How to Use ChatGPT for Competitor Analysis (Step by Step)
🔍 Want the best deal? Check current prices and availability.
Compare Prices →When you buy through links on our site, we may earn a commission.
How to Use ChatGPT for Competitor Analysis (Step by Step)
Most competitive research advice assumes you have a budget for Crayon or Klue and a dedicated analyst to run them. Most small businesses have neither. They have a founder or a marketing manager with eleven other things on their plate and a vague sense that a competitor just changed their pricing page.
That's the gap ChatGPT fills. Not perfectly, and not without some real caveats I'll get into — but well enough that a solo operator can produce a competitor teardown in an afternoon that would have taken a consultant a week.
I've run this workflow for a handful of SaaS and ecommerce clients over the past year. Here's exactly how it works, including the prompts, the parts where ChatGPT will confidently lie to you, and what to do about it.
What ChatGPT Is Actually Good At (And What It Isn't)
Let's get the limitations out of the way first, because they shape the entire workflow.
ChatGPT is genuinely good at:
- Synthesizing messy inputs — pasted web pages, PDFs, CSV exports — into structured analysis
- Turning raw observations into frameworks: SWOT, positioning maps, battlecards
- Drafting the actual deliverable your sales team will read
- Playing devil's advocate on your own positioning ("here's why a prospect would pick them over you")
- Running the same analysis across 8 competitors without getting bored
ChatGPT is bad at:
- Knowing current facts without browsing enabled. Its training data has a cutoff, and pricing pages change constantly
- Giving you real traffic numbers, ad spend, or revenue figures. It will estimate if you push it, and those estimates are fiction
- Accessing gated content — G2 reviews behind a login, paywalled analyst reports, your competitor's internal Slack
- Anything requiring a live, continuously updated feed of competitor moves
The practical takeaway: ChatGPT is the analyst, not the data source. You feed it facts; it does the thinking. If you skip the feeding step, you get plausible-sounding garbage.
That's why I pair it with a live-research tool. Perplexity for real-time web lookups, or a data platform like Semrush if I need hard numbers. Check Tool -> for a current comparison of those options.
Step 1: Define a Competitive Set You Can Actually Analyze
Before you open ChatGPT, write down your competitors in three buckets:
- Direct — same product, same buyer, same price band (3–5 companies)
- Adjacent — different approach, overlapping buyer (2–3 companies)
- Status quo — "we just do it in a spreadsheet" or "we hire an intern" (1–2 entries)
That third bucket matters more than people think. In most market analysis work I've done, the biggest competitor is inertia, and ChatGPT is useful for articulating why someone would switch from doing nothing.
Cap the list at 8. Beyond that, the output gets shallow and you stop reading it.
Step 2: Set Up a Project With Persistent Context
If you're on ChatGPT Plus or above, use Projects. It keeps all your competitor research in one place with shared instructions, so you don't re-explain your business every session.
Create a new project and paste in custom instructions like this:
You are a competitive intelligence analyst supporting [COMPANY NAME].
About us:
- We sell [PRODUCT] to [TARGET BUYER]
- Our price is [PRICE]
- Our main differentiator is [DIFFERENTIATOR]
- Our weakness is [HONEST WEAKNESS]
Rules for every response:
- Never state a fact about a competitor unless I have provided it in this
conversation or you have cited a source you browsed.
- If you don't know something, say "UNKNOWN — needs verification."
- Flag any claim you're inferring rather than sourcing.
- Write in plain English. No marketing adjectives.
Rule #2 is the important one. Without it, ChatGPT fills gaps with confident invention, and if you paste that into a board deck you'll regret it.
If you're on the free tier, you don't get Projects, but you can paste the same block at the top of each new chat. Annoying, but it works.
Step 3: Feed It Real Data
This is the step people skip, and it's the one that determines whether the output is worth anything.
Gather raw material first:
- Pricing pages — copy the full text of each competitor's pricing page, including FAQ sections and fine print
- Homepage and product pages — the headline, subhead, and feature bullets
- Recent changelog or release notes — signals where they're investing
- Job postings — a company hiring three enterprise AEs is moving upmarket; that's a strategic tell
- G2 or Capterra reviews — copy the 3-star reviews specifically. Five-star reviews are marketing; three-star reviews are product feedback
- Funding announcements and press releases — context for why they're behaving a certain way
Drop these into a single document per competitor, then upload the files to ChatGPT or paste them in. If you have a CSV of review data, the Data Analysis feature will parse it directly and give you coded themes instead of vibes.
One warning: don't paste anything you wouldn't want in a cloud system. Competitor info is fine. Your own unreleased roadmap or customer contracts are not.
Step 4: Run the Core Analysis Prompts
Here are the four prompts I actually use, in order.
Prompt 1 — Positioning teardown
Below are the homepage and pricing page copy for [COMPETITOR].
Extract:
- Who they say they're for (quote the exact language)
- The primary pain they claim to solve
- The outcome they promise, and any specific numbers they cite
- What they conspicuously do NOT mention
- The 3 words they repeat most often
Then tell me: what buyer are they optimizing for, and what buyer are
they implicitly deprioritizing?
Point 4 is where you find openings. If nobody in your category mentions implementation time, and your product is fast to deploy, that's your wedge.
Prompt 2 — Pricing and packaging analysis
Here is the pricing page for [COMPETITOR A], [COMPETITOR B], and
[COMPETITOR C].
Build a table with: entry price, price at 10 users, what's gated
behind the next tier, overage charges, contract requirements, and
free tier limits.
Then answer: at what company size does each competitor become the
cheapest option? Where is the pricing gap in this market?
This one consistently produces something useful. ChatGPT is very good at normalizing messy pricing structures into a comparable grid — per-seat vs. usage-based vs. flat-rate is exactly the kind of thing it handles cleanly.
Prompt 3 — Objection simulation
You are a [TARGET BUYER] evaluating us against [COMPETITOR].
Using only the materials I've provided, write the five strongest
reasons you would choose [COMPETITOR] over us. Be specific and
unflattering. Do not soften anything.
This is the prompt that earns its keep. It surfaces objections your sales team hears but doesn't report upward.
Prompt 4 — Synthesis into a deliverable
Using everything in this project, write a one-page competitive
battlecard for [COMPETITOR] with these sections:
- Who they are (2 sentences)
- Where they win (3 bullets, each with a source)
- Where they lose (3 bullets, each with a source)
- Our counter-positioning (3 bullets)
- Trap-setting questions our reps can ask prospects
Mark any claim lacking a source as [VERIFY].
That last instruction is doing real work. It forces the model to distinguish what it knows from what it's guessing, and it gives you a to-do list instead of a document full of holes you didn't notice.
Step 5: Turn Findings Into a Monitoring Rhythm
A one-time analysis goes stale in about six weeks. Competitors ship, reprice, and reposition constantly.
Set up a lightweight loop:
- Weekly (15 min): Ask ChatGPT to diff this week's competitor changelog, blog RSS, or pricing page against last week's saved copy. Paste both versions and ask what changed and what it implies.
- Monthly (1 hour): Refresh the pricing table. Repricing is the single most actionable signal in most markets.
- Quarterly (half day): Rerun prompts 1 through 4 with fresh inputs and update the battlecards.
For the weekly diff, a dedicated page-monitoring tool beats manual copy-paste. Visualping or Google Alerts handle the "tell me when it changes" part for free or close to it, and then ChatGPT handles the "what does this change mean" part.
ChatGPT Plans and Alternatives Compared
Pricing below reflects what I found at the time of writing. Verify before you buy — this category reprices often.
| Tool | Best For | Entry Price | Live Web Data | Competitive Research Fit |
|---|---|---|---|---|
| ChatGPT Free | Trying the workflow | $0 | Limited | Weak — no Projects, capped browsing |
| ChatGPT Go | Casual use | ~$8/mo | Limited | Basic synthesis only |
| ChatGPT Plus | Solo founders, small teams | $20/mo | Yes | Best value starting point |
| ChatGPT Pro | Heavy daily research | $200/mo | Yes | Overkill unless you live in it |
| ChatGPT Team | 2+ person marketing teams | $25–30/user/mo | Yes | Shared Projects, good for agencies |
| Perplexity Pro | Live web research | $20/mo | Strong, cited | Best companion to ChatGPT |
| Claude Pro | Long-document analysis | $20/mo | Limited | Strong on big PDFs and reports |
| Semrush | Hard traffic and keyword data | ~$140/mo | N/A (own dataset) | Needed for real numbers |
| Crayon / Klue | Dedicated CI teams | Custom (typically $15k+/yr) | Yes | Continuous monitoring at scale |
Check Tool -> if you want to see how these stack up on current pricing.
Pros and Cons
Pros
- Costs $20/month versus five figures a year for a dedicated CI platform
- Handles messy, unstructured inputs better than any spreadsheet workflow I've tried
- The objection-simulation prompt surfaces things internal teams are too polite to say
- Produces a finished deliverable, not just raw notes — the battlecard prompt alone saves hours
- Scales across 8 competitors without a proportional increase in effort
Cons
- Will invent specifics if you don't explicitly forbid it. The guardrail instruction is mandatory, not optional
- No reliable numbers on traffic, ad spend, or revenue. You need a separate data source for that
- Browsing is inconsistent — sometimes it cites sources, sometimes it works from memory without saying so
- Projects and the best features sit behind the $20 tier; free users get a noticeably worse experience
- Output quality tracks input quality almost linearly. Garbage in, confident garbage out
- No continuous monitoring. It's a point-in-time tool you have to re-run
Common Mistakes
Skipping the source material. Asking "analyze my competitors" with no inputs produces a generic essay. Feeding it four pricing pages produces a pricing teardown.
Trusting numbers. If ChatGPT tells you a competitor has 40,000 monthly visitors, that's a guess dressed up as a statistic. Get traffic data from Semrush or Similarweb, then bring it back into ChatGPT for interpretation.
One-shot analysis. The value compounds when you revisit it. A single teardown is a snapshot; a quarterly rhythm is actual intelligence.
Analyzing too many competitors. Eight is the ceiling. Five is better.
Publishing raw output. Everything ChatGPT produces needs a human pass
🔍 Want the best deal? Check current prices and availability.
Compare Prices →