AEO Specialist Playbook: The Weekly Operating Routine (2026)

An AEO specialist spends most of the week on four jobs: keeping AI crawlers unblocked, turning buyer questions into pages that answer them, keeping the business's facts consistent everywhere, and measuring where those pages get cited.

None of it is magic, and Google says so plainly in its guide to optimising for generative AI features: AEO and GEO are "still SEO" from Google Search's point of view.

That is good news for operators, because it means the work runs on checklists you can repeat and evidence you can show a client.

If you would rather have my team walk your site through this routine, book a free SEO strategy session and bring your domain.

The four lanes of the job

Every task an answer engine specialist does fits into one of four lanes.

Access: robots.txt, CDN and firewall rules, indexing and preview controls.
Answers: question research, answer pages, page structure and freshness.
Entity facts: business name, address, services, structured data and third-party profiles telling the same story.
Measurement: Search Console, Bing Webmaster Tools and referral analytics, reviewed on a fixed day.

When a week feels busy but nothing moves, it is usually because one lane has been ignored.

Most often it is access, because nobody looks at a firewall rule until a crawler gets blocked by it.

Monday: the access check

I start every week here, because one bad deploy can quietly shut out a crawler.

OpenAI's crawler overview says sites opted out of OAI-SearchBot will not be shown in ChatGPT search answers, and that robots.txt changes can take around 24 hours to register.

Perplexity's crawler docs recommend allowing PerplexityBot and its published IP ranges, and they include firewall setup notes for providers such as Cloudflare.

Anthropic's crawler page warns that disabling Claude-SearchBot may reduce visibility in Claude's search results.

Here is the baseline I want to see for search crawlers:

User-agent: Googlebot
Allow: /

User-agent: Bingbot
Allow: /

User-agent: OAI-SearchBot
Allow: /

User-agent: PerplexityBot
Allow: /

User-agent: Claude-SearchBot
Allow: /

Training crawlers are a separate business decision.

OpenAI says GPTBot and OAI-SearchBot are independent, so a client can block GPTBot and still be eligible for ChatGPT search.

The same split exists for ClaudeBot versus Claude-SearchBot.

Google-Extended works the same way, and Google states it "does not impact a site's inclusion in Google Search nor is it used as a ranking signal in Google Search", per its common crawlers page.

Monday, part two: preview controls that silently remove you

Robots.txt is only half the access lane.

Google's robots meta tag documentation says nosnippet applies to AI Overviews and AI Mode too.

So a template that adds this tag to service pages kills those pages for AI answers:

<meta name="robots" content="nosnippet">

Bing has its own trap, listed in its Webmaster Guidelines: NOARCHIVE prevents content from being used in Copilot responses, and NOCACHE limits Copilot to the URL, title and snippet.

My Monday list for every priority URL:

Returns a 200 status and is indexed in Search Console.

2. No noindex, nosnippet or tight max-snippet value unless the client chose it.

3. No data-nosnippet wrapped around the actual answer paragraph.

4. No NOARCHIVE or NOCACHE meta on pages meant to be cited in Copilot.

5. CDN or firewall logs show the search crawlers getting 200s, not challenges.

📞 Want your access lane checked this week? Book a free SEO strategy session and we will look at your crawler rules, preview tags and priority pages together on the call. → Claim your free strategy session

Tuesday: building the question map

Answer engines do not answer one query at a time.

Google describes "query fan-out", where the model fires off a set of related searches to gather more results for one question.

For a client, that means the page has to cover the follow-ups a real buyer would ask, not just the headline query.

My question map has five columns:

The buyer's question in their own words, pulled from sales calls, support tickets and reviews.
The stage: researching, comparing or ready to buy.
The page that should answer it, or "none yet".
The proof that page can offer: a number, a process, a photo, a named method.
The date the answer was last checked.

Be careful with volume here.

Google's guide warns that building separate pages for every fan-out variation, mainly to manipulate results, breaks its scaled content abuse policy.

One strong page per real buyer problem beats twenty thin ones.

Wednesday: shipping one answer page

I aim for one finished answer page a week rather than five drafts.

The format that works for most service businesses:

The direct answer in the first two sentences, in plain words.

2. The conditions: when the answer changes, and why.

3. First-hand detail only this business has, such as its process, local knowledge or real job examples.

4. A short comparison or checklist if buyers weigh options.

5. A clear next step for the reader.

Google calls the goal "non-commodity content", meaning content that goes beyond common knowledge with an expert or first-hand view.

That is where the specialist earns their keep, because interviewing the owner for those details takes skill that a generic writer rarely has.

Thursday: keeping the entity facts straight

AI answers mention businesses whose facts line up across the web.

On the site, that means one consistent business name, address and service list, and markup that matches the visible text.

Google's Organization structured data guide says markup helps Google tell organisations apart and supports sameAs links to official profiles.

The schema types I reach for most:

Organization on the home or about page, with logo, url and sameAs.
LocalBusiness or a more specific subtype for businesses with a physical location.
Person for the named expert behind the content, where it is accurate.

Google says there is no special schema.org markup needed for AI features, so I treat schema as hygiene, not a lever.

Bing adds that markup must accurately reflect visible content, or it may be ignored.

Off the site, the Thursday check covers the Business Profile, key directories and, for US local businesses, OpenAI's Business Hub, which lets an authorised owner claim and manage business details in ChatGPT.

OpenAI is clear that claiming doesn't guarantee more visibility.

🔥 Want to run this routine with your own agents? Inside the AI Profit Boardroom you get step-by-step video trainings, the installable Agent OS and four live coaching calls a week, so you can learn AI SEO yourself and hand the busywork to agents. → Join the AI Profit Boardroom

Friday: measuring citations, not vibes

Friday is for numbers, logged in the same sheet every week.

Google's Search Console generative AI performance reports, rolled out to all sites on 31 August 2026, show impressions, pages and countries for AI Overviews and AI Mode.
Bing's AI Performance report shows total citations, cited pages and sample grounding queries across Copilot and Bing's AI summaries.
Referral analytics for AI assistants; in our own accounts, ChatGPT visits usually show up with chatgpt.com as the source.

Compare week on week, page by page.

A page that gains impressions but no citations usually has a weak first answer or thin proof.

Where agents save the most hours

The routine above is perfect agent work, as long as a person approves the output.

Access sweep: an agent fetches robots.txt, headers and meta tags for every priority URL and flags changes since last week.
Question mining: an agent clusters call notes and reviews into the question map, with the source line for each.
Schema drafts: an agent drafts JSON-LD from the visible page, and a person checks each value.
Brief writing: an agent turns a row of the question map into a brief with the owner interview questions.
Report assembly: an agent pastes the Friday numbers into the sheet and writes a short change note.

Outreach is the exception: agents can research targets and draft messages, but nothing gets sent without a person reading it.

Skills to look for when you hire one

If you are hiring rather than doing, the routine tells you what skills matter.

Can read robots.txt, response headers and firewall logs without help.
Knows the difference between search crawlers and training crawlers for each platform.
Interviews well, because first-hand detail comes from the business owner.
Writes clear, direct answers and edits AI drafts hard.
Cites official docs from Google, Bing and OpenAI when recommending changes.
Reports with real exports, not screenshots of a tool's private score.

Google's guide also warns that no third-party tool has access to its internal ranking or AI systems, so be wary of anyone selling an "internal" metric.

Specialist or agency: how the workload splits

A solo specialist is a good fit when the site is small, the owner is involved, and the main gap is answers and access.

An agency fits better when the work also needs a developer, regular content, digital PR or link building at the same time.

Goldie Agency describes itself as an SEO, GEO, AEO and link building agency, so all four lanes sit under one roof, and its link building is described as "manual outreach, no PBNs".

Whichever you choose, link building has to stay inside Google's spam policies, which allow paid placements only when the link carries rel="sponsored" or nofollow.

For the same comparison from the generative side, see the GEO specialist breakdown.

A worked example: a local plumbing firm

Here is how the first four weeks run for a two-van plumbing firm.

Week 1: the access sweep finds the CDN challenging unknown bots, so the specialist allows the published IP ranges for OAI-SearchBot and PerplexityBot and rechecks the next day.

2. Week 2: the question map pulls ten questions from call notes, such as "do you fix boilers on weekends" and "how fast can you come out".

3. Week 3: the first answer page goes live for the emergency call-out question, opening with the plain answer and the firm's real response process.

4. Week 4: LocalBusiness markup is fixed so the phone number and hours match the page, and the Business Profile hours are updated to match.

No one can promise this firm a mention in ChatGPT, but every step removes a reason not to cite it.

The gotchas that waste a month

Blocking all AI crawlers in one wildcard rule, then wondering why ChatGPT never cites the site.
Adding an llms.txt file and calling it AEO; Google says Search doesn't use it, and it neither helps nor harms there.
Chunking every page into tiny sections; Google lists chunking as something you can ignore.
Chasing fake mentions; Google says inauthentic mentions are less helpful than they seem.
Reporting a tool's "AI visibility score" instead of exports from the platforms themselves.

Questions to ask before you sign

Google's Do you need an SEO? page now asks whether AEO and GEO advice lines up with its official guidance on generative AI features.

Add these to your interview:

Which crawlers will you allow, and which training bots do you recommend blocking?

2. Which official docs back each recommendation?

3. What will you report each week, and from which tools?

4. Who writes the answers, and how do you get first-hand detail from us?

5. Will you list every change you make to the site?

If anyone guarantees a first-place ranking or a ChatGPT mention, walk away; Google says no one can guarantee a #1 ranking, and OpenAI says placement is not guaranteed.

📞 Want an expert to run the routine for you? Book a free SEO strategy session and my team will map the first four weeks for your site, lane by lane. → Book your free session now

🎁 Two free extras for your routine Grab the 200+ AI SEO Prompts for your briefs and audits, and the free link building book for the outreach side. → Or learn AI SEO yourself inside the AI Profit Boardroom

Also On Our Network

FAQ: AEO Specialist Playbook

What does an AEO specialist check first?

Crawler access. If OAI-SearchBot, PerplexityBot, Claude-SearchBot or Googlebot cannot fetch the page, nothing else in the plan matters.

Does an AEO specialist need special schema for AI answers?

No. Google says there is no special schema.org markup for AI features, so the job is accurate Organization or LocalBusiness markup that matches the visible page.

How often should citation data be reviewed?

Weekly is a sensible rhythm: Search Console's generative AI report, Bing's AI Performance report and your analytics referrals, logged in one sheet.

Can AI agents do the AEO work on their own?

Agents are fast at fetching, comparing and drafting. A person still decides what to publish, signs off schema and owns every outreach message.

Should I give an AEO specialist full Search Console access?

Not at the audit stage. Google's hiring guidance says to grant read access only until you trust the work.

📺 Video notes + links to the tools 👉

🎥 Learn how I make these videos 👉

🆓 Get a FREE AI Course + Community + 1,000 AI Agents 👉

About Julian

I'm Julian Goldie, an SEO entrepreneur, author and founder of the AI Profit Boardroom, which has 3,400+ members.

I help business owners scale with AI agents, automation and SEO.

I've built a 7-figure SEO and link building agency, Goldie Agency, with a team of 70+ SEO experts.
I've grown a YouTube channel to 400,000+ subscribers.
I wrote "Link Building Mastery".
My Udemy courses have taught over 29,000 students.

→ Get my best AI training inside the AI Profit Boardroom

Run the five checks on Monday, ship one answer page a week, log citations on Friday, and if you want a team to run it with you, book a free SEO strategy session with an experienced AEO specialist.