Instagram Competitor Analysis with Claude or ChatGPT

Can Claude or ChatGPT do an Instagram competitor analysis?#

Yes, once they can read Instagram. On their own, neither can open a profile, so they analyze whatever you paste. Connect StoriesFly and you can name three to five competitors in one message; the assistant reads their public profiles, posts, reels, comments and hashtags, and returns a side-by-side comparison with numbers. A five-competitor analysis costs about 50 requests and takes a few minutes.

Updated October 2026.

This article is the AI version of the job. If you want the method itself (what to look at and why), read the Instagram competitor analysis guide first. Here we cover setup, a five-step workflow with prompts you can copy, request counts, and the limits.

Setup in one minute#

Add StoriesFly to your assistant once. Full steps are on the connect page.

• Claude: Customize → Connectors → Add custom connector → https://storiesfly.com/mcp → Connect → sign in with Telegram.
• ChatGPT: Plugins → Add → Add custom MCP server → same address → OAuth → Connect → sign in with Telegram. Works on ChatGPT's Free plan.

The first 100 requests are free, once per Telegram account, which covers a two-competitor analysis with room to spare. You never log in to Instagram with your own account, and your account never appears among a story's viewers.

Claude tends to be better at long, structured comparisons; ChatGPT is quicker for short checks. Both call the same tools, so the data is identical.

Step 1: Pick the competitors and set the frame#

The assistant will read more than you need if the question is open. Tell it the accounts, the time window and the output you want.

I run @ourbrand (specialty coffee, Lisbon). My competitors on Instagram are
@harbor.roasters, @northtrail.coffee and @fabrica.cafe.
For each, read the profile and the last 27 posts (3 pages).
Don't read anything else yet.

Cost: 1 profile + 3 pages of posts per account = 4 requests × 4 accounts (yours included) = 16 requests.

Step 2: The comparison table#

Now ask for the numbers. The data is already in the chat, so this step costs nothing.

Make a table: followers, posts per week, % reels / carousels / photos,
median likes, median comments, median engagement rate by followers,
and the date of the newest post. Then three lines on what stands out.

A typical answer (sample numbers):

"@northtrail.coffee has a third of @fabrica's followers but four times the engagement rate, mostly from reels shot behind the bar. @fabrica posts rarely and its photos get little response."

Engagement rate here is likes plus comments divided by followers, using the median so one viral post does not distort it. You can check single numbers with our engagement rate calculator guide.

Step 3: What their best posts have in common#

For each competitor, take the 3 posts with the highest engagement rate.
What do they share: format, length, topic, first line of the caption,
time of posting? Then tell me which of those patterns we never use.

Cost: 0 if the assistant uses what it already read. If it wants full details for a post (shares, saves where Instagram gives them, the video length), each post is 1 request. Ask it to say before reading more.

Step 4: What their audience says#

Comments tell you what people want from a brand in your niche: questions they ask, complaints, what they praise.

Read one page of comments on the top post of each competitor.
Group the comments by theme (questions, praise, complaints, tags of friends).
What are customers asking for that nobody answers?

Cost: 1 request per post = 3 requests.

For reels-heavy competitors you can also ask what they repeat:

Read the last 18 reels of @northtrail.coffee. List the hooks
(first line of caption and what happens in the first seconds, if the caption says).
Which hooks got above-median views?

Cost: 2 requests.

Step 5: Hashtags and discovery#

Which hashtags do the competitors use most in the posts you read?
For the top 3, read how many posts use each and one page of top posts.
Is any of them a tag where small accounts still reach the top?

Cost: about 2 requests per hashtag = 6 requests.

You can also ask who they work with:

Read the first page of accounts @harbor.roasters follows.
Which of them look like suppliers, cafés or creators they collaborate with?

Cost: 1 request per page of up to 25 accounts.

Total cost#

A deeper version with five competitors and reels for each is closer to 50. On the free 100 you can run two or three such analyses. For monthly checks, Basic gives 100 requests a month for $4.99 and Pro 500 for $19.99. An agency checking many clients' markets will spend less on a prepaid balance: $1.50 per 1,000 requests at the entry price, no monthly cap, and the balance never expires. See plans.

Make it a monthly routine#

Save the prompts from steps 1–5 as one message and run it on the first of each month. Ask the assistant to compare with last month's table if you paste it in. What changes month to month is more useful than any single snapshot: a competitor switching from photos to reels, a sudden jump in posting, a new type of collaboration.

In Claude, a Project with the prompt and last month's results as files keeps this in one place.

What the analysis can't tell you#

Public data shows what competitors post and how people react in public. It does not show:

• Reach, impressions and profile visits. Only the account owner sees these.
• Saves on every post and audience demographics. Insights only.
• Ad spend and paid reach. Look at Meta's Ad Library for which ads are running; spend is not public.
• Private accounts. The assistant gets the profile header and says the account is private.
• Why a post worked. The assistant finds patterns in the numbers; whether the cause was the hook, the timing or a share from a big account is a guess. Test before you copy.

Tips for better answers#

• Name the window. "Last 27 posts" is clearer than "recent posts" and caps the cost.
• Ask for medians. Averages get pulled by one viral post.
• Ask it to show the numbers it used. Then you can spot a misread.
• Compare like with like. A café with 8K followers and a chain with 400K have different normal engagement.
• Ask what you don't do. The useful output is the gap list, not the leaderboard.

The same connection reads TikTok. For that side see TikTok competitor analysis with AI. For single-profile analysis, see can ChatGPT analyze an Instagram profile.