How to use Claude for competitor research

Start with a decision

Suppose you run a small coaching business and want to promote a new workshop. The question is not ‘What are my competitors doing?’

It is ‘What do their workshop pages explain that mine does not?’ That gives the research a finish line.

Choose three businesses serving a similar customer.

A celebrity with a huge audience may have interesting content, but their reach is not evidence that the same approach will work for you.

Once you know what works, the free 7-day posting routine turns it into a week of posts.

Collect comparable evidence

For each business, collect its workshop page, pricing if public, and three recent posts promoting the offer.

Paste the text into separate documents. Put the source URL and collection date at the top.

Use the same time period and content types for every competitor.

Start at Claude and attach those documents to a conversation. Supported uploads are described in Claude’s file guide.

This walkthrough uses supplied documents; it does not assume Claude can access every social account or private metric.

Ask for a comparison you can check

A useful table has one row per business and columns for customer, promise, format, price, proof, objections answered and call to action.

Every factual entry should point back to a source. ‘Not stated’ is a useful answer: it shows where you need more information.

Watch for inference masquerading as fact. A competitor saying ‘for busy founders’ tells you their positioning.

It does not prove that founders buy their offer.

Turn the comparison into a better page

Illustrative finding: two competitors show what participants leave with, while your page only describes the topic.

The next experiment is a before-and-after example of the thing attendees build—not a new price or a complete rebrand.

Ask Claude for three possible improvements, with supporting evidence and effort involved.

Choose one, record what you changed, and compare results over a useful period.

If you have very little traffic, start with direct feedback from likely customers instead of declaring a conversion winner.

Make the next review easier

Save the comparison and the original sources.

Next time, collect updated versions of the same pages and ask what changed. A Claude Project can hold reference documents and instructions.

Saving context is different from scheduling a recurring check; automatic collection needs a separate setup.

For larger research batches, see which tools handle collection and analysis.

For turning findings into posts that sound like you, read the content voice guide.

Try this prompt.

Compare these three businesses using only the attached sources. My decision: improve the landing page for a live workshop. Build a table covering audience, promise, deliverable, format, public price, proof and CTA. Cite the source filename and relevant excerpt for each entry. Write ‘not stated’ for missing information. Separate observations from your interpretations. Then suggest three changes I could test on my own page, with evidence, effort and what result to measure. Do not infer sales from views or follower counts.

Copy this into your assistant after adding the source material described above.

Keep going.