How to do a competitor SWOT analysis with ChatGPT

Why a bare SWOT prompt produces generic bullets

Ask ChatGPT to "do a SWOT analysis of Competitor X" and you'll get a tidy four-box answer within seconds. Read it closely and most of the bullets could describe any company in the category: "strong brand recognition", "dependence on a single market", "growing demand for AI features", "intense competition". The model isn't being lazy. Without evidence in front of it, it completes a pattern it has seen thousands of times, and that pattern is the generic SWOT.

Two things fix this. First, collect evidence before you ask for judgment. When the model has to write down "Acme's pricing page lists a free tier up to one million events, with the URL" before it's allowed to call anything a strength, the strength has to rest on that sentence. Second, make the SWOT relative to you. A competitor's strength is only a strength if it matters against your product, so the analysis needs one line about what you sell, to whom, and at what price. The prompt at the top of this page does both. Its evidence step uses the same eight dimensions a Competely report has, so the structure you learn here carries over.

One disclosure: Competely wrote this guide, and Competely is a competitor analysis tool. It appears twice near the end, in the section on keeping the SWOT current and in a section about where it fits and where it doesn't. Everything before that works with ChatGPT alone. If you want a plain SWOT grid to fill in by hand, the free SWOT analysis template is the companion to this guide, and the broader guide to competitor analysis with ChatGPT covers a full eight-dimension analysis rather than the SWOT alone.

What ChatGPT gets right, and where it hits a ceiling

ChatGPT is good at the parts of a SWOT that reward breadth and speed. With web search on, it can read a pricing page, a features page, a careers page, and a handful of review pages in the time it takes you to open the first one. OpenAI's help center states that responses using web search include citations you can select to open the source, plus a Sources button that lists what the answer drew on. It's also good at the synthesis step. Given fifteen sourced facts, it will group them into strengths and weaknesses sensibly and notice patterns you might miss, like three reviews complaining about the same thing.

It hits a ceiling in four places.

• Unsourced claims still slip in. ChatGPT may search on its own when a question would benefit from current information, and OpenAI's own documentation warns that search results and citations can be incomplete, outdated, or incorrect. A fact with no URL next to it is a guess until you've checked it.
• Answers drift between runs. Ask about the same competitor twice and you'll get two differently shaped answers. That's fine for one competitor and a problem for six.
• Pricing goes stale fastest. The model reads the page at the moment you ask. A price change a week later doesn't reach you.
• Nothing watches for changes by default. ChatGPT does have scheduled tasks, covered below, but the ordinary chat doesn't tell you when a competitor moves.

The method, step by step

1. Define your own product in one line

Write one sentence: what you sell, who buys it, and roughly where you sit on price. "We sell product analytics to B2B SaaS teams under 50 people, flat monthly pricing" is enough. Without this line, ChatGPT has no reference point, and "strength" and "weakness" collapse back into general praise and criticism. With it, a competitor's free tier becomes a threat if you charge from day one, and a non-event if you have one too.

2. Gather evidence per dimension with web search on

Start a new chat and turn on web search. In ChatGPT you select the + icon and choose Web search, or type / and pick it from the menu. Paste the prompt with the competitor's name and URL. Step 1 of the prompt makes ChatGPT collect one to three sourced facts under each of the first seven dimensions before it writes a single judgment. Read this section before you read the SWOT, and open the source behind anything surprising. A line that says "no public source found" is the prompt working as intended.

If you have your own material, such as a screenshot of a competitor's pricing page from a sales call or a PDF of their one-pager, attach it. ChatGPT reads uploaded PDFs and images. Two cautions from OpenAI's help center: on most plans a PDF is read as text only and images inside it are discarded, and small text in a screenshot is harder for the model to read, so paste pricing as text when you can. The prompt tells ChatGPT to prefer the live web over attached files where the two disagree.

For a deeper pass, OpenAI's deep research mode reads many more sources and returns a longer report with citations. It starts from the same tools menu. Which plans include it and how many runs you get has changed more than once, so check OpenAI's help center rather than assuming.

3. Build the SWOT from that evidence only

Step 2 of the prompt forces every item into the form "item. Rests on: dimension, fact." Read each item and ask two questions. Is the fact really in the evidence section? Does the fact actually support the label? "Acme has a mobile app" in the evidence section doesn't make "strong mobile presence" a strength unless the app matters to the buyers you both sell to. Delete any item that fails either test.

Keep the boxes honest. Strengths and weaknesses describe the competitor itself: their product, their pricing, their team, their reputation. Opportunities and threats describe forces outside them: a market shift, a regulation, a new entrant, a change in how buyers behave. Any item the competitor could change by shipping a feature belongs in the top half.

Then read the "Implication for us" lines. The prompt asks for a concrete action rather than a direction, so push back on "Implication for us: consider our pricing". What you want reads like "Implication for us: add a flat-price row to the comparison page and test it against Acme's usage-based plan".

4. Run every competitor through the identical structure

Open a new chat for each competitor and paste the identical prompt, changing only the competitor line. Don't ask follow-up questions that reshape one competitor's output, because a follow-up on competitor two won't be applied to competitor one. If you want an extra dimension or a different fact count, change the prompt and re-run everyone. This is the only defense against drift between runs, and it's what makes the next step possible.

5. Synthesize across competitors into one landscape view

Once you have three or more SWOTs in the same shape, paste them into one new chat and ask for a landscape view: which strengths appear at more than one competitor (those are table stakes, not differentiators), which weaknesses are shared (that's your opening), and which threats hit everyone including you. Ask for the same "Rests on" trace on every landscape claim, pointing back to a named competitor's SWOT.

A worked example on a fictional competitor

This example uses a made-up company, Acme Analytics, and made-up URLs, so no real company's data is involved. Say your one-line definition is "product analytics for B2B SaaS teams under 50 people, flat monthly pricing".

Strength (Product & Features). Acme ships session replay and funnel analysis in every plan, including the free one. Rests on: Product & Features, the features page at acmeanalytics.example/features lists both under "Included in all plans". Implication for us: stop leading with "replay included" in our comparison copy, since it no longer separates us, and lead with flat pricing instead.

Weakness (Customer Sentiment). Recent reviewers describe surprise overage charges when event volume spikes. Rests on: Customer Sentiment, four of the ten most recent reviews on the G2 listing at g2.example/acme-analytics mention unexpected bills, and Pricing & Sales, the pricing page at acmeanalytics.example/pricing bills per million events above the plan limit. Implication for us: publish a "your bill never changes" section on our pricing page and quote it in outbound email to teams evaluating Acme.

Threat (Market & Opportunity). A large suite vendor announced bundled analytics at no extra charge for its existing customers. Rests on: Market & Opportunity, the announcement dated 2026-08-12 at suitevendor.example/blog/analytics. Implication for us: this squeezes Acme and us equally at the low end, so decide this quarter whether we compete on price or move to larger teams the suite doesn't serve well.

Each item has a label with its dimension, a specific fact with a URL, and an action with a verb. A generic SWOT has none of those.

A checklist before you trust the SWOT

Run these checks before the SWOT reaches a planning meeting.

1. Every SWOT item traces to a sourced fact in the evidence section, and you've opened the source behind every item you plan to act on.
2. Strengths and weaknesses are about the competitor itself. If an item describes the market, move it down.
3. Opportunities and threats are about forces outside the competitor. If the competitor could fix it by shipping something, move it up.
4. No item is a restated feature. A feature becomes a strength only when the item says why it matters against your product.
5. Pricing facts match the competitor's live pricing page today, plan by plan, including billing period and usage limits.
6. Every "no public source found" line is either resolved by a fact you found yourself or accepted as unknown. None was quietly filled in.
7. Sentiment items rest on more than one review, and the reviews are recent enough to describe the current product.
8. Every "Implication for us" line names an action someone could start this week, and you've edited it to fit what your team can do.
9. The SWOT reflects your one-line definition. If changing that line would change some items, the analysis is truly relative to you.
10. The research date and the stale-first list are on the document.

Keeping the SWOT current

A SWOT is a snapshot, and the pricing fact behind your weakness item can change without telling you. You have three honest options.

Do it by hand. Save the exact prompt and your one-line definition, set a monthly reminder, re-run each competitor, and compare the new evidence section against the old one. It costs nothing and works as long as you keep doing it.

Build it in ChatGPT. ChatGPT has scheduled tasks, which OpenAI's help center describes as one-time or recurring tasks that can also monitor for changes and notify you when something relevant happens. Free accounts get up to three active tasks running no more than once a day, and paid plans can run tasks more often. OpenAI's release notes say monitoring tasks can search the web. You could set a weekly task that re-checks a competitor's pricing and features pages and reports differences. What you take on is the setup and the upkeep: choosing the pages, defining what counts as a change worth reporting, and keeping the output in the same shape every week. Check the help center for which plans include monitoring tasks before you build on them.

Buy it. A monitoring product re-reads the competitor's pages on a schedule and sends you what changed. You trade flexibility for not maintaining anything. That is where our own product comes in.

Where Competely fits, and where it does not

Competely automates the method above. You enter your product and your competitors, and it analyzes each one across the same eight dimensions the prompt uses, more than 100 data points with a source link on each, in about 30 minutes. Its eighth dimension is the SWOT Analysis, built from the facts in the other seven, so every item rests on something you can open. Because the structure never varies, competitors line up side by side without the discipline step 4 asks of you. Monitoring is included on every paid plan. On a schedule you set, every two weeks or monthly on any plan and weekly on Pro and Scale, it re-reads each competitor's public pages, flags what changed, and emails a short brief. You can read a full sample analysis of Figma vs Adobe XD without an account.

Its limits matter too. Competely's SWOT describes the competitor as seen from its public web pages. It doesn't know your internal strategy, your roadmap, or what your sales team hears on calls, so the "implications for us" step stays with you. It isn't a strategy consultant and won't tell you whether to move upmarket. It watches public web pages only, so there is no social listening, no press tracking, and no win-loss interviews. If ChatGPT plus a monthly reminder covers you, you don't need it. Self-serve, from $39/mo, billed monthly or yearly, cancel anytime, no demo call.