideally from a small pilot: a few people generate real decks and record how much correction each needed. This is the number the template affects most.

Estimate fix-up time after the rebuild. Use a conservative assumption, clearly labeled as such, and commit to measuring the real figure.

Value the difference using a loaded hourly cost, and be explicit about what’s capacity and what’s cash. Reduced agency or contractor spending on routine decks is cash; freed employee time is capacity.

Subtract the cost of the rebuild, written rules, testing and a year of upkeep.

Show a range โ€” conservative, expected, optimistic โ€” rather than a single figure.

If you want to reference outside research, cite it with its conditions. The most-cited study, Forrester’s Total Economic Impactโ„ข of Microsoft 365 Copilot, reported 34.2% time savings in content creation for a composite organization โ€” but it was commissioned by Microsoft, it describes a modeled organization rather than a real one, it gives no PowerPoint-specific figure, and Forrester itself “strongly advises that readers use their own estimates.” It’s supporting context, not your forecast.

5. The risks

A credible case names its risks:

  • Adoption. If people don’t use Copilot for presentations, the template’s value is limited. Mitigation: pilot group, short guidance, measure usage alongside fix-up.
  • Platform change. Microsoft updates Copilot continuously โ€” twice in September 2026 on template behavior alone. Mitigation: quarterly re-testing and upkeep, budgeted.
  • High-stakes misuse. Generated drafts used for decks that need experienced people. Mitigation: clear guidance on which decks go to Copilot.
  • Measurement. Savings claimed but not proven. Mitigation: the measurement plan below.

6. How you’ll prove it

End with a commitment finance can hold you to:

  • Baseline: fix-up scores for a fixed set of realistic prompts, and for a sample of real generated decks, before the rebuild.
  • After: the same measures after the rebuild.
  • Ongoing: quarterly re-testing and a sample of real decks, reported in writing.
  • Spending: routine presentation spending with agencies and contractors tracked before and after.

A business case that promises measurement is more persuasive than one that promises savings, because it doesn’t ask anyone to take the numbers on faith.

Common questions

What if we can’t run a pilot before the business case?
Use conservative assumptions, clearly labeled, and make measurement a condition of the investment. A case that commits to measuring is more credible than one that claims certainty.

Should we include the Forrester figures?
As context, with their conditions: commissioned by Microsoft, a composite organization, no PowerPoint-specific figure. Don’t use them as your forecast.

How big should the ask be?
Start with an audit if scope is uncertain. It turns the rebuild from an estimate into a plan, and it gives finance a smaller first decision.

Key takeaways

  • Lead with the problem, grounded in a real review of your template against Microsoft’s guidance.
  • Explain why now: licenses are already paid for, strict brand adherence raises the stakes, and volume is growing.
  • Show the options; only the template changes what Copilot learns from.
  • Use your own numbers, focus on fix-up time, separate cash from capacity, and show a range.
  • Name the risks, and commit to measuring the result.

Need findings for your business case? Have your template scored.**

Related reading

Sources

What the research says about Copilot time savings โ€” and what it doesn’t

If you’re building a case for Microsoft 365 Copilot, you’ll quickly meet a handful of figures that appear in almost every vendor deck and article: a triple-digit ROI, hours saved per user per month, a payback period measured in months. Most trace back to one family of studies: Forrester’s Total Economic Impactโ„ข (TEI) research, commissioned by Microsoft.

These studies are useful. They’re also frequently quoted in ways their authors wouldn’t endorse. This post walks through what the main study found, how it was produced, and what it can and can’t tell you about presentations specifically.

The study

The Total Economic Impactโ„ข of Microsoft 365 Copilot, published by Forrester Consulting in March 2025, is the most-cited source. Its method, as Forrester describes it:

  • interviews with 16 decision-makers across 12 organizations using Microsoft 365 Copilot;
  • a survey of 367 respondents at organizations using it;
  • a composite organization built from those inputs: $6.25 billion in annual revenue, 25,000 employees, with Copilot rolled out to 40% of the workforce over three years.

The headline findings

For that composite organization, over three years, Forrester reports:

  • ROI of 116%;
  • net present value of $19.7 million;
  • benefits of $36.8 million against costs of $17.1 million (present value);
  • payback in 10 months.

On time, the study reports that the average user saves 9 hours per month, and gives task-level time savings including:

  • content creation: 34.2%;
  • information search: 29.8%;
  • meeting summarization: 18.6%.

It also reports faster onboarding for new hires โ€” up to 25%, or 11 days per employee.

A separate Forrester study for small and medium businesses, published in October 2024, projected a three-year ROI range of 132% to 353%.

How to read those numbers

The figures are real outputs of a careful method. They also come with conditions that should travel with them whenever they’re quoted.

It’s commissioned research. Forrester states that the study was commissioned by Microsoft. That doesn’t make it wrong. It does mean it’s the vendor’s evidence, and a reader should weigh it as such.

It’s a composite, not a customer. The financial results describe a modeled organization assembled from interviews and survey data, not a single real company. Your organization’s size, rollout, roles and adoption will differ.

It’s risk-adjusted. Forrester applied downward adjustments to account for adoption and integration uncertainty. That’s good practice, and a reminder that the unadjusted estimates were higher.

Forrester tells you to use your own numbers. The study states that Forrester “strongly advises that readers use their own estimates within the framework provided.” The framework is the durable part; the composite’s results are an illustration.

The SMB figures are projected. The small and medium business study describes projected ranges, not measured outcomes.

What it says about presentations

Here’s the specific point for anyone focused on presentations: the main study doesn’t report a PowerPoint-specific figure. Presentation work is part of what “content creation” covers, and 34.2% is the relevant number in that sense. But it’s an average across content-creation tasks, not a measurement of how long generated presentations take to finalize.

That matters because presentations have an unusual cost profile. The time to produce a first draft can fall sharply with Copilot. The time to finish the deck โ€” correcting structure, emphasis, layout and brand, and checking facts โ€” depends heavily on the template Copilot learned from and on the stakes of the deck. A study that averages across content tasks can’t tell you how that balance works out in your organization.

What it can’t tell you

No external study can answer the questions that decide your return on Copilot for presentations:

  • How much fix-up time do your generated decks need? That depends on your template, examples and rules.
  • What share of your presentations are routine? Routine decks benefit most; high-stakes decks still need experienced people.
  • How deep are your licenses? Savings accrue to people who create documents and decks; licensing roles that rarely do changes the math.
  • What does review cost? Microsoft states that generated presentations should be human-reviewed. That time is real.

How to use the research responsibly

Quote it with its conditions. “A Microsoft-commissioned Forrester study of a composite organization reported 34.2% time savings in content creation” is accurate. “Copilot saves 34% of presentation time” is not.

Use the framework, not the result. Forrester’s own advice is to put your estimates into its framework. Do that.

Measure your own baseline. Before rollout, estimate how much time your people spend on presentations and how long typical decks take. After rollout, measure again โ€” including the time spent fixing generated drafts.

Separate draft time from finish time. The biggest risk in presentation ROI is counting the fast first draft and forgetting the correction that follows.

We follow the same rule ourselves: we cite the studies with their conditions, and we don’t quote savings figures of our own until we’ve measured them.

Common questions

Is commissioned research unreliable?
Not necessarily. Forrester’s methods are documented. But commissioned research is the vendor’s evidence, and its results describe a composite organization. Treat it as context, not as a forecast for you.

Is there research specific to PowerPoint?
The main Forrester study doesn’t report a PowerPoint-specific figure. Content creation, at 34.2%, is the closest category, and it’s an average across many content tasks.

What should we measure ourselves?
Time per deck before rollout, then draft time and fix-up time after, by deck type. Fix-up time is the part external studies can’t tell you.

Key takeaways

  • The most-cited Copilot ROI figures come from Forrester TEI studies commissioned by Microsoft.
  • The main study reports 116% ROI, 9 hours saved per user per month and 34.2% content-creation time savings for a composite organization.
  • It’s commissioned, composite and risk-adjusted, and Forrester advises readers to use their own estimates.
  • It gives no PowerPoint-specific figure; presentation value depends on finish time, not just draft time.
  • Quote it with its conditions, use its framework, and measure your own baseline.

Want to know how much finishing your generated decks will need? Have your template scored.**

Related reading

Sources

Estimate your own presentation savings: a model you can fill in

Most estimates of what Copilot saves on presentations start with someone else’s number โ€” a percentage from a study, a vendor’s claim, a colleague’s anecdote โ€” and multiply it by your headcount. The result is precise-looking and almost entirely borrowed.

Forrester, whose Microsoft-commissioned study is the most-cited source of Copilot savings figures, advises the opposite: readers should “use their own estimates within the framework provided.” This post gives you a simple framework for presentations specifically, with the one term most estimates leave out.

We don’t supply the inputs. They’re yours to measure or estimate, because the answer depends on your organization, your people and your template. The worked example at the end uses invented round numbers purely to show the arithmetic.

Diagram of the savings formula: hours saved per deck times decks per year times loaded hourly cost, minus the cost of running the system; hours saved equals hours today minus draft and fix-up time

The model in one line

Annual net saving = (hours saved per deck ร— decks per year ร— loaded hourly cost) โˆ’ annual costs of running the system

Everything below is about estimating those terms honestly.

Step 1: Count the decks, by type

Not all presentations behave the same way under Copilot, so split them. At minimum:

Deck type Examples Decks per year
Routine Status updates, briefings, team meetings, internal proposals ____
Important internal Leadership reviews, planning decks, cross-functional proposals ____
High-stakes external Board, investor, major pitch, keynote ____

How to estimate: sample a few teams for a month, or use the number of PowerPoint files created in a typical month if your IT team can report it. Don’t try to be perfect; be consistent.

Step 2: Time per deck today

For each type, estimate the hours a typical deck takes today, from start to “ready to use” โ€” including any designer or agency time.

Deck type Hours per deck today
Routine ____
Important internal ____
High-stakes external ____

How to estimate: ask a handful of frequent presenters and your design team. Ranges are fine; use the midpoint.

Step 3: Time per deck with Copilot โ€” split into two parts

This is where most estimates go wrong. Time with Copilot has two components:

  • Draft time: prompting, refining the outline, generating slides.
  • Fix-up time: everything needed to make the generated deck usable โ€” correcting structure, emphasis, layout, brand and density, and checking facts.
Deck type Draft hours Fix-up hours Total with Copilot
Routine ____ ____ ____
Important internal ____ ____ ____
High-stakes external ____ ____ ____

Fix-up time is the term that makes or breaks the case, and it depends heavily on the template Copilot learns from. Microsoft’s guidance is that Copilot learns primarily from a template’s sample slides, and falls back to the Slide Master when there aren’t enough โ€” “which may result in less optimal output.” A template with good examples, real placeholders and a theme-defined brand produces drafts that need less fixing. A template without them can push fix-up time close to the time saved on drafting.

How to estimate: run a pilot. Have a few people produce real decks with Copilot and record both components. If you can’t pilot yet, estimate fix-up time conservatively.

For high-stakes decks, be cautious. These are hard because of argument and stakes, not formatting. Copilot can speed up a first draft, but assume the judgment-heavy work remains.

Step 4: Hours saved per deck

For each type: hours today โˆ’ total hours with Copilot. If the answer is negative or near zero for a type, that’s a finding in itself โ€” usually about the template, or about deck types that shouldn’t be generated.

Step 5: Put a value on the hours

Use a loaded hourly cost โ€” salary plus benefits and overhead โ€” for the people who make each deck type. Your finance team will have a standard figure. If designers or an agency handle part of the work, include their cost for that part.

Be honest about what saved hours mean. They’re capacity, not cash, unless they reduce spending โ€” agency fees, overtime, contractors โ€” or are redirected to work with measurable value. A business case that treats every saved hour as cash will be challenged, rightly.

Step 6: Subtract the costs of running the system

Annual costs to include:

  • Licenses attributable to presentation work, if they’re being justified by it. (Microsoft’s US list price for Microsoft 365 Copilot enterprise is $30.00 per user per month paid yearly, as of 22 September 2026, on top of a qualifying Microsoft 365 plan.)
  • Template work, amortized over its useful life.
  • Written rules โ€” note instructions and Brand Skills โ€” and their testing.
  • Review time not already in fix-up.
  • Upkeep โ€” a quarterly cycle of re-testing and updating.

Worked example (invented numbers)

To show the arithmetic only. These inputs are made up. They are not benchmarks, not measurements, and not a prediction for any organization.

Routine decks only:

  • Decks per year: 1,000
  • Hours per deck today: 4
  • With Copilot: 1 hour drafting + 1.5 hours fix-up = 2.5 hours
  • Hours saved per deck: 1.5
  • Loaded hourly cost: $60

Gross value: 1,000 ร— 1.5 ร— $60 = $90,000 of capacity per year.

Now change one input. Suppose the template teaches Copilot poorly and fix-up rises to 2.5 hours. Total with Copilot becomes 3.5 hours; hours saved per deck falls to 0.5; gross value falls to $30,000. Same licenses, same people, a third of the value โ€” because of fix-up time.

That sensitivity is the point of the model. Fix-up time is one of its most sensitive inputs, and the template Copilot learns from is one of the main things that drives it.

Keep it honest

  • Measure, then re-measure. Record a baseline before rollout and repeat the measurement after. Replace estimates with data as it arrives.
  • Show ranges. Give low, middle and high cases rather than a single number.
  • Separate capacity from cash. Say which savings reduce spending and which free time.
  • Cite external studies with their conditions. They can frame expectations; they can’t replace your inputs.

Key takeaways

  • Use your own inputs in a simple framework, as Forrester itself advises.
  • Split decks by type; routine decks and high-stakes decks behave very differently.
  • Split time with Copilot into draft time and fix-up time โ€” fix-up is where estimates fail.
  • Value saved hours honestly as capacity unless they reduce spending.
  • Fix-up time is one of the most sensitive inputs, and the template is one of the main things that drives it.

Want a realistic read on your fix-up time? Have your template scored.**

Related reading

Sources

What Copilot for PowerPoint costs in 2026

There’s no such thing as “Copilot for PowerPoint” on Microsoft’s price list. Copilot in PowerPoint comes as part of Microsoft 365 Copilot, licensed per user on top of a qualifying Microsoft 365 plan. So the question “what does it cost to use Copilot for presentations?” has two answers: the license price, which is published, and the cost of making it produce presentations you’d actually use, which isn’t.

This post covers both. License figures are Microsoft’s US list prices as published on 22 September 2026. Prices, bundles and promotions change, and your organization may have negotiated terms, so check current figures with Microsoft or your reseller before budgeting.

The license

Enterprise

Microsoft 365 Copilot for enterprise customers is listed at $30.00 per user per month, paid yearly โ€” or $31.50 per user per month paid monthly on an annual commitment. It’s an add-on: Microsoft states that “a separate license for a qualifying Microsoft 365 plan is required.”

Business

For smaller organizations, Microsoft lists business plans (check Microsoft’s page for the current seat limits):

  • Microsoft 365 Copilot Business (add-on): $21.00 per user per month at the regular price, with a promotional price of $18.00 per user per month paid yearly for the first year, available until 31 December 2026. Paid monthly, $25.20. It also requires a qualifying Microsoft 365 plan.
  • Bundles combining a Microsoft 365 plan with Copilot: Business Standard with Copilot at $23.50 and Business Premium with Copilot at $32.00 per user per month, paid yearly.

What’s included for presentations

Copilot in PowerPoint lets licensed users create presentations from prompts and source material, refine them in conversation, and work with an organization’s templates and Brand Kit. Microsoft’s documentation lists support in PowerPoint for Microsoft 365 on Windows and Mac and on iPad, subject to organizational settings. Brand Kit features, including official kits managed by brand managers, carry their own licensing notes in Microsoft’s documentation โ€” its official Brand Kit page, for example, names a “Microsoft Copilot Premium license” as the requirement. License names change often, so confirm exactly which license your organization needs with your Microsoft administrator.

What a license budget looks like

Licenses are simple arithmetic. As an illustration only โ€” not a recommendation for any particular organization:

Seats Enterprise list ($30/user/month) Per year
100 $3,000 per month $36,000
500 $15,000 per month $180,000
2,000 $60,000 per month $720,000

Those are the add-on costs alone, before the base Microsoft 365 plans. You don’t have to license everyone at once; one approach is to start with the roles that create the most documents and presentations, and expand from there.

The costs that aren’t on the price list

Here’s the part that surprises budget holders. A license makes Copilot available. It doesn’t make generated presentations good. The gap between the two has its own costs, and they’re worth planning for rather than discovering.

1. Setting up the brand

Designating brand managers, building an official Brand Kit, connecting the SharePoint asset library, deciding template settings. Mostly internal time from brand and IT. Modest in size, but it needs to be someone’s job.

2. Making the template teachable

This can be the largest one-time cost, and it’s the easiest to skip. Copilot learns from a template’s example slides, placeholders and theme; Microsoft says that without enough sample slides it falls back to the Slide Master, “which may result in less optimal output.” Many corporate templates need rebuilding for how Copilot reads them: brand colors moved into the theme, layouts rebuilt with real placeholders, and a set of example slides designed at the densities people actually use. It’s design work โ€” done in-house or with outside help.

3. Written rules

Note instructions in template speaker notes and Brand Skills published through Brand Kit need writing, testing and governance review. Microsoft recommends reviewing Brand Skills for security, privacy, accessibility, legal and brand before publishing, and auditing them periodically.

4. Enablement

Short, practical guidance for employees: where to start, what Copilot is for, and the two habits that matter most โ€” briefing Copilot on audience and purpose, and fixing the outline before generating slides.

5. Review time

Microsoft states that generated presentations “should be human-reviewed and edited accordingly.” Review isn’t free, and it’s where weak templates cost most: every generated deck that needs heavy correction consumes time the license was supposed to save.

6. Upkeep

The platform changes โ€” Microsoft made two significant template-related changes in September 2026 alone โ€” and organizations change. Templates, examples and rules need re-testing and updating, at least quarterly.

Why the hidden costs matter to the license decision

The license is a fixed cost per seat. The value of each seat depends heavily on the hidden costs being handled well. A licensed employee working from a template that teaches Copilot well gets usable first drafts. The same employee working from a template that teaches it nothing gets generic slides and spends time fixing them.

That’s why we’d encourage anyone budgeting for Copilot to put the template in the same conversation as the licenses. The template is a one-time investment, with upkeep, that affects the return on every seat.

A simple budgeting checklist

  • โ˜ License cost: seats ร— current list or negotiated price, plus any base-plan upgrades.
  • โ˜ Brand setup: named brand managers and time allocated.
  • โ˜ Template: audit against Microsoft’s guidance; rebuild if needed.
  • โ˜ Rules: note instructions and Brand Skills written, tested and reviewed.
  • โ˜ Enablement: short guidance and a pilot group.
  • โ˜ Review: who reviews which decks, and how much time that takes.
  • โ˜ Upkeep: a quarterly cycle of re-testing and updating.

Key takeaways

  • Copilot in PowerPoint comes with Microsoft 365 Copilot, licensed per user on top of a qualifying Microsoft 365 plan.
  • US list prices on 22 Sep 2026: $30.00 per user per month (enterprise, paid yearly); $21.00 for Copilot Business, with an $18.00 first-year promotion through 31 Dec 2026.
  • The license makes Copilot available; the template, rules, review and upkeep decide whether its presentations are usable.
  • Budget the template alongside the licenses โ€” it affects the return on every seat.

Budgeting for Copilot? Have your template scored so you know what the template side will need.

Related reading

Sources