Answer engine optimization checklist for small sites

Answer engine optimization (AEO) means writing pages so that search assistants and AI tools can pull a correct, quotable answer from them. This checklist has 12 steps a small site can start in a weekend, and none of them can guarantee a citation.

Fig.Measure the AI sources before and after you change anything. Sample data, not a real site.

Work through it in order. The first group changes your pages, the second your brand footprint, the third your measurement.

Group 1: page rules

1. Answer in the first two sentences. Put the direct answer to the page's question at the top, then explain. If the title is a question, the lede is its answer.
2. Use question-style headings where natural. "How much does it cost?" is easier to match to a real query than "Pricing overview".
3. Add a table of checkable facts. Price, limits, supported platforms, last updated date. One small table beats three paragraphs of adjectives.
4. Add a short FAQ with 3 to 5 real questions. Use the questions people actually type, and answer each in two to four sentences. If you mark them up as FAQ structured data, make sure the visible page matches it.
5. Show dates. Say when a page was published and updated, and keep numbers current. Stale prices can lead to wrong answers.
6. Include one honest limitation. "This doesn't apply if..." makes a page more trustworthy to a reader and is hard to fake.

Group 2: brand footprint

1. Write a single facts paragraph. What the product is, who it is for, price, how setup works, and what it is not. Copy it word for word onto every profile and directory listing.
2. Check for conflicting facts. Search your own name and product. If an old price or a wrong description appears on a third-party page, ask for a correction, politely and with the right fact.
3. Be present where people compare. Directories, review sites with a free tier, community answers where you disclose who you are. Assistants may repeat what third parties say, so off-site mentions can matter alongside your own copy. We cover this in generative engine optimization basics for small sites.

Group 3: access and measurement

1. Check your robots.txt. Make sure it does not block pages you want read. Decide deliberately about AI crawlers, and check each vendor's current documentation for crawler names, since they change.
2. Skip the file hunt. llms.txt is optional, and whether any given vendor reads it is something to check in that vendor's current documentation. See llms.txt: what it is and whether it matters.
3. Measure weekly for 15 minutes. Details below.

Example (made-up numbers for illustration)

Here is an invented before-and-after of a weekly log for a fictional bookkeeping tool over eight weeks. The numbers are made up.

What you can conclude from a table like this: mentions are moving up slowly. What you cannot conclude: that any single step caused it. Answers vary from run to run, and the other assistants change on their own schedule. Keep notes on what you changed each week so you can at least look for patterns.

Fig.AI assistants as their own channel in the source table. Sample data, not a real site.

What does the weekly check involve?

1. Choose 10 to 20 prompts. Mix category questions ("best one-time analytics"), comparison queries ("X alternative"), and brand queries ("what is X").
2. Run them in ChatGPT, Perplexity, Claude, Gemini and Google's AI results, in a private session so your history does not skew results.
3. Log: were you mentioned, were you linked, which competitors appeared, which sources were cited.
4. Pick at most three actions for the week.
5. Look at AI referrals in your analytics. See how to track ChatGPT referrals, the ChatGPT traffic guide and the Perplexity guide. Ownlytics groups ChatGPT, Claude, Perplexity, Gemini and Copilot visits into an AI channel; the AI traffic analytics page describes it. Google's AI Overviews and AI Mode send visitors from google.com, so they stay in Organic Search and are not separated out.

Give brand mentions and page changes weeks to months before judging them. We have no measured timeline to offer.

What goes wrong most often?

• Doing everything at once. You will not know what worked.
• Writing for the bot. If a sentence reads oddly to a person, cut it.
• Overclaiming. Do not state facts you cannot verify, and do not invent comparisons. Wrong facts get repeated.
• Ranking yourself first on a "best tools" page you wrote about your own category. Readers notice, and assistants may too.
• Expecting traffic. Answers are often read without a click, so mentions can rise while visits stay flat.

When is this not worth doing?

If you have no pages worth citing yet, build those first. If your buyers come from direct outreach or a small community, the weekly check is optional. AEO is a long game that pays back through being clear, which is already worth doing.

At a glance

Times below are our rough estimates, not benchmarks.

If you want to see how AI visits and revenue appear together on a sample site, there is a live demo.

FAQ

What is the difference between AEO and GEO?

The terms overlap and people use them loosely. AEO usually refers to being the source for direct answers, and GEO to being included in generated responses. The steps are mostly the same.

Can I guarantee my site will be cited?

No. Nobody can. Assistants choose sources in ways that are not public and change over time. The steps here improve clarity, not certainty.

Do I need structured data for this?

It is not required. FAQ markup can help machines read your questions, but only if it matches the visible text. Clear writing comes first.

How often should I update pages?

Update when a fact changes, and review key pages every few months. A visible "last updated" date helps readers and avoids stale answers.

Keep reading

• Do AI-referred visitors convert better? How to check
• Generative engine optimization basics for small sites
• How to track ChatGPT referrals in your analytics