How to spy on your competitors with Claude

Search the web and read every page you can reach on [competitor URL]. I run [describe your business in one line, including who you sell to], so read them as a direct competitor of mine, not as a neutral observer.

Work through their site properly: the homepage, the pricing or packages page, the about page, at least three product or service pages, and their blog or resources index if they have one. Open the pages, do not infer them from search snippets.

Then give me a teardown in these six sections. 1. Positioning: the exact promise they lead with, in their words, plus the customer they are clearly writing for and the customer they are clearly not. 2. Offers: every product, package or tier they sell, with what is included in each. 3. Pricing: the real numbers if published, and if not, every clue on the page about how they charge, such as "starting from", "custom quote", or a minimum spend. 4. Proof: the case studies, logos, review counts, guarantees and named results they use to make you believe them. 5. Funnel: what they ask a first-time visitor to do, every free thing they offer to capture an email, and how many clicks it takes to talk to a human. 6. Language: ten words or phrases they repeat across pages, which tells me the vocabulary their buyers use.

Rules: quote them verbatim wherever you can, put a source URL next to every claim, and write "not published" instead of estimating anything you could not find on a page. Finish with the three things they do better than the average competitor in this market and the two places their message is weakest.

Search the Meta Ad Library for [brand name] and read what they currently have running. If you cannot load the library directly, tell me plainly and I will paste screenshots of the results for you to analyse instead.

For every ad you can see, give me: the hook or first line, the format (static image, carousel, video, or UGC style), the offer or call to action, and how long it has been running. Group ads that are clear variations of one another so I can see which angle they are testing hardest.

Then answer three things. Which angle are they betting the most on, judging by how many variations and how long they have been live? Which audience or objection is each angle written for? And what is the one angle nobody in this market is running, that I could own?

Fetch the raw HTML of [competitor URL] and tell me which tools and platforms you can positively identify from it, such as the ecommerce or CMS platform, the email or SMS provider, the review widget, the chat widget, the analytics and any advertising pixels.

Only list something if you can point to the specific evidence in the page: a script source, a class name, a meta tag, or a visible widget. Show me that evidence next to each tool, and mark anything you are unsure about as "possible" rather than confirmed.

Then tell me what their stack implies about how they operate: roughly what that stack costs a month, whether it suggests a small team or a big one, and which two tools I would need if I wanted to match their setup.

Search the web for public reviews of [brand name] across review sites, app stores, marketplace listings and forum threads, and read the actual review pages rather than summaries of them.

Pull out the twenty most repeated specific complaints, in the customers' own words, and group them into themes. Ignore anything vague like "bad experience" and keep anything concrete, such as a delivery time, a missing feature, a support delay or a pricing surprise. Put a source link next to each theme.

Then do the same for the praise: what do happy customers name as the reason they stayed? Finish by telling me which of their complaint themes I could credibly fix or avoid in my own offer, and write me the one sentence of marketing copy that speaks directly to the biggest one.

Using everything in this chat, the teardown, the ads, the stack and the reviews, give me three gaps I can attack in the next 30 days. A gap is something their customers clearly want that this competitor is not offering, not saying, or not saying well.

For each gap, give me: the evidence from our research that it exists, the specific thing I would do about it (an offer change, a piece of content, an ad angle or a page rewrite), what it costs me in time or money, and how I would know within 30 days whether it worked.

Then rank the three by how fast I could ship them, and write the first draft of the highest-ranked one so I have something to edit tonight rather than a to-do list.