Copilot Leaves Office Apps April 15. Pay or Pivot?

In 48 hours, the AI features inside your Word, Excel, PowerPoint, and OneNote apps either cost money or disappear.

Microsoft confirmed in a March 18 announcement that starting April 15, 2026, Copilot Chat will be removed from those four Office apps for users without a paid M365 Copilot license. The free tier gets rebranded to “Copilot Chat (Basic)” and pushed to a standalone app. The in-app experience becomes paid-only under the “M365 Copilot” label.

If your team built any workflows around Copilot inside Word or Excel over the past few months, you’re about to lose access to them unless you pay up. And the bill isn’t small.

What Changes on April 15

	Before April 15	After April 15 (No License)	After April 15 (Paid License)
Copilot in Word	Available	Removed	Full access
Copilot in Excel	Available	Removed	Full access
Copilot in PowerPoint	Available	Removed	Full access
Copilot in OneNote	Available	Removed	Full access
Copilot in Outlook	Available	Still available	Full access
Copilot Chat (standalone app)	Available	Still available	Full access + Work IQ
Branding	”Copilot Chat"	"Copilot Chat (Basic)"	"M365 Copilot”

The one-line version: Outlook and the standalone chat app survive for free. Everything else behind a paywall.

The Fine Print: Organization Size Matters

Here’s the part most coverage has glossed over. The change doesn’t hit every organization the same way.

2,000+ seats: Hard cutoff. Copilot Chat in Word, Excel, PowerPoint, and OneNote goes away entirely for unlicensed users. No workaround. No degraded mode. Gone.

Under 2,000 seats: You keep in-app access, but it gets relabeled as “standard access” with a vague qualifier that “availability may vary during periods of high demand.” Licensed users get priority. Translation: your free Copilot experience is about to get slower, less reliable, and intentionally worse than the paid version sitting right next to it in the same interface.

Microsoft hasn’t published what “standard access” means in measurable terms. That’s deliberate. It gives them room to throttle the experience until the gap between free and paid feels unbearable enough to trigger an upgrade. Classic freemium pressure.

If you run a team of 15 people, you’re technically not losing access on April 15. You’re losing consistent, reliable access. And if you’ve built workflows that depend on Copilot being available when your team needs it, “may vary” is the same as “plan for it to break.”

What It Costs to Keep Copilot

The math depends on your size.

Tier	Price	Who Qualifies
M365 Copilot (Enterprise)	$30/user/month	Organizations on E3/E5 plans
M365 Copilot Business	$18/user/month (promo)	≤300 users, through June 30, 2026
M365 Copilot Business (after promo)	$21/user/month	≤300 users, starting July 1, 2026

These prices sit on top of your existing Microsoft 365 subscription. A business already paying $22/user/month for M365 Business Premium is looking at $40/user/month total to keep Copilot in their Office apps. For a 20-person team, that’s an extra $4,320 to $7,200 per year depending on your tier.

Microsoft extended the $18 promotional pricing through June 30, which tells you everything about how this conversion is going. If demand were strong at $30, they wouldn’t still be running a 40% discount eight months after launch.

The Pattern You Should Recognize

This is the third major Copilot adjustment in under a month. I covered Microsoft’s March rollback when they pulled Copilot from Notepad, Snipping Tool, Photos, and Widgets after the “Microslop” backlash. That was a surface area reduction — removing AI from places where users didn’t want it.

This April 15 change is different. Microsoft isn’t removing AI from places where it doesn’t fit. They’re removing it from places where it does fit and charging for it. Word, Excel, and PowerPoint are exactly where Copilot should live. Microsoft knows it. That’s why they’re putting the paywall there and not in Outlook (where free access drives habit formation and eventual upgrades).

The sequence tells you where Microsoft’s AI strategy landed:

1. January-February 2026: Copilot everywhere, free for all M365 users, hoping adoption drives upgrade revenue
2. March 2026: Retreat from utilities (Notepad, Photos) where users revolted — 3.3% pull-through rate on 450 million users
3. April 2026: Paywall the productivity apps where Copilot actually adds value, force the conversion

It’s a classic bait-and-switch pattern. Distribute free, build dependency, then monetize. The only question for your business is whether you took the bait.

What Should SMBs Do Before April 15?

Five steps. Start today.

1. Audit Your Actual Copilot Usage

Before you decide whether to pay, find out if anyone’s actually using it. Check the Microsoft 365 admin center usage reports. Look at Copilot interactions per user over the last 30 days.

In my experience, most organizations find the same thing Microsoft found across their entire user base: a small percentage of heavy users and a large majority who tried it once and went back to their normal workflow. If only 4 of your 20 people use Copilot regularly, you need 4 licenses, not 20.

2. Identify Workflows That Break

Make a list of every process that depends on Copilot inside an Office app. Common ones:

• Excel data analysis prompts your finance team relies on
• PowerPoint deck generation from outlines
• Word document summarization or drafting
• OneNote meeting note organization

If these workflows exist, they need a solution by Tuesday. If they don’t exist (be honest), this change costs you nothing and the decision is simple.

3. Price Out the Alternatives

Copilot isn’t the only AI that works with Office documents. Before you commit to $18-30/user/month, compare:

• Claude with the M365 Connector: Anthropic recently made its M365 connector free on all plans, letting Claude pull context from your OneDrive, SharePoint, Outlook, and Teams data in the Claude.ai chat interface. For in-app Excel and PowerPoint work, Claude’s separate add-ins require a Pro ($20/month) subscription with access to Opus-class models.
• ChatGPT Plus with file upload: $20/month. Handles Excel analysis, document review, and presentation content. No native in-app integration, but for many workflows the copy-paste tax is minor.
• Google Gemini Advanced: $20/month. Strong on spreadsheet analysis through Sheets integration. If you’re not locked into the Microsoft ecosystem, this is worth evaluating.
• Open-source options: For technically capable teams, running a local model through something like LM Studio or using an API with a document pipeline costs significantly less per user at scale.

The in-app integration that Copilot offers is genuinely valuable. I’m not pretending the alternatives are identical. But “genuinely valuable” and “worth $30/user/month on top of your existing subscription” are different conversations.

4. License Selectively, Not Universally

If you decide to pay, don’t buy licenses for everyone. Buy them for the people who actually use Copilot in Office apps daily. Your finance lead who runs Excel analysis prompts 20 times a week needs a license. Your office manager who opens Word twice a month does not.

Microsoft’s licensing page lets you assign per-user. Use that granularity. A 20-person company buying 5 licenses at $18/month costs $1,080/year. That same company buying 20 licenses costs $4,320/year. Same productive output. Four times the cost.

5. Build the Workflow Outside the App

This is the structural move that matters most long-term. If your AI workflows live inside Microsoft’s apps, Microsoft controls the pricing, the access, and the experience. That’s exactly the vendor lock-in risk I’ve been writing about.

The better architecture: keep your documents in Microsoft 365 (switching costs are too high for most organizations), but build your AI workflows through tools that sit alongside the documents rather than inside them. An n8n automation that pulls an Excel file, runs analysis through Claude’s API, and drops results back into a shared folder accomplishes the same outcome as Copilot in Excel — without the per-user licensing dependency.

That takes more setup time. But it makes your AI investment portable. When Microsoft changes pricing again (and they will), you shrug instead of scramble.

The Bigger Signal for AI Budgets

Microsoft introduced E7 at $99/user/month last month. They’re running a promotional discount on Copilot Business because conversion at full price is weak. They pulled AI from utility apps after user revolt and are now paywalling it in productivity apps where the value is clearer.

All of these moves point in the same direction: Microsoft is done giving away AI and is building a tiered pricing structure that makes the free experience progressively worse until you upgrade.

That’s their right as a business. But it changes the calculation for anyone who treated Copilot as a free perk of their M365 subscription.

The organizations that treated Copilot access as a temporary trial period and built workflows that don’t depend on it — they’re fine. The ones who embedded Copilot into daily operations assuming it would stay free — they’re the ones scrambling this week.

I wrote about your AI tool stack getting smaller and more expensive in February. This is exactly the scenario. The free tier shrinks. The paid tier grows. The gap between them widens deliberately until the free option isn’t viable for production use.

The Decision Framework

Run this on Monday. You have one business day.

Pay for Copilot if:

• You have users who actively use Copilot in Word/Excel/PowerPoint daily
• Those workflows generate measurable business value (proposals, analysis, client deliverables)
• You’ve calculated the per-user cost against the output and it’s net positive
• You’re already on M365 and switching costs make alternatives impractical

Pivot away from Copilot if:

• Your team’s Copilot usage is light or experimental
• The per-user cost stacks uncomfortably on your existing M365 bill
• You can accomplish the same workflows with Claude, ChatGPT, or Gemini at lower cost
• You want your AI investment to be portable rather than locked to one vendor

Do nothing if:

• You’re under 2,000 seats and your team barely uses Copilot anyway
• You’re comfortable with “standard access” performance for light, non-critical usage
• You’d rather evaluate for 90 days under the degraded tier and then decide

There’s no universally right answer here. The wrong answer is not making a deliberate choice and letting the April 15 deadline make it for you.

Your Next Move

Open the M365 admin center today. Pull the Copilot usage report. Count how many people on your team used Copilot in Office apps more than five times in the last 30 days. That number is your licensing decision.

If the number is zero, congratulations — this change costs you nothing and you just learned something important about how your team actually uses AI. If the number is meaningful, price out selective licensing versus an alternative tool, make the call, and move on.

Microsoft trained your team on free AI, then pulled the rug. The only mistake is being surprised by it. Every vendor does this. The defense is building workflows that survive the inevitable pricing change.

Build portable. License selectively. And stop treating any vendor’s free tier as a permanent benefit.

---

Related Reading:

• Microsoft’s Copilot Rollback: Deploy AI Differently
• 90% Don’t Trust AI With Their Data. Copilot Shows Why.
• Microsoft 365 E7: What SMBs Need to Know
• Your AI Stack Has an Expiration Date