Microsoft Copilot for small business

What Microsoft is selling small businesses

Microsoft offers two layers of AI to businesses on Microsoft 365. The first is Microsoft 365 Copilot Chat, which Microsoft describes as a web-grounded AI assistant included at no additional cost with eligible Microsoft 365 plans. Staff sign in with their work account and get a chat assistant with enterprise data protection, plus Copilot in Outlook.

The second is the paid license, Microsoft 365 Copilot Business, aimed at organizations with up to 300 users. It adds Copilot inside Word, Excel, PowerPoint, Outlook and Teams, grounded in the user's own work data: email, files, chats and meetings. Microsoft's pricing page also lists pre-built agents for research and analysis and the ability to build internal agents with Copilot Studio. Features change often, so check Microsoft's current plan comparison before you buy.

Current prices (check before buying)

As of November 2026, Microsoft's US business pricing page lists the following. These are list prices for annual subscriptions unless noted, and promotions change.

Microsoft also notes usage-based billing for select advanced AI features, so total cost can exceed the license fee if staff use those features heavily. Ask your Microsoft partner or IT provider to confirm the price that applies to your tenant.

• Microsoft 365 Copilot Business (add-on to an existing Business plan): from $18.00 per user per month paid yearly, shown as reduced from $21.00. Microsoft notes promotional pricing applies to the first year only. Monthly billing is listed at $25.20.
• Microsoft 365 Business Standard with Copilot: $23.50 per user per month, paid yearly.
• Microsoft 365 Business Premium with Copilot: $32.00 per user per month, paid yearly, or $38.40 on a monthly subscription.
• Microsoft 365 Copilot Chat: included at no additional cost with eligible Microsoft 365 plans.

Is it worth it for a small firm?

Copilot is most useful for people who spend their day in email, documents and meetings. In a 30-person accounting firm, partners and managers who write client letters, summarize long email threads and prepare meeting notes are likely candidates. Staff who mostly work in a line-of-business system, such as a technician in ServiceTitan or a hygienist in Dentrix, may get little from a paid license.

A rough way to think about value, as an illustration only: at $18 to $25 per user per month, a license needs to save a person well under an hour a month to cover its fee at most professional wage levels. The harder question is whether the time saved turns into more billable work, faster client responses or fewer errors. That is what a pilot should measure.

Think about which roles spend the most time on first drafts and summaries. In a law firm, attorneys preparing client update letters or associates reviewing long email chains are natural candidates. In an engineering firm, project managers writing meeting minutes and proposal sections may benefit most. Front-line roles that live in one specialist system may be better served by AI features inside that system, if it offers them.

Fix permissions before you switch it on

Copilot works across whatever each user is already able to open. That is the point, and it is also the main risk. In many small firms, SharePoint sites and Teams channels have been shared broadly for years. A folder of payroll reports or a partner's draft compensation plan may technically be open to everyone, simply because nobody ever locked it down. Before Copilot, nobody stumbled on it. With Copilot, a casual question might surface it.

Before rollout, ask your IT provider or technology lead to review sharing settings on SharePoint and OneDrive, remove organization-wide links on sensitive folders, check who has access to HR, finance and client-confidential sites, and confirm multi-factor authentication is on for every account. Remove licenses and access for departed staff. This work is worth doing even if you never buy Copilot.

Set rules for use

Write a short AI usage policy that covers Copilot and any other AI tools staff use. Say which tools are approved, what kinds of data may be used with them, and that a person must review anything sent to a client. Copilot drafts can be confident and wrong, especially with numbers and dates.

Regulated firms should involve counsel. A law firm should consider confidentiality duties and ABA Formal Opinion 512. A health practice should confirm how HIPAA obligations apply to its Microsoft agreements and configuration. A tax or accounting firm should check the work against its written information security plan under the FTC Safeguards Rule. These are questions for your advisers, not reasons to avoid AI altogether.

Run a small pilot

Start with five to ten people who handle a lot of email, documents or meetings, and who are willing to try new things. Before licensing them, write down three tasks to measure, such as time to draft a client follow-up email after a meeting, time to summarize a long thread, or time to prepare a first draft of a proposal from notes.

Measure those tasks for two weeks without Copilot and four to six weeks with it, using a simple log. Hold a short weekly session where pilot users share prompts that worked. At the end, compare times, ask about quality, and check whether drafts needed heavy correction. License more people only where the numbers and feedback support it. Some firms find a few heavy users justify licenses while everyone else does fine with Copilot Chat.

Pay attention to quality, not just speed. Ask reviewers to note how often a Copilot draft needed significant correction, and whether any draft contained a factual error that could have reached a client. A tool that saves ten minutes but introduces errors is not saving time once someone has to catch them.

Alternatives worth comparing

Copilot is not the only option. Firms on Google Workspace have Gemini features available in their plans, and standalone business plans for ChatGPT and Claude are widely used. The right choice depends on where your documents and email live, which controls you need, and what staff already use. A firm running entirely on Microsoft 365 often benefits from Copilot's access to its own files, while a firm on Google Workspace would rarely switch suites just for AI.

Whatever you choose, the readiness steps are the same: clean permissions, a clear policy, a measured pilot, and a person reviewing client-facing output. InTheory can help a firm run that pilot and decide who needs licenses, but a capable internal owner and a good IT provider can do it too.