AEO (answer engine optimization) is everyone’s favorite buzzword. Brands ask us every day how to get their business to show up in LLMs for relevant prompts.
We’ve helped dozens of ecommerce brands with AEO. Today, we’re diving into the AEO audit we conduct for DTC brands as phase one of maximizing their AI visibility.
Whether you want Intergrowth to set up a proper AEO foundation or you want to take the DIY approach, here’s everything you need to know.
Expectation Setting: Can You Influence AI Visibility Today?
In some cases, you can influence AI visibility in the next 3-6 months. In others, you’re playing a longer game.
Here’s how to determine how long it will take you:
Write a list of questions that you hope for LLMs to recommend your brand in response to (ex. “what’s the best razor for men?” or “what’s the longest-lasting razor that costs under $100?”).
Run them through ChatGPT, Claude, Gemini, and Google AI Overviews. Watch how each LLM builds its response.
Three ways that LLMs answer questions :
- Parametric memory only. The LLM answers entirely from what it learned during training. No web search happens. Its own internal memory is the only source the LLM uses.
- Retrieval-augmented generation (RAG). The LLM runs a live web search to inform its answer. This is the scenario where your on-page content, your product feed, and your recent press mentions can move the needle right now.
- A hybrid of both. Training data forms the backbone of the answer, with a web search layered on top to fill gaps or verify details.
Why does this matter?
If an LLM answers your priority queries from parametric memory alone, no amount of schema markup or product feed optimization changes what it says today. That content lives inside a model that was trained months ago and won’t retrain tomorrow.
LLMs don’t automatically expand their parametric memory. This only happens when developers add to their training set or update existing memory.
If an LLM doesn’t run a live web search to inform your priority queries, nothing you do will influence whether your brand shows up until that LLM expands its training data.
This is exactly what our audit’s query fan-out analysis does for you. We run this test across up to three of your core solution sets, not just one guess at a priority query, which gives you a clear read on where your AI visibility actually sits today.
There’s a second output most brands don’t expect: this same exercise flags which traditional search engine queries deserve priority ranking attention. If LLMs are answering a query from parametric memory alone, that query’s near-term opportunity is still sitting in Google, and we’ll tell you to focus your SEO budget there instead.
The Crawler Mistake That Locks You Out of Parametric Memory
Many brands ask whether they should block AI bots from crawling their site (via robots.txt). We strongly recommend against it. AI crawler access is not the place to play it safe.
There are two types of LLM crawlers to audit (we’ll skip agentic crawlers for now):
- Retrieval crawlers
- Training crawlers
Retrieval crawlers, like OAI-SearchBot, PerplexityBot, and Claude-SearchBot, search the web for prompt responses using RAG.
Training crawlers, like GPTBot, ClaudeBot, and Google-Extended, analyze pages to expand LLMs’ parametric memory.
If your website’s robots.txt file blocks the retrieval crawlers, your webpages won’t help you get cited in LLM prompt responses that run RAG. If your robots.txt file blocks training crawlers, your webpages won’t be included in an LLM’s future training data.
Why you should avoid blocking training crawlers
LLM training sets keep expanding. The version of your business trained into the next major model update depends on what these crawlers could access when they ran. Block them now, and you’re absent from the next generation of parametric memory, not just invisible in today’s answers.
Check your robots.txt file at yourdomain.com/robots.txt. Confirm you’re not disallowing the training or retrieval crawlers anywhere in the file. Most CMS platforms don’t block these by default, but plugins and past developers sometimes add blanket AI crawler rules without realizing what they cut off.
This is one piece of a larger crawlability and indexability picture. If you haven’t audited discoverability, crawlability, and indexability recently, start here.
Schema Markup Is Training Wheels, Not a Cure-All
Schema markup, particularly SameAs schema, is one of my favorite tools for helping brands get more customers from search engines and LLMs. It’s driven massive wins for several clients and can be put together in under an hour. Add it correctly, and you’re telling machines that your website, your Instagram, your YouTube, and your other brand assets all belong to the same entity.
We’ve watched this translate directly into revenue for ecommerce clients. One example from our own work involved a client with a strong YouTube following and comparatively little website traffic. Adding relevant schema connected the two properties in the eyes of search engines and LLMs, and rankings moved within 48 hours.
However, schema isn’t a “cheat code” that always helps businesses double overnight.
Schema markup is training wheels for machines like Google and ChatGPT.
Coca-Cola doesn’t need SameAs schema. LLMs and search engines already spend enormous resources analyzing every corner of Coca-Cola’s web presence. A line of code confirming that the Coca-Cola website and the Coca-Cola Facebook page belong to the same business doesn’t tell them anything new.
Your $3M ecommerce brand is a different story. LLMs and search engines haven’t spent years indexing your business the way they have for Coca-Cola. Schema gives them the entity clarity they need to connect the dots on their own.
Product Feeds Are Becoming the New PDP
Product feed hygiene used to be a traditional SEO lever. It’s turning into something bigger.
Agentic commerce is still in its early stages. In the near future, we’ll likely see AI agents that shop on a customer’s behalf, compare products, and complete purchases without a human clicking through a category page. Product feeds are shaping up to be the primary way these agents understand what you sell.
If you haven’t logged into your product feed backend recently, start there.
Shopify makes this easy to access. Pull up your feed and check your top priority product pages first. Look for completeness and accuracy:
- Are titles specific?
- Are descriptions detailed
- Are prices and availability current?
An incomplete or inaccurate product feed hurts more than just Google Shopping visibility. It’s the data source AI agents and LLMs increasingly draw from to understand your products, your pricing, and whether you’re a fit for what a customer is asking for.
Your PLPs Need to Get More Specific, Not More Generic
Product listing pages carry more AEO weight than most brands realize. They’re often the clearest signal you give search engines and LLMs about the specific use cases your products solve.
A generic “Shoes” category page tells an LLM almost nothing about who should buy from you. A dedicated page for “Trail Running Shoes for Wide Feet” tells LLMs which narrow, specific query you’re the answer to.
LLMs reward specificity far more than traditional search engines. Broad category pages compete against every other broad category page on the internet. Narrow, well-built PLPs compete against almost no one.
Deindex the Pages Dragging Your Whole Site Down
Search engines and LLMs don’t evaluate your pages in isolation. They form an impression of your entire site, and thin or outdated pages can make your entire site look bad.
Here are some pages you’d want to deindex:
- An old blog post with three sentences of content.
- A filtered product URL generating a near-duplicate of your main category page.
- A discontinued product page still live with no redirect.
Every one of these tells search engines and LLMs your site has more noise than substance.
Your Shipping and Returns Policy Is AEO Content, Too
Ask an LLM to recommend a product that ships to a specific region within a week, or that uses eco-friendly packaging, and it can only recommend brands that have actually published that information somewhere it can find it.
If your shipping and returns policy is vague, buried, or missing key details, you’re invisible for every prompt where policy specifics are the deciding factor. That’s a meaningful category of purchase decisions.
Check whether your policy clearly covers shipping windows, regions served, return timelines, and any conditions that matter to your specific customers. If it’s thin, write an updated version that answers the questions that customers are asking.
Fixing Broken Backlinks
Every backlink pointing to a dead page on your site is wasted authority. The site that linked to you meant to vouch for something specific. That vote of confidence disappears when it points to a broken page.
Search engines and LLMs both treat backlinks and brand mentions as trust signals when deciding who to cite and recommend. A backlink profile full of broken links is authority you’ve already earned and are actively throwing away.
Run your site through Ahrefs and pull your full backlink profile. Flag every link pointing to a 404 or a dead page, then redirect each one to the next most relevant working page on your site (not just your homepage by default).
Core Web Vitals Are an AI Problem Now, Not Just a Google One
Site speed used to be a Google ranking factor and a conversion rate lever. It’s becoming an AI agent problem too.
As agentic commerce grows, AI agents that browse and shop on a customer’s behalf will favor sites that are easiest to navigate and parse. A slow, bloated page frustrates a human shopper and slows down the agents increasingly standing in for one.
Check PageSpeed Insights to see where your Core Web Vitals scores currently stand. If you’re falling short, compress your image files and convert them to WebP format. These are two of the most straightforward, non-technical updates you can make to improve load times, no developer required.
Know How LLMs Are Already Describing You
Everything above is technical. None of it tells you whether LLMs are currently saying good things about your brand, bad things, or nothing at all.
This is where LLM Sentiment Analysis comes in.
Ask an LLM:
- What it knows about your business.
- Whether it has your facts right.
- What it’s citing when it talks about your reputation.
- Whether it recommends you when a customer asks a broad or narrow product question.
We break down our four-part LLM Sentiment Analysis framework here, including exactly how to prompt each audit and what to do with what you find.
Consider building an llms.txt file.
This is a plain-text positioning document you publish at yourdomain.com/llms.txt, describing what your business does and who it’s for.
To be direct about where things stand as of September 2026: LLMs do not read this file today, and it won’t move your visibility in ChatGPT or Claude this quarter. The value is the positioning exercise itself, forcing your team to agree in writing on what you’re exceptional at and who you’re really for.
More on building an LLMs.txt file here (and why I recommend that challenger brands take an afternoon to build this), if you want to get ahead of it before it becomes standard practice.
See also: SEO vs. AEO (and the levers you can pull to influence rankings
Why We Recommend a One-Time AEO Foundation, Even at $500K in Revenue
AI-powered search is in its infancy. Those who invest in AI search visibility today will reap the returns for years to come as LLMs become a more popular shopping tool.
If your brand is generating $500K a year, you don’t need to commit to an ongoing AEO retainer. What you need is a one-time AEO foundation: schema in place, crawlers unblocked, product feed clean, sentiment understood. Then reassess.
Should a brand this size really be spending on AI visibility instead of channels like PPC with a proven return today? Fair question.
Meta Ads and other proven acquisition channels should still get the bulk of an early-stage ecommerce brand’s budget. AEO isn’t a replacement for that. But a one-time foundation is inexpensive compared with the downside of doing nothing while larger competitors lock in the crawler access, schema, and content structure that compound as LLM training sets expand.
Get the foundation right once. Then decide, with your own query fan-out data in hand, whether an ongoing investment makes sense for your business.
What’s Actually Inside Intergrowth’s Ecommerce AEO Audit
We break our AEO audits into four groups, based on what each piece actually solves. You won’t walk away with a single meaningless AEO score. You’ll walk away with fixes.
Technical foundation:
- Crawlability audit: confirming robots.txt allows both standard search crawlers and training bots like GPTBot and ClaudeBot.
- Indexability check: confirming meta robots and X-Robots-Tag are set correctly
- Discoverability optimization, including Google Search Console setup and sitemap submission.
- Broken backlink audit, with implementation of approved redirects.
- Core Web Vitals analysis, with image compression and WebP conversion recommendations if you’re not passing.
- Technical SEO audit, with up to 3 hours of direct error fixes
Structured data:
- Full schema markup audit, with custom schema built for your site and handed to your development team to implement
Content and product data:
- Product feed audit, including your Google Merchant Center integration, with up to one hour of direct fixes
- PLP Creation/optimization: optimization of up to four top-priority product listing pages, or recommendations for four new PLPs worth creating
- Shipping and returns policy review, with a rewritten policy for your approval if yours is thin
- Page quality audit, with confirmation that low-quality pages are deindexed
Sentiment and query behavior:
- Query fan-out analysis across up to 3 of your core solution sets, so you know whether LLMs are pulling from the web for your priority queries, and if so, which traditional search engine queries you need to prioritize ranking for
- LLM Sentiment Analysis, with recommended opportunities specific to your brand
Our AEO audit runs $4,700 for ecommerce brands generating up to $10M in annual revenue.
Conclusion
AI search is still early enough that a one-time AEO foundation, not an open-ended retainer, is the right first move for most early-stage ecommerce brands.
Unblock the right crawlers, add schema where it actually earns its keep, clean up your product feed and PLPs, tighten your policy pages, and find out what LLMs are already saying about you before you spend another dollar guessing.
From there, assess the long-term potential for investing in AEO. Get recommendations from a team like ours on whether you should invest in AEO, SEO, PPC, or a different channel entirely.
Contact us today to see if working with Intergrowth for an AEO audit is right for you.
AEO Audit FAQs
What is an AEO audit?
An AEO audit assesses whether your website and brand are set up to be found, understood, and recommended by LLMs and AI search engines like ChatGPT, Claude, and Gemini. It covers technical factors like crawler access and structured data, content factors like product feed accuracy, and brand factors like how LLMs currently describe your business.
How is an AEO audit different from a traditional SEO audit?
A traditional SEO audit focuses on rankings, crawlability, and on-page optimization for search engines. An AEO audit adds layers specific to LLMs: training crawler access, query fan-out behavior, structured data built for entity recognition, and sentiment across AI platforms.
The two overlap in plenty of places, but AEO asks questions traditional SEO never had to.
Should I block AI crawlers like GPTBot or ClaudeBot?
No. Blocking training crawlers doesn’t protect you today, but it does exclude your website from the training data that shapes how LLMs understand your brand in future model versions.
Confirm your robots.txt file allows GPTBot, ClaudeBot, and similar training bots the same access you’d give any standard search crawler.
Does an llms.txt file improve my brand visibility in AI search?
Not directly, and not today. As of September 2026, LLMs don’t read llms.txt files as part of generating their responses. The value is in the positioning exercise of writing one, not in an immediate visibility bump.
How much does an ecommerce AEO audit cost?
Intergrowth’s ecommerce AEO audit runs $4,700 for brands generating up to $10M in annual revenue. It covers technical foundation, schema markup, product feed and content audits, and LLM Sentiment Analysis.
How long does it take an AEO audit to show results?
It depends entirely on whether LLMs are pulling from the live web for your priority queries right now. If they are, on-page and technical fixes can influence answers within months. If your priority queries are answered from parametric memory alone, expect a longer runway, closer to 12-18 months, as LLMs retrain on newer data.
Is an AEO audit worth it for a small ecommerce brand?
Yes, even at $500K in annual revenue. AI search is still in its infancy, so a one-time AEO foundation is an inexpensive way to get ahead of competitors who wait. You don’t need an ongoing retainer at that size, just the foundational work done once, followed by a reassessment.