Buying an AI Tool

ChatGPT vs Copilot: the comparison depends on the job, not the tool

The prices, side by side

Both vendors changed prices in 2026, so most comparisons online are out of date. These are list prices as of September 14, 2026, in US dollars, before tax.

The per-seat price is a tie. That is the first thing to notice, because it means price will not decide this. Two things will: what you already pay for, and where the job happens.

What you already pay for changes the answer

If your firm is on Microsoft 365 Business Standard or Premium, you already have Copilot Chat at no extra cost. It is the version that does not see your files, but it does the general drafting and summarising that a lot of small firms buy a tool for. The paid Copilot Business add-on is what connects it to your documents, email and meetings.

If you are on Google Workspace, the same logic applies with Gemini, which is included in Business Standard and above. Copilot is not a realistic option for a Google firm; the comparison there is ChatGPT versus Gemini.

This is why the method starts with a test rather than a comparison. Before you price either tool, run the actual job through the assistant you already have, three times, with real inputs. Then write down, in one sentence, the thing it could not do. If you cannot write the sentence, you do not need either tool yet. If you can, that sentence is what you are comparing ChatGPT and Copilot on. Not their feature lists.

Where the output lands

The real difference between the two is structural, and it decides adoption more than any benchmark.

Copilot's output lands inside the thing you were already working in. The draft appears in the Word document. The summary appears in the Outlook thread. The meeting notes appear in Teams. There is no copying. For a job that lives in those apps, that is the whole case for Copilot, and it is a strong one: people keep using tools that are already open.

ChatGPT's output lands in ChatGPT. It is often the better draft, on a wider range of tasks, with more room to think out loud, and then someone copies it into the document, the email or the CRM. For a job that is mostly thinking, analysis, or writing across many kinds of files, that trade is worth it. For a job that is the same document every week, every copy is a step, and steps are where a tool goes quiet by week three.

The question to ask is not which tool is better. It is: where does the job happen now, and does the output need to arrive there?

Data, and what each vendor does with it

Both business tiers state that they do not train on your inputs by default. That is the headline, and it is the same for both. The details differ and they are the details a firm handling client files should read.

Copilot Business inherits the permissions and admin controls of your Microsoft 365 tenant. If your firm already manages who can see which files, Copilot respects that. It also means Copilot can surface any file a user already has access to, which is a reason some firms tighten permissions before switching it on.

ChatGPT Business has its own admin console, its own retention settings and its own data processing terms. Single sign-on and some audit features are on the Enterprise tier, not Business. For a firm without an IT person, that means one more set of terms to read and one more admin console to own.

Neither is unsafe. Both require someone to read the data processing terms rather than the landing page, and to check where data is stored if a client or regulator cares.

How to actually decide between them

Run both through the same four questions, from a trial, with the person who will do the job.

Score each one to five, weight the job highest, and total. If the two finish within a few points of each other, the decision is not about the tool. It is about who will own it, where the draft needs to land, and whether the free one already did the job. That is usually where this comparison ends.

What this is not

This is not a verdict. Both tools do the general job well and both change monthly. Any post that tells you one is better is telling you which one was better on the author's job, run by the author, on the author's data. The free five-article method at practicalmethods.ca/guides/buying-an-ai-tool gives you the questions to run on yours.

The list is not the decision. It is how you find the one or two things that are.

Not a review, not rankings, not software advice. A method.