Ecommerce

The ChatGPT SEO Playbook

It’s a recurring theme at the moment, “Dan, how do we ensure we are optimised for visibility in ChatGPT?” or “How do we SEO for ChatGPT?”.

I get it it’s important, and at time of writing ChatGPT is the favourite to become the Google or Hoover of next gen information access on the web. This is an exciting opportunity for old wizards like me who are obsessed with understanding how these systems work.

It’s maybe a less exciting time for more process & structure led SEO agencies. They have to find a way to productise something quickly within their billable frameworks, despite there not being enough known about how to predictably rank in GhatGPT yet. Plenty of high profile agencies are already flogging their “GEO” solutions, I’ve yet to see one that really nails it.

That said, I’m going to plant my own flag somewhat with this article and suggest that there are a couple of things that will give you a big advantage when it comes to being seen in this brave new world:

  1. Traditional SEO Skills – yep, I’m firmly in the camp that “GEO” is actually just “SEO”, just not quite SEO as you know it. Your keyword first processes won’t cut it here, but most of the techniques that you have spent years aquiring (choughdecades in some cases) are still important
  2. Marketing Skills – hate to break it to you, but your average SEO agency lacks much of a grounding in marketing. We’re no longer in a world of links and traffic, so much as marketing yourself to the machines
  3. Passion and Curiosity – yes, I know everyone claims to have this, but to echo the Beckham meme “be honest”. How much time are you really spending diving into this stuff? This isn’t some grind-culture 5am club 80 hour week hustlebro bullshit, it’s that if you are from the generation who learned SEO in an agency following SOPs and relying on SEMRush then you’ve got to change your mindset quickly

Personally, I’ve not been this excited about the opportunities arising from a fundamental shake-up in marketing since “the year of mobile” finally (after about half a decade) became a reality. Putting my money where my mouth is, I’ve made a hell of a lot of time for step 3 recently (I’ve been around long enough that we can consider 1 & 2 a given), and I figured why not jot down what we know so far. If you are involved in digital marketing, then you need to be thinking about this stuff.

One thing is certain, in 6 months time I’ll re-read this and cringe, but for now here is my attempt to make sense of SEO for ChatGPT, starting with THE CHAT GPT SEO Playbook

What does that mean?

Well, if you are going to use SEO techniques to gain visibility in ChatGPT, it’s going to be pretty important for you to understand why and how the bot goes out to the open web during chats. That’s what I’m covering here. I’ve witnessed many people make many incorrect statements about this (“If you rank on Google you’ll rank in ChatGPT” anyone?) so let’s break it down.

Imagine you are in a conversation with ChatGPT (or one of the other adjacent OpenAIs like Copilot) and you ask it a question that results in a load of recommendations coming back, as an ecommerce SEO specialist I’m mainly interested in how it recommends the right product for you. So how does it do that, and how does that differ from a human web searcher on a traditional search engine?

Broadly speaking ChatGPT always follows these steps, aka The ChatGPT Search Playbook:

Infer first what you are actually asking

ChatGPT starts by reading your wording as a kind of brief. Terms like “budget”, “UK”, “beginner”, “quality” act as clues to what matters. From there, the AI maps your question to a simple framework, for example “best” might be interpreted as performance, reliability, ease of use, cost of ownership, support, safety, and availability.

If the topic is stable – say, “what specs matter for a beginner guitar” – then the robot stays offline. It answers directly using its internal knowledge and defaults to a “best for most people” recommendation.

Decide if the web is actually needed

The AI doesn’t browse unless there’s a good reason. It goes online only when the answer could be wrong or out of date.

Things like current models, pricing, recalls, warranty terms, or UK-specific standards. If the topic is evergreen, the robot keeps the workflow tight and fast.

When the AI does browse, a common misconception is that it simply goes to Google on your behalf, which leads to a lot of people incorrectly assuming that ranking on Google is still the magic bullet. But what it actually does is quite different.

Targetted Retrieval

The bot isn’t Googling for you, and the bot doesn’t hit one giant fire-hose. It actually builds from Bing’s search index as a foundational layer – meaning if a page isn’t indexed by Bing, it’s much less likely to surface in ChatGPT’s results. That’s right, if you are ignoring Bing with your SEO efforts, you’d better change track.

On top of Bing, the AI also uses OpenAI’s own crawlers and other trusted feeds to enrich, verify and expand beyond purely Bing-indexed data. However, as it stands, your primary entry-point to visibility in ChatGPT is wide ranging visibility in Bing.

So, what do we mean by “targetted”? Well, where you, as a human, might go to Google and search “electric guitar for sale”, ChatGPT would construct far more detailed searches. It might specify features, limit the search to trusted retailers, or construct verification searches that are designed to ensure the initial retrieval contains trustworthy information. As an example it could be using any combination of these verticals and more:

  • General Web – manufacturer sites, documentation
  • News – for time-sensitive updates or regulation
  • Product Data – spec sheets, trusted retailers
  • Finance / Metadata – when cost or company context matters
  • Images – for visual confirmation when relevant

How many search results does the robot typically retrieve? This is another key difference between SEO on Google and SEO for ChatGPT. Your absolute ranking for a keyword isn’t quite so important.

The AI focuses on the top 15-20 Bing results for a given query, however this varies based on how specific the query is. It can be as low as the top 8, or as high as top 30. Either way the days of “ranking top 3 or nothing” are on the way out.

So in practice, the robot is likely retrieving around two twenty or so top results from Bing (plus maybe a few extras for verification) and then selecting the most relevant sources from that set. As it’s performing multiple queries the trick is to appear often for a range of searches around your focus topic.

How ChatGPT Searches Differ From a Human’s

Most people start broad (“best X 2025”) and then skim the top results. The robot structures its searches differently:

Builds tight queries with entity + attribute + qualifier (e.g., ModelName warranty length site:manufacturer).

Uses filtering by domain/region (site:.uk, manufacturer domains).

Adds disambiguation tokens like “datasheet”, “manual”, “CE/UKCA”, “BS EN”.

De-duplicates aggressively, avoiding low-trust blogs or affiliate spam.

Runs verification queries just to confirm claims (manufacturer PDFs, standards lists).

Performs regional checks for UK stock, compliance, plug type, VAT.
In short: it is running an accelerated research phase using smarter queries that surface the right source fast.

Score Against Your Intent

Once retrieval is done, the AI takes your inferred brief + the verified facts and scores each option against what you care most about: performance, reliability, price, or support. It highlights trade-offs – “cheaper but shorter warranty”, “better spec but no UK stock”.

Re-Rank As Context Changes

If you later add “my budget is £600” or “needs UKCA compliance”, the robot re-weights and re-ranks. It keeps the logic, just refines the priorities.

This means that you need to ensure your information architecture matches a deep contextual hierarchy. You may be recommended early in the conversation with ChatGPT, but if the potential customer adds requirements, if you meet those requirements you’ll need to ensure that you communicate that.

Example, if the customer narrows down their options, and then asks ChatGPT “find one with next day delivery”, well even if you offer that service, if it isn’t clear enough to a stupid, literal robot then you might lose the sale.

In Summary

ChatGPT doesn’t just hallucinate answers – it uses a disciplined workflow. It infers what matters, only browses when it should, builds on Bing’s index (and its own tools), typically retrieves from the top ~15-20 Bing results, performs structured searches, verifies facts with primary sources, and tailors the result to you.

It’s methodical, transparent, and built to save time – it is built to emulate how you would perform research for a considered purchase, only it does it so much faster.

Practical Takeaways

This is all very well Dan, but what can we do with this information? Well, here’s my adapt to the playbook playbook

  1. Figure out what you have a right to be visible for – product, market fit, those traditional marketing skills we talked about earlier
  2. Replace “keyword research” with “market research” – when people in your target market are considering your product or service what are the “best x for y” queries they are likely to be using to understand what is the right product for them? Run as many of these searches as possible and identify the common features that are referenced. Build a matrix of the product data that is referenced and count up the mentions to weight them
  3. Ensure that you surface the important product data in a structured format on your site
  4. Pay attention to your trust signals – all other things being equal, you need to need to look like a quality website, good reviews, established history, good stock coverage…aaand we’re back in traditional SEO land

I hope you’ve found this overview useful. Right now, nothing is certain in this space, but if you’d like to discuss how to apply some of these learnings without diverting budget away from the channels that are delivering for you today, it might be a good time to chat to me.

Protect the present, while taking care of the future is how Coleman the Wizard works.

Good luck.

Coleman Marketing Services

Digital Marketing Consultancy

Ecommerce Consultancy

  • Project Specification, Discovery & Requirements
  • Technical Architecture
  • Martech Integrations
  • Conversion Optimisation
  • Platform and Supplier Selection
  • Agency Relationship Management

Ecommerce Platform Expertise

I've worked with most major platforms including