All tool profiles

Microsoft 365 Copilot: Licence Tiers and a Retail Sales Review

Microsoft 365 Copilot licence tiers, documented features in Excel, PowerPoint, Word and Teams, admin controls and an illustrative retail sales review workflow.

Independent editorial guide, written with AI assistance. Examples are hypothetical; no hands-on product testing or owner interviews were performed. Features and prices may change. Our editorial method.

ShiftToAGI | Shifting Founders Beyond AI

Evidence note: Features were checked against Microsoft's official overview on 2 November 2026 (Microsoft Learn: Copilot overview). On the same date, the official business pricing page could not be retrieved (Microsoft pricing page). This article therefore gives no price figures. ShiftToAGI is not affiliated with Microsoft.

Microsoft 365 Copilot is Microsoft's AI assistant for work. It is built on large language models, it can ground answers in web data and, depending on licence, in organisational data, and Microsoft says access is scoped by each user's permissions (source). For a business that already keeps its files, email and meetings in Microsoft 365, it puts AI inside apps staff already open every day, rather than adding a separate analytics product.

A hypothetical Monday problem

This story is fictional. It shows the kind of decision the tool might support, not a real customer.

Priya owns four homeware shops and an online store. Every Monday her operations lead, Tom, exports last week's sales into one Excel workbook, and on Tuesday morning they review it together. This week the online return rate looks high. Tom cannot tell whether the cause is one product line, one courier or a reporting error. The review is in 18 hours, and Priya has to decide whether to pause a reorder worth several thousand pounds. Acting on a pattern that turns out to be a spreadsheet mistake would cost her either stock or credibility with the supplier.

Copilot will not make that decision. At best it helps Tom question the workbook faster and draft the review pack. Priya still makes the call, and someone still checks the numbers.

What Microsoft 365 Copilot is, and which version you have

The name covers several licence levels that behave differently, so the first question is which one each person holds (source):

Microsoft says in-product labels show users which experience they have. For planning, this matters more than any single feature. A workflow that assumes Copilot can work inside Excel fails for someone on Copilot Chat (Basic) alone, and a workflow that assumes Copilot finds last month's files unprompted needs Premium.

Documented features worth understanding

Copilot in Excel

Microsoft lists data analysis, insights, formulas and visuals as Excel capabilities (source). Keep the layers separate. Pivot tables, filters, SUMIFS and charts are ordinary Excel features that predate Copilot. The AI layer lets you ask in plain language and get a suggested formula, chart or observation. Whether that suggestion matches what you meant is your job to confirm.

Drafting in PowerPoint and Word

Copilot can create presentations from prompts or existing content and can draft and summarise documents (source). For an operating review, the sensible use is a first-pass deck or memo built from figures already checked by a person.

Teams meeting summaries and action items

In Teams, Copilot can transcribe and summarise meetings and capture action items (source). That helps after a review, when each decision needs an owner. A person should still confirm the list, because a summary can attribute agreement to the wrong person.

Model choice

The overview describes a model selector with an automatic default, a faster option and a deeper reasoning option (source). The deeper option sounds suited to multi-step questions about returns or margin. We have not measured whether it performs better.

Admin and governance controls

These are control features, not AI generation. Admins can manage web search grounding and how users create and use agents. Microsoft also describes Purview for sensitivity labels and reviewing Copilot prompts and responses, oversharing tools for SharePoint, and usage reports (source). Some of these sit in the Premium description, so confirm availability for your licences. A small firm without IT staff should name one person who owns these settings.

Who it suits

It fits businesses whose records already live in Excel, SharePoint, Outlook and Teams, where managers spend hours each week turning exports into summaries. It also fits founders who want an operations lead to run reviews without first becoming a spreadsheet specialist, provided that lead has time to check results.

Potential applications by role

  1. Owner or founder: turn the weekly sales workbook into three questions worth discussing, rather than a wall of numbers.
  2. Operations lead: draft formulas that flag stock lines with rising returns, then test them against a few rows already known to be right.
  3. Store or area manager: summarise a Teams call with store staff into action items, and send them only after confirming each one.
  4. Finance or bookkeeper: draft variance commentary in Word from figures already reconciled. Copilot handles wording; the bookkeeper owns the numbers.
  5. Marketing lead: compare promotion weeks with normal weeks to frame a hypothesis for the next campaign, not a conclusion.

Illustrative workflow: weekly retail sales review

This is a proposed workflow built on documented features. It is not a product test, and the data is synthetic. It assumes the user has Copilot inside Excel and PowerPoint (Microsoft 365 Copilot Basic or Premium). File storage, format and size requirements are not covered in the source used here, so check Microsoft's requirements before relying on it.

Fictional input: one workbook with columns Date, Channel (Store A, B, C, D, Online), SKU, Category, Units Sold, Revenue, Units Returned, Return Reason and Courier. It holds 1,840 rows for one week. Twelve rows have a blank Return Reason, and Store C has no rows for Saturday.

  1. Prepare the data. Format the range as an Excel table with clear headers. Remove customer names, emails and addresses first. Product and channel data is enough here.
  2. Check completeness. Ask Copilot to list blank cells by column and any channel with missing dates. A useful answer would read: "Return Reason blank in 12 rows; no Store C rows for Saturday." If Copilot cannot produce this reliably, use filters instead. Tom then confirms whether Store C was closed or the export failed.
  3. Handle missing data explicitly. If Saturday is a failed export, stop and re-export. Never let a summary treat it as zero sales. Mark blank return reasons as "Unknown" rather than guessing.
  4. Ask focused questions. Use the prompt brief below. Sample output, invented for illustration: "In this week's online channel, Cushion returns warrant review; 61% of Cushion returns list 'damaged', and most of those used Courier X."
  5. Verify by hand. Tom builds a pivot table, an ordinary Excel feature, to recompute the Cushion return rate, the reason split and the courier split. Where figures disagree, inspect source rows, filters and formulas independently. Neither result is accepted until it reconciles to the source totals; record the discrepancy in the review notes.
  6. Draft the review pack. Ask Copilot in PowerPoint for a short deck from the verified figures only, including a slide headed "What we do not know yet".
  7. Human approval. Priya decides whether to pause the reorder or contact the courier. Nothing in this workflow sends a supplier email or changes an order. A named person approves and sends any message.
  8. Record the outcome. After the meeting, check the Teams action items against what was agreed before sharing them.

Pilot checks over four weekly reviews:

Stop route: pause if an unverified Copilot figure reaches a decision, if customer personal data appears in a prompt, or if Copilot's figures fail the pivot check in two or more of the four weeks. Return to the manual review, then look at table structure and prompt wording before trying again.

Copyable prompt brief

You are helping prepare a weekly retail operating review.
Data: the Excel table "WeeklySales" (product and channel level, no customer data).
1. List blank cells by column and any channel with missing dates. Do not fill gaps.
2. Calculate return rate (Units Returned / Units Sold) by Channel and Category.
3. For the category with the largest return rate, break returns down by Return Reason and Courier.
4. Show the formulas you used so I can check them.
5. State what this data cannot tell us. Do not recommend purchasing or supplier actions.

The brief asks for the highest rate rather than a week-on-week rise, because the fictional workbook holds only one week.

Costs, access and limits

Pricing status: numeric price not verified. The business pricing page could not be retrieved on 2 November 2026, so check the official pricing page. Do not assume any tier is free for you. Copilot Chat (Basic) is described as included with eligible licences, and eligibility depends on what you already hold.

Documented access and cost dependencies (source):

General limits apply too. AI answers can be wrong, messy headers and merged cells make analysis harder, and loose file permissions can surface content people should not see.

Who should skip it

Frequently asked questions

Can Copilot replace our bookkeeper or analyst?

No. Treat it as a drafting and questioning aid. Reconciled figures, tax treatment and judgement calls stay with qualified people.

Should we put customer data into prompts?

Avoid it unless there is a clear need and you have checked your data policies. Most sales reviews work at product and channel level, where personal data adds risk without adding insight.

How do we know if the pilot worked?

Compare four weeks of assisted reviews against your manual checks and your own time log. If figures held up and nothing unverified reached a decision, you have grounds to continue. Otherwise, stop and revert.

Practical next action

This week, ask whoever manages your Microsoft 365 account which Copilot experience each team member has. Then make a copy of last week's sales export with customer details removed, run the prompt brief on it, and check every figure against a pivot table before the next review.


Methodology: AI-assisted draft based on Microsoft's official documentation, checked 2 November 2026. The workflow, people and data are illustrative. This is not a hands-on product test, and ShiftToAGI is not an official Microsoft publisher.

Keep exploring

Compare the tool profiles

Find a workflow for your business

Explore ShiftToAGI courses in development