How I'm using LLMs in my SEO work (mostly without 3rd party data)

   By Chris Byrne (6.26 - updated 9.26)

Note I have broken some URLs below.

   Other than sometimes using LLMs to summarise long articles / research (which some might consider lazy and not the optimal way to learn), I am also using Anthropic's Claude or Google's Gemini in my day-to-day work with simple quick prompting in the following ways (mostly without using 3rd party data and without using MCPs) for qualitative data analysis:

1.  using Google Gemini to research the network of websites of a lead generation agency (competing with one of my clients) disguised as what appears to be B2C vendors . Lead generation agencies are not eligible to be listed on the Google Map (aka Google Business Profile), thus we can report their Map profiles to Google . 

  R.e. disguised lead gen websites created by lead gen agencies with Bing Map listings (aka Bing Places): these listings do not provide a “physical address open to customers OR provide an area of service where employees travel to visit the customer” thus are not eligible to be listed on Bing Places as far as I can see.

We can also report "pages that do not deliver promised online or real-life services with intent to mislead or defraud the user" to GOOG using https://search.google.com/search-console/report-spam. This requires identification of one of the lead generation websites first.

2. using Claude to analyse the authenticity & likelihood of incentivisation of competitors' Google Business Profile aka Google Map listings (this requires uploading / pasting the Google Map reviews in question) / Trustpilot reviews and reporting these reviews to Trustpi1ot and GOOG. Fake & undisclosed incentivised reviews are now illegal in the UK, so you can also report them to the UK's Competition and Markets Authority (CMA) .

For Trustpilot review analysis you now have to copy and paste the reviews as they now block Claude.

   You can also use Gemini with a prompt like “analyse the authenticity & likelihood of incentivisation of following reviews. analyse all reviews and reviewer's profiles on https://uk.trustpilot .com/review/xcvwindows.co.uk for suspicious patterns . do same for all other pages e.g. https://uk.trustpilot .com/review/xcvwindows.co.uk?page=2" 

3. using Gemini to analyse competitors backlink profiles for what are non-editorial (i.e. paid backlinks) which are non-compliant with Google and / or Microsoft Bing guidelines. This involves using 3rd party data. You can use a prompt like:

"analyse following urls and page content to determine if the website is part of a pbn [paste urls]" 

You can report the relevant links to Google & Microsoft.

4. using Claude to analyse websites in the light of Google's relevant UX guidelines for websites in that sector . You can use a prompt like:

"analyse this [ecommerce] website https://www.XXXX .com/ in the light of https://services.google .com/fh/files/events/pdf_retail_ux_playbook.pdf" . 

  All of the above have provided useful feedback and saved time - not all the inference / analysis provided by these tools is correct. With all feedback from these tools you need to check the facts.