The Ultimate AI Shopping Guide For Ecommerce Managers
😎 TLDR
- Track target questions on a GEO tracker: Research questions customers ask AI in your problem space and track how your brand is showing up and what citations AI uses.
- Get mentioned positively in the citations AI uses: AI chooses brands that are spoken the most highly of in the highest authority websites in your problem space.
- Become a citation AI uses: AI chooses brands with topical authority content that answers questions and helps make decisions in concise and structured ways.
- Optimise your data for the AI comparison showdown: AI chooses brands that best meets the customers needs with the most detailed, structured and trustworthy data.
🤔 How to think about it
95% of the time when you search for a solution online you just pick one of the top 3 options on Google or Amazon. Maybe you’ll have a look at some reviews or skim a blog post if you have time but then you just buy it and move on with your life. This is the world of SEO. It’s dominated by Google and Amazon who collectively made €220m this year from search ads.
But then remember that time you bought that thing that you really really really cared about. You started on Google but spent hours on Reddit, days on Youtube, checked reviews on Trustpilot or G2, ended up in deep dark blogs or news from 6 years ago or some weird Facebook or LinkedIn community or even Boards.ie somehow where you found the answers to the nichest of questions about the tiniest insignificant details. You probably got the very best solution on the very best package at the very best price for your exact use case. Now imagine you had an assistant who was offering to do all that research for you, except they could do it in minutes instead of hours and they could do it for free. You’d get them to go through that intense research process every time you wanted to buy something right? This is the world of GEO and ChatGPT is that assistant. It follows the exact same intense research process you would do had you the time, it just does it a lot faster and assesses things like a robot rather than a human.
The leads Webflow get’s from GEO convert 6x better than leads they get from SEO
🧐 What’s changing
Turning up on Google → Turning up everywhere
- A lot of the search focus has been distributed to Reddit, Youtube, Trustpilot, Forbes, industry blogs, etc. ChatGPT doesn’t even use Google to search, they are currently moving from Bing to their own search engine. Ranking high in search engines (SEO) is still important because AI still starts with a series of web searches but it’s now also critical to find out what citations (websites) AI uses to answer questions about your problem space and build strategies for turning up on those websites.
- There’s no point turning up everywhere if everything being said about you is bad. Brand and product sentiment across citations is equally important.
- Brands are experiencing a nose dive in SEO traffic, especially since Google AI overviews and AI chatbots completely skip search ads. They’re now shifting a huge amount of the SEO and search ads budget into GEO strategies.
- While it takes years to build up the domain authority to turn up on Google, a new brand with a buzzy launch that gets mentioned on a bunch of relevant forums or news outlets can start turning up on ChatGPT straight away. This is one of the major differences between backlinks and citations, small guys can win early.
Optimise for keywords → Optimise for questions, decision making and topical authority
- AI is searching a lot more questions with a lot more detail than traditional SEO. The long tail of SEO (longer more complex queries) that was previously insignificant is now very significant.
- AI very often generates comparisons between a shortlist of brands. It’s important to understand how your brand and products look in these comparisons and how your competitors look beside you. Creating decision making content including competitor comparisons, technical breakdowns, use cases, etc. can help influence the narrative of these comparisons. Understand how you’re differentiating and communicate it strongly.
- AI has a much better understanding of topics than traditional SEO so there is no need to target specific keywords anymore, AI will understand when users are searching for a specific topic no matter what keywords they use.
- AI has a much better understanding of authority than traditional SEO so it greatly rewards brand who position themselves as authorities on topics with topical authority content and EEAT (experience, expertise, authority, trustworthiness).
Webflow achieved a 24% boost in AI visibility from onsite optimisation alone
Catchy slogans & lifestyle pics → Structured data & trust signals
- AI prefers sites that are fast, accessible, technically compliant, good schema markup, etc.
- AI prefers structured data and rich factual details over marketing fluff. It’s focus is to find the best possible solution that matches as many customer details as possible so the more detail, the more attributes, the more info around customer concerns the better. Customers get very detailed in these conversation so you never know what detail you might lose out on.
- AI prefers and quotes data and claims being backed up, verified or proven. These are called trust signals and they can be certifications, standards, awards, influencer/expert endorsements, reviews, case studies, press mentions, mentions across the web, external links to authoritative resources, etc.
🤓 How it works
1. Customer conversation
AI shopping conversations tend to average about 90–140 words from the customer. They might start with “Best light waterproof hiking boots 2025” but as the conversation goes on they often add that they would prefer them to be brown, under €200, long lasting, have some specific strap or cert, from a somewhat ethical brand, etc. The AI is essentially recording all these attributes on scorecard and will be trying to tick each box as comprehensively and as confidently as possible.
2. Problem exploration
Each time the customer replies the AI will make a series of web searches, “best waterproof hiking boots 2025”, “types of waterproof certifications”, “gore-tex more expensive?”, “best light hiking boots brands”, “hiking boots weights light medium heavy”, “light hiking boots reviews 2025”, etc. Again, think of all the searches you would be making if you were willing to spend all day at it. Most AI’s don’t show you the searches they are making but there are some Chrome Extensions you can install to uncover these searches. As you can imagine, here it is mostly ending up on educational and analytic content like social forums, third-party review sites, blogs, expert/influencer takes, news, etc. Each time it searches it first finds a collection of “citation candidates”, it then reviews each candidate for depth, EEAT, consistency, etc. and shortlists a few chosen citations to actually generate it’s answer from.
3. Shortlisting options
Next, once the AI is happy with it’s understanding of the landscape it will start searching with more buying intent. At this point it will be searching recommended brands, recommended types of hiking boots, recommended certs or standards like gore-tex, but it will be looking for product pages this time. Again here it will choose citation candidates and shortlist it to citations based on the product attributes matching the customers needs, data consistency with other citations, trust signals like reviews, certifications, endorsements, etc.
4. Generating an answer
The AI will then take the knowledge bank and products it has shortlisted and form an answer directly answering what the customer asked. It will be looking for citations to quote in it’s answer, especially in answering questions and concerns the customer had.
🤠 How to win
Winning an AI shopping purchase essentially amount to, firstly, getting your brand into the citations AI uses to generate it’s answers, either through external sites (step 2) or through your own site (step 3), and secondly, getting your brand shortlisted among the options AI finds (step 4).
1. Track target questions on a GEO tracker
Research questions customers ask AI in your problem space and track how your brand is showing up and what citations AI uses.
- Research the kinds of questions your customers are asking on google “people also ask”, customer reviews, sales data, keyword data, etc.
- A good topic or question is: popular but not too competitive (keyword data), consistent with your brand (AI will drop anything not consistent across platforms), has good buying intent, etc.
- Track customer prompts around your target topics and questions on a GEO tracker. These are new tools that help you track how your brand is turning up in AI conversations. They include how often your brand is visible, who are turning up as competitors, your ranking vs competitors and sentiment around your brand. They also crucially track what citations (websites) AI is using to find brands and generate these answers.
- One important thing to note in tracking progress in general is that AI attribution is not very reliable. It’s very common for AI users to discover a brand or product in a conversation and then go directly to their URL rather than clicking through the UTM. If you want to track AI attribution better you could include a “how did you here about us” form.
2. Get mentioned positively in the citations AI uses
AI chooses the brands that are spoken the most highly of in the highest authority websites in your problem space.
- Adopt strategies for getting your brand turning up on the citations coming up for your tracked prompts. Both owned and unowned content are useful. For owned content, you could create Youtube/Vimeo videos answering unserved tracked questions, post blog on LinkedIn or Medium, launch on ProductHunt or Kickstarter, become very active in relevant Reddit forums providing real value and signing off with your name, your role and your company (EEAT), etc. For unowned content, you could form partnerships with relevant blogs or industry publishers, form content sharing partnerships with partners, reach out to popular listicles for inclusion, work with influencers/experts, etc.
- Adopt strategise to control customers sentiment across the web. Think taking control of any review platform and replying to reviews, responding publicly to any negative commentary, being more explicit in product information about anything getting negative reviews (long delivery times), generally upping customer support and being explicit about how it’s improved, etc.
3. Become a citation AI uses
AI chooses brands with topical authority content that answers questions and helps make decisions in concise and structured ways.
- Some of the most impactful work you can do here is just traditional SEO work. AI still starts by querying search engines like Google and if you are ranking high in these search results you will be very likely to be included in the citations where AI will consider including you in its answer.
- AI prefers content that is Q&A based, well structured, fresh, brand consistent and has topical authority.
- Since AI is searching a lot more questions with a lot more detail its now very worthwhile to be answering as many questions as possible in a concise, comprehensive and quotable way (AI prefers 40-60 word snippets). Across all your content, onsite and offsite, across every different format be it landing pages, blogs, product pages, news articles, help centres, social forums, etc. you should be thinking about answering as many of the most relevant questions as possible in the best way possible. For quantity think FAQ’s, definitions, glossaries, etc. For relevance think comparisons, listicles, HowTo’s, use case language i.e. “for mothers”, “for sensitive skin”, etc. For quality think rich factual details, structured data tables, bullet lists, etc.
- AI prefers content that is well structured. So this is mostly basic SEO stuff but brands have seen significant lifts in AI visibility by changing the title of a blog to a question, including a keyword in the title of a blog, adding examples to a blog and improving internal linking.
- Since there is more content out there than ever before AI prefers content that is fresh. Updating content regularly also gives you the chance to catch recent trends. For example, adding a FAQ about whether the product has high protein as protein becomes a popular attribute.
- AI will drop any content that isn’t consistent with your brand across platforms so prioritise topics and messaging that is consistent for your brand and new content is strongest when distributed in many forms and media formats across platforms. Think publishing a blog on your website but also a supporting blog on Medium, making a Youtube or Vimeo video on it, getting some PR in the news, having partners share it on LinkedIn, etc.
- To optimise for topical authority and EEAT (experience, expertise, authority, trustworthiness) think topic clusters, deep dive expert content, expert authors, unique content, etc.
- To influence AI comparisons its now very worthwhile creating content that directly targets these comparisons and helps with decision making. Think comparison landing pages vs your main competitors, content on your differentiation, your product attributes, your sustainability, your integrations, management, after care, etc. The more the better since customers get very detailed in the conversations so you never know where you might win or a lose a comparison.
4. Optimise your data for the AI comparison showdown
AI chooses brands that best meets the customers needs with the most detailed, structured and trustworthy data.
Technical
- Ensure your site properly allows and introduces AI with Robots.txt & LLMs.txt
- Ensure your site is fast, accessible, SEO compliant, standard structure including breadcrumbs, knowledge graphs, secure checkout, standardised everything, etc.
- Max out your site schema/json-ld data as much as possible. This is the structured data layer of your site and what AI understands as the source of truth for your brands and products. Any data missing here will result in missed opportunities. For example, if a consumers prompts “coffee beans delivered within 4 days” AI’s will keep searching though products until they finds one with delivery info in the sites product schema data even though most brands don’t currently use this part of the schema. You should be commonly using schema elements like FAQPage, ItemList, ProfilePage, etc.
Medium achieved a 20% boost in AI visibility by improving their structured site schema data alone
Attributes
- Again, since AI’s focus is to find the best possible solution that matches as many customer details as possible, the more rich factual details and product attributes you have included the better. The winning and losing of products turning up or products losing the AI comparison table battle is very often decided in how you present your brand and product attributes. Also, customers often get very detailed in these conversations so you never know what attribute you might lose out over. Think clear product specs, clear differentiators, quantifiable attributes (weight, dimensions, battery life, etc.), integrations, management, after care, etc.
- Try to simulate the experience your customers are having during AI shopping. When they look for your category what is turning up and why? When they ask for attributes you’re strong on (local, sustainable, long lasting, for beginners, etc.) ask why the AI chose those brands or products for those attributes. This will uncover AI’s reasoning logic, sources and how you can optimise.
Trust signals
- A big part of having strong attributes is having strong trust signals for those attributes. These include certifications, standards, awards, influencer/expert endorsements, reviews, case studies, press mentions, mentions across the web, external links to authoritative resources, etc.
- AI has a strong incentive to prioritise brand trust and customers experience since it gets a lot of the blame when things go wrong. They prefer brands with good return policies in the site schema data, good customers service reputation, etc.
🫡 Tools that help
1. Track target questions on a GEO tracker
Ecommerce specific
- azoma.ai (3m raised)
- wild-card.ai (YC)
- higoodie.com
General
- tryprofound.com (59m raised)
- peec.ai (30m raised)
- bluefishai.com (24m raised)
- scrunch.com (19m raised)
- evertune.ai (15m raised)
- brandlight.ai (6m raised)
- searchable.com (4m raised)
- athenahq.ai (2m raised)
- gauge.ai (2m raised)
- telepathic.com (1m raised)
- 70+ more → graphite.io/five-percent/aeo-tools
Old guard
2. Get mentioned positively in the citations AI uses
Unfortunately (thankfully!) there’s no way to hack turning up across the internet. Many startups have tried to automate this in different ways but they are always eventually blocked by the platforms they’re automating or the LLMs find a way to bypass them. Similar to the years of whack-a-mole between Google and SEO hackers. This is an area for smart innovative strategies rather than technology.
3. Become a citation AI uses
These are new solutions or newly pivoted solutions specifically for targeting AI citations. Obviously there is also a whole industry of tools in the content space who will all be moving in this direction.
- airops.com (60m raised)
- frase.io (10m raised)
- surferseo.com
4. Optimise your data for the AI comparison showdown
Simulate and optimise the comparison showdown
- upshelf.io (that’s us!)