How to Audit Umbraco Content for AEO: A Practical Checklist
AI answers have changed how people meet your content. A reader may land on the whole page. An answer engine may retrieve one passage and place it beside other sources. Same content, much less context.
That can make Answer Engine Optimization (AEO) sound like a collection of new tricks: split every page into tiny chunks, add more schema, publish an llms.txt file, and hope an answer engine notices. Google’s current guidance is much less dramatic. Established SEO practices still apply to generative AI search, and there is no special AI markup or required content format.
I use AEO as the umbrella term in this guide. You will also see GEO, short for Generative Engine Optimization, used for much the same area.
I wrote the first version of this guide in March. Since then, the advice has become clearer: make the page genuinely useful, make the published result easy to discover and understand, show where important claims come from, and measure what happens after you improve it.
Use this guide to audit one important Umbraco page at a time. It will not guarantee a citation or an AI Overview. It will help you find the content, technical, and governance problems that are within your control.
Also, do not expect every improvement to produce a click. A page can appear as a source or influence a generated answer without sending a visitor to your site. Treat search impressions, citations, referral visits, and on-site behaviour as different signals. None of them proves on its own that the content influenced a buying decision.
Last reviewed: 30-09-2026
Source: Google Search Central: Optimizing your website for generative AI features on Google Search
How do you run an AEO content audit in Umbraco?
To audit an Umbraco page for AEO, check whether it:
answers a real audience question;
gives a useful answer early;
adds original expertise, examples, or evidence;
uses clear headings and coherent sections;
shows ownership, review dates, and sources;
uses important terms consistently;
is crawlable, indexable, canonical, and readable in the published HTML;
uses structured fields and blocks where they help editors work consistently;
connects to related pages and supporting sources; and
has a measurement and review plan.
You can use the 10-minute checklist near the end for a quick first pass. A proper audit may take longer when a developer needs to inspect templates, rendering, or structured data.
Key takeaways
AEO builds on SEO; it does not replace it.
Helpful, original content matters more than formatting tricks.
Clear structure helps readers first and can also help machines interpret the page.
Umbraco can support the workflow, but the CMS does not make a page visible by itself.
Schema supports SEO when it matches visible content; there is no special AEO schema.
Measure search performance, AI-search visibility, referrals, and on-site behaviour separately.
1. Start with the question the page should answer
I start every audit by writing down the main question the page should answer and who is asking it. Do this before you inspect headings or schema.
That sounds basic, but it catches a common problem: a page can be technically perfect and still avoid the question that brought the reader there. I still find pages where the introduction explains the company, the market, and the background before it gives the visitor anything useful.
Look at the language people already use in search data, sales conversations, support tickets, community discussions, and user research. Pick one primary intent for the page. Related questions can support it, but they should not pull the page in five different directions.
In Umbraco, check that the page name, hero headline, teaser, meta title, meta description, and opening section all point to the same intent. They do not need to repeat the same sentence. They do need to make the same promise.
What good looks like: A reader can understand who the page is for and what it will help them do before they start scrolling.
2. Give a useful answer early, and add something worth finding
Put a direct answer near the top when the page targets a clear question. A definition page should define the term. A how-to page should show the process. A comparison should state the important difference before the detailed analysis.
Do not stop at a polished summary. Search and AI systems already have plenty of generic content to choose from. I look for something grounded in real knowledge: an example, a limitation, a decision rule, a source, a screenshot, a short case, or experience from the team that does the work.
For example, “structured content improves consistency” is generic. Show the exact fields your editors maintain, such as answer, source, content owner, and reviewed date. Now another team can evaluate the idea and use it.
Here is the difference in practice.
Before: “An AEO content audit improves your visibility in AI search.”
After: “An AEO content audit checks whether a page answers a real question, supports important claims, can be crawled and indexed, and has a measurement plan. It improves the parts of AI-search visibility you can control, but it cannot guarantee that an answer engine will cite the page.”
The second version is not better because it contains more AEO terminology. It is better because the reader can see what the audit includes, what it changes, and where the promise stops. I still have to edit vague claims like the first one out of my own drafts.
Google’s guidance now calls this “non-commodity content”: useful material with first-hand knowledge or a distinct point of view, rather than another summary of what is already online.
What to check:
Does the page answer its main question in the opening section?
Does it contribute experience, evidence, or a useful example?
Does it explain limitations and exceptions where they matter?
Would the page still be worth reading if the visitor already knew the basic definition?
3. Structure the page for readers, not for imaginary chunks
Clear headings and focused sections make a long page easier to scan. They can also make individual passages easier to interpret when a search or AI system retrieves specific information.
That does not mean cutting every page into tiny “AI-friendly” fragments. Google says there is no required chunk size or ideal page length for generative AI search. Write for the reader and use structure where it improves comprehension.
Use one H1 for the page title, H2s for the main sections, and H3s only when a section genuinely needs a subsection. Replace headings such as “Overview” or “More information” with labels that name the subject.
I use a simple test: if a heading says “Overview”, can I tell what it is an overview of without reading the paragraph below it? If not, the heading needs more work.
Paragraphs do not need to pass an artificial no-pronouns test. They do need enough context to remain understandable. If a paragraph begins with “This”, “It”, or “They”, check whether the subject is still obvious when someone lands directly on that section.
In Umbraco, editors can apply headings in the Rich Text Editor. Repeating content patterns such as definitions, steps, evidence, warnings, or FAQs may be better modelled as blocks. Use whichever approach produces the clearest published page. The backoffice structure is a means, not the outcome.
4. Make important claims easy to verify
Trust is easier to judge when the page shows who wrote or reviewed it, when it was checked, and where important facts came from.
These details are not a magic citation signal. They are good publishing practice. They help readers evaluate a claim, help editors know who owns the page, and make future reviews much easier.
When you work with pages that can become outdated, consider fields for:
author;
reviewer or content owner;
published date;
last reviewed date;
relevant product or documentation version; and
sources for factual claims.
Make the information visible when it helps the reader. A last-reviewed field hidden in the backoffice does not build public trust, and a date that changes after a cosmetic edit does not prove the substance was reviewed.
I treat this as governance rather than decoration. If nobody owns the page, “we should update that sometime” tends to become the long-term content strategy.
5. Keep names and facts consistent across the site
AI answers can combine information from product pages, documentation, support articles, and third-party sources. Inconsistent terminology makes that job harder for people too.
Check whether your site uses several names for the same product, feature, audience, or process. Also look for facts that differ between pages: supported versions, availability, requirements, dates, or definitions.
I have lost count of how often I have seen three names for the same feature across a product page, documentation, and an old blog post. Each name may make sense on its own. Together, they create doubt.
In Umbraco, you can support consistency with controlled fields, reusable content where appropriate, editorial help text, and a maintained terminology guide. Do not force every sentence to use one exact keyword. The goal is a consistent meaning, not robotic repetition.
Repeating a target phrase does not make a passage more useful to an answer engine. Use the language your audience uses, then spend the rest of your effort on the answer, the evidence, and the reading experience.
For company and product claims, decide which source is authoritative and give someone responsibility for keeping it current. A polished page built on an outdated fact is still an outdated page.
6. Inspect the published output, not only the backoffice
Search and answer engines work with the page you publish. A tidy Umbraco editing experience cannot compensate for incomplete, contradictory, or inaccessible public output.
I always open the public URL. You should check:
the main content appears in the rendered page;
the heading hierarchy is coherent;
the page is allowed to be crawled and indexed;
the canonical points to the preferred URL;
important pages appear in the sitemap;
internal links resolve correctly;
metadata matches the visible content; and
old URLs redirect to the current page after a URL change.
Review robots.txt and any crawler rules in your CDN or web application firewall as deliberate policy decisions too. Search crawlers, AI training crawlers, and tools that fetch a page for a user may use different controls. Do not assume that blocking or allowing one user agent has the same effect everywhere, and do not switch everything on simply because it has “AI” in the name. First decide which uses you want to support, then confirm the current documentation for each service.
Umbraco records old URLs when publishing changes a document’s route and can use them to return a 301 redirect to the new URL. That helps preserve access after a move, but it does not remove the need to update important internal links.
Canonical and robots settings can have significant consequences. If you are unsure, involve the person responsible for technical SEO or the implementation before changing them.
Useful references:
7. Use structured content to make good practice repeatable
A large rich-text field gives an editor freedom. It gives the organisation fewer guarantees about sourcing, ownership, reuse, or consistency.
Structured fields and reusable blocks can give editors a stronger starting point. If you see the same pattern across many pages, consider modelling it once. Useful patterns may include:
a short answer or definition;
numbered steps;
key takeaways;
a claim with its source and review date;
a quote with attribution;
a warning or limitation; and
visible questions and answers.
In Umbraco, Block List and Block Grid can support these patterns. I prefer this approach when a pattern repeats across many pages. Editors do not have to remember the required pieces every time, and developers can render the same type of content consistently.
Structure does not create visibility by itself. A perfectly modelled block can still contain a vague answer, and a poorly implemented template can still produce weak markup. Always check the finished page.
Useful references:
8. Use schema as supporting SEO, not as an AEO shortcut
Structured data can help search engines understand eligible page types and can support rich-result features. It should describe content that is actually visible on the page.
There is no special AEO or GEO schema, and structured data is not required for Google’s generative AI features. That makes schema useful, but not magical.
For a blog article, the implementation might include BlogPosting, Person, Organization, and BreadcrumbList, depending on the site setup. Use only schema types that match the visible content and current search-engine guidance.
In Umbraco, I would generate repeated schema from maintained fields and templates rather than ask editors to hand-write JSON-LD on every page. If you add schema, validate the output and compare it with what a visitor can see.
Read more in What is Schema Markup?.
9. Connect the page to the rest of the topic
An isolated page is harder for readers to discover and gives them nowhere useful to go next.
I check links in both directions. You should add internal links from relevant pages, not only links out of the article. Connect the audit guide to technical SEO guidance, schema explanations, implementation documentation, and related articles about how teams should approach AEO.
Use anchor text that explains the destination. “Read our AEO content audit checklist” is more useful than “click here”. Do not repeat the same keyword-heavy anchor everywhere; write links that fit the surrounding sentence.
External authority also matters, but it is not something you can manufacture with an on-page checklist. Earn references by publishing material worth referring to, contributing expertise, and keeping important facts accurate. Chasing artificial mentions is a side quest you can safely skip.
10. Measure the result and decide what to change next
An audit is not finished when the page is updated. Record what you changed, when you changed it, and what you expect to learn.
If you are moving beyond a single-page pilot, establish a baseline before you edit everything. Pick a manageable first batch of business-critical pages rather than the entire website, and write down the main customer question for each one. Record the current search data, any identifiable AI referrals, and a small set of controlled prompt observations. Include the date, model, settings, answer, and cited sources. This gives you a repeatable starting point, not a perfect score for “AI visibility”.
Keep different measurements separate:
Search demand and discovery: impressions, queries, clicks, and landing pages in Search Console.
Generative AI search visibility: use first-party reporting where the platform provides it. Google’s Generative AI performance report in Search Console shows impressions for AI Overviews and AI Mode, with page, country, date, and device breakdowns. It does not prove that one content change caused the movement.
AI-referred visits: identify referral traffic where the source is available, while remembering that a mention without a click will not appear in web analytics.
Controlled prompt observations: if you track selected questions, keep the prompts, models, settings, dates, answers, and citations visible. Treat the results as observations, not market-wide share of voice.
On-site outcomes: check what visitors do after they arrive.
Review the page after enough time has passed for the data to become useful. For a lower-traffic page, a few clicks can be noise, so look for a combination of query relevance, impressions, referrals, engagement, and qualitative evidence.
My preferred closing question is simple: What did we learn, what did we change, and what will we check next?
A 10-minute AEO audit checklist
Use this as triage for one important page. It is not a replacement for a full technical or content audit.
☐ Write down the main audience question in one sentence.
☐ Check that the page answers it near the top.
☐ Identify one piece of original expertise, evidence, or practical detail.
☐ Replace vague headings and fix any broken heading hierarchy.
☐ Confirm the author, owner, review date, and important sources.
☐ Check names, product facts, and definitions against authoritative sources.
☐ Inspect the published page for indexability, canonical, rendered content, and broken links.
☐ Confirm that crawler access rules reflect an intentional policy.
☐ Confirm structured data matches visible content.
☐ Add or improve relevant internal links.
☐ Record the change and choose the measurement you will review.
If several boxes fail, fix the highest-impact issue first. One thoroughly improved page is more useful than twenty pages that received a rushed “AEO pass”.
Frequently asked questions
What is an AEO content audit?
An AEO content audit checks whether a page answers a real question, provides useful and verifiable information, meets technical search requirements, and can be measured across search and AI experiences. GEO is another common label for this area. The audit extends a normal SEO and content review rather than replacing it.
How is an AEO audit different from an SEO audit?
AEO places extra attention on how content may appear in generated answers, but the foundation remains SEO. Crawlability, indexability, helpful content, clear structure, authority, and accurate information still matter. Google describes optimisation for its generative AI features as part of optimising for search.
Do I need an AEO plugin for Umbraco?
You do not need to begin with a dedicated AEO plugin. Start by checking the published page and the fields, blocks, templates, and technical controls already available in your implementation. Additional tools may help with auditing, observation, or workflow, but a tool does not replace useful content and sound implementation.
Does schema make a page appear in AI answers?
Schema does not guarantee inclusion in an AI answer. For Google, structured data is not required for generative AI search and there is no special AEO schema. Continue using accurate structured data where it supports your wider SEO strategy and eligible search features.
How long does an AEO content audit take?
The checklist in this guide gives you a 10-minute first pass on one page. A full review takes longer when it includes query research, source verification, template inspection, technical testing, internal-link analysis, or measurement setup.
Can an audit guarantee that an AI system cites my page?
No. Search and AI systems decide what to retrieve and present, and those decisions can vary by query, context, product, and time. An audit improves the parts you control; it does not guarantee crawling, indexing, visibility, traffic, or citations.
Summary: what AEO readiness means in practice
Start with a real audience question and answer it clearly.
Add original expertise, examples, and evidence instead of another generic summary.
Use structure to help readers, not to chase an imaginary ideal chunk size.
Make ownership, sources, and review dates visible where they matter.
Inspect the published page for crawlability, indexability, canonicals, rendering, and working links.
Review crawler controls as a deliberate policy, not a generic “allow AI” switch.
Use Umbraco content models to support consistent work, not as a substitute for it.
Treat schema as supporting SEO, not as a special AI signal.
Measure different outcomes separately and keep improving the page.
The practical goal is not to make a CMS “AEO compliant”. It is to help your team publish useful, technically sound, well-supported content and maintain it after the first audit is over.