DIY AEO for 30 Days — Hire When You’re Blind to Your Own Gaps

Yes — you can do Answer Engine Optimization yourself for the first 30 days if someone on the team will own a frozen prompt panel and ship truth-layer fixes. Hire help when you are blind to your own contradictions, need an outside prioritization hammer for stakeholders, or the panel stops moving after you have done the obvious work.

This is not a pitch to DIY forever or to outsource on day one. It is a scope split under the Answer Engine Optimization playbook. The cousin spend question — whether AEO is even worth a budget — lives on is AEO worth it. This page is the operating gate: run the month, then decide.

The short answer

• DIY works for baselines, fact reconciliation, answer-first rewrites, and a weekly measurement ritual.
• DIY fails when identity conflicts are invisible from inside, when legal and comms cannot align, or when “we posted more blogs” is the only tactic.
• A hired audit should return scores, a ranked backlog, and a 30/60/90 — not a slide deck of new acronyms.
• Keep execution in-house when you have an SEO or content owner with hours. Hire execution when that owner does not exist.
• Stop DIY and get an outside pass when two re-baselines show no citation or accuracy movement after you shipped the checklist.

Can you do AEO yourself?

Yes, for a defined slice. The first month is a measurement-and-truth program, not a content factory.

Google’s site-owner guidance is the floor, not a secret AEO trick: to appear as a supporting link in AI Overviews or AI Mode, a page must already be indexed and snippet-eligible. There is no extra technical bar (Google Search Central: AI features). If ordinary Search eligibility is broken, DIY AEO is theater.

OpenAI splits the same idea across two bots. OAI-SearchBot surfaces sites in ChatGPT search answers; opting out removes you from those answers. GPTBot is the training crawler. The settings are independent, and robots.txt changes can take about 24 hours to apply for search (OpenAI crawler docs).

Minimum DIY kit:

Piece	Pass	Fail
Named owner	One person on the calendar	“Marketing will handle it”
Prompt panel	25–40 frozen IDs	A new list every week
Log	Date, engine, citation, accuracy	Screenshots in a Slack graveyard
Access	CMS, Search Console, analytics	“We can get that next quarter”
Authority	Can fix About, schema, conflicting claims	Legal parks every edit

If you lack the owner or the access, you do not have a DIY program. You have a wish.

• One name on the 30-day calendar
• CMS + Search Console login in that person’s hands
• Authority to change About, offer copy, and published facts
• A competitor set of 3–5 names you will not keep adding
• A written stop rule for when the month ends

Three empty boxes → do not start a “DIY AEO sprint.” Fix ownership first, or hire the audit and admit you are buying a scoreboard.

What a 30-day DIY panel actually is

A panel is a frozen list of prompts you will re-run, not a brainstorm. If the list moves, the delta is fake.

Build it from revenue language, not from keywords you wish buyers typed.

Bucket	Example shape	Why it belongs
Recommendation	“best [category] for [job]”	Where shortlists get written
Comparison	“[you] vs [competitor]”	Where contradictions get quoted
How-to	“how to [job your offer finishes]”	Where citeable passages live
Brand	“who is [brand] / what does [brand] do”	Where hallucinations show up first
Local (if real)	“[category] in [city]”	Where NAP drift kills you

Rules that keep the panel honest:

1. Freeze IDs on day 1. Add a parking lot for new prompts. Do not swap IDs mid-month.
2. Tag each prompt revenue or vanity. Vanity can sit in the log. It cannot drive the hire decision.
3. Run the same wording. “Best automation agency” and “best AI automation studio” are two IDs.
4. Record engine, date, cited URL, mention-only, accuracy note, and who else got the citation.
5. Screenshot the answer once per run. Memory is not a source.

I have been SEO-certified since 2021 and I have shipped hundreds of production sites. The teams that get this right treat the panel like a test suite. The teams that fail treat it like a content calendar.

What you freeze before day one

Do not open ChatGPT until the fact packet exists. Models quote whatever public page is easiest to lift. If your About, footer, and a 2023 press mention disagree, DIY month one becomes an argument about which lie is official.

Fact	Source of truth	Where it must match
Legal / brand name	Articles or brand guide	Home, About, schema name
What you sell	Offer page, not a tagline	First screen + FAQ
Who it is for	ICP sentence you would say out loud	About + service pages
Where you operate	One NAP string	Site, directories, GBP if local
Founded / size claims	Only numbers you can defend	About, bios, schema
Same-as profiles	Live URLs	Organization sameAs

Google’s Organization markup exists to disambiguate you, not to “rank you in AI.” Place it on the home page or a single About page. There are no required properties; add what is true (Google Organization structured data; schema.org Organization).

• One fact sheet in a doc the owner can edit
• Conflicting public claims listed, not “we’ll remember”
• Legal or founder sign-off on the three sentences AI already mangles
• A decision on training vs search crawlers, written down
• Search Console property verified so Google’s own AI-feature report is even an option

Search Console now has a Search generative AI control for inclusion in AI Overviews, AI Mode, and related Discover features. Default is include. Exclude is a site-owner choice, not a ranking boost, and it does not replace noindex or Google-Extended (Search Console: Search generative AI control). DIY teams skip this screen, then argue about “why we are not in Overviews” while the property is excluded or unverified.

The 30-day DIY plan

Skip the vanity of twenty new posts in month one. Truth layer and extractable answers beat volume. Google’s own AI-optimization note is still ordinary Search quality: crawlable text, internal links, structured data that matches the visible page (Google AI optimization guide).

Days	Outcome	Proof you keep
1–3	Panel frozen + competitor set	Sheet with prompt IDs and owners
4–7	Baseline on ChatGPT and sample AI Overviews	Screenshots + citation / mention / accuracy
8–14	Fact sheet + schema / llms.txt / About fixes	Diffs or published URLs
15–22	Five money pages rewritten answer-first	Before / after lead paragraphs
23–27	Corroboration outreach list (10 targets)	Sent vs pending
28–30	Re-baseline + go / no-go on hiring	Same panel IDs, delta noted

Day-by-day is less important than the gates:

1. No baseline → you cannot judge day 30.
2. No published fact fixes → the re-baseline is noise.
3. No answer-first money pages → you measured a site that still hides the offer.
4. No written go / no-go → the month becomes “we should keep posting.”

llms.txt is a community Markdown map at /llms.txt, not a Google ranking file (llms.txt spec). Chrome Lighthouse will check for it in agentic-browsing audits; a missing file is currently N/A, not a fail (Chrome Lighthouse: llms.txt). Ahrefs’ read is the honest DIY hedge: treat it as cheap hygiene, not a citation switch (Ahrefs on llms.txt). Ship it after the fact sheet is true. Do not spend week two decorating a file that points at contradictory pages.

How to log a baseline without buying a dashboard

Buy tools after the ritual exists. Logos do not create citations.

A working log is a table you can sort. Columns that matter:

Column	What you write	Trap
prompt_id	Stable ID, never recycled	Renaming mid-month
prompt_text	Exact string	“Close enough” paraphrases
engine	ChatGPT, AI Overview, other	Mixing engines in one row
run_date	ISO date	“Sometime in June”
cited	Yes / no + URL	“They kind of mentioned us”
mentioned	Yes / no	Counting a competitor’s mention as yours
accuracy	Correct / wrong / mixed + the error	Skipping the quote
who_else	Cited domains	Empty because it felt rude
ticket	The fix you opened	None, so the fail dies

Run cadence for a single-brand DIY month:

1. Week 1: full panel, both primary engines you care about.
2. Weeks 2–3: only the fails plus any prompt that hit a published fix.
3. Week 4: full panel again. Same IDs. Same engines.

Pew’s July 2025 analysis of March 2025 browsing is why this log cannot be “did we get the click.” On visits with an AI summary, users clicked a traditional result on 8% of visits versus 15% without one, and clicks on links inside the summary happened on 1% of those visits (Pew Research Center). Google’s own AI-features page still reports those appearances inside Search Console’s web performance data and says clicks from Overview-bearing pages can be higher quality when they happen (Google AI features). DIY takeaway: log inclusion and accuracy first. Do not wait for a traffic cliff to reverse.

• Week-1 sheet exists before anyone rewrites a page
• Every wrong brand fact is a ticket, not a vibe
• Competitor citation URLs are saved so you know what to displace
• Search Console generative-AI report is opened at least once, if the control is available
• No vendor dashboard is required to call the month done

Truth-layer work a marketer can ship

You do not need a developer to start measuring. You need honesty about what AI already says.

Marketer-owned in the first 30 days:

• Prompt panel and weekly log
• About / offer copy that states who, what, and for whom in the first screen
• FAQ copy that matches questions people already ask
• Internal links across a small cluster
• Directory and Google Business Profile consistency, if you are local
• Outreach list for corroboration (roundups, associations, partner pages)
• CMS fields for author, dates, and canonicals that do not need code
• A one-page fact sheet legal can approve in a sitting

Answer-first is a writing job. Lead with the sentence a model would quote. Then the proof. Then the offer. If the first 80 words are atmosphere, DIY month one will “ship five pages” and change nothing extractable.

FAQ copy is still worth writing even when Google’s FAQ rich result is not the prize. FAQPage remains a schema.org type for classifying Q-and-A content (schema.org FAQPage). Write the answers for humans and for quote-ready extraction. Do not build a DIY program around a rich-result screenshot.

What marketing should not pretend to ship alone:

Task	Why it stalls without eng
Organization / LocalBusiness JSON-LD in the template	Theme or framework ownership
llms.txt on the edge or host	Deploy pipeline
robots.txt bot allow / deny matrix	One typo can hide you from search crawlers
JS-only money copy with no HTML text	Engines may miss the passage
Multi-host canonicals	Easy to publish two truths

Marketing writes the fact sheet. Engineering ships the machine-readable layer. Mixing those jobs is how silent drift starts.

When DIY fails

Watch for these failure modes. Any two is a hire signal. I have seen the same pattern across 500+ automations and the website book of work: unowned ritual looks cheap and costs a quarter.

Failure	What it looks like	Cost if ignored
Insider blindness	You swear About is clear; AI still invents a sibling brand	Wrong recommendations for a quarter
Ritual collapse	Panel skipped three weeks in a row	No idea what changed
Blog reflex	“Ship 12 posts” with no quote-ready answers	Crawl bloat, zero citation lift
Stakeholder fog	SEO, brand, and legal disagree on facts	Schema and PR blocked
Tool cosplay	Dashboard logos without a human log	Budget spent, no decisions
Bot panic	Blanket-block “all AI crawlers” after a webinar	You paid for AEO and opted out of the surface

DIY is not free. Unowned DIY is the most expensive option.

Insider blindness is the one founders underestimate. You cannot see the contradiction because you know which page is “the real one.” The model does not. It will quote the press mention that still lists the old product name. An outside reader — or a hired audit — is the fix, not another internal workshop.

Bot panic is the other expensive miss. Google’s robots.txt file is a crawl-traffic control, not a hide-from-Search switch. A disallowed URL can still be indexed without a snippet if other pages link to it (Google robots.txt intro). Write a training-versus-search policy. Do not “block the bots” as a personality.

When to stop DIY and hire

Stop self-serve and bring in an outside audit when any of these gates trip. Hiring earlier is fine if time is the scarce resource. Hiring instead of running a single baseline is usually premature.

• Two re-baselines (about 30 days apart) show no meaningful citation or accuracy movement after you shipped the day 8–22 work
• You cannot get a single ranked backlog agreed in a 90-minute meeting
• Brand facts conflict across site, directories, and PR, and no internal owner can force reconciliation
• Multi-location or multi-product complexity explodes the entity matrix
• Leadership wants an external receipt before funding the next quarter
• You suspect competitor citation dominance and need a gap map you will not bias
• The panel owner disappeared and nobody else will take the log

The hire is for sight and force, not for “someone else to do AEO” with no owner.

Gate	What you buy	What you still own
Blind to contradictions	Entity / fact conflict list + source of truth	Publishing the corrections
Panel stalled after real fixes	Gap map + ranked 30/60/90	Weekly log after the audit
Stakeholder fog	External backlog leadership will fund	A named operator
Complexity (locations, products)	Entity matrix you will not finish alone	Local / product owners
No operator	Audit plus a staffing recommendation	The decision to hire or stop

If you already know the top five fixes and refuse to ship them, do not hire. An audit will reprint your avoidance in better typography.

What a hired audit should include

Demand this deliverable set. If the proposal cannot map to it, you are buying SEO theater with an AEO sticker.

1. Scope: ICP, markets, competitor set, prompt panel
2. Baseline tables: citation rate, mention rate, share of voice, accuracy issues per engine
3. Entity / fact conflict list with a recommended source of truth
4. On-site truth-layer review (llms.txt, schema, About, offer pages)
5. Content citeability sample — not a full rewrite of the blog
6. Citation-gap / corroboration targets
7. Technical fetchability notes (bots, indexation, JS traps)
8. Section scores + prioritized 30/60/90 with explicit non-goals

Deliverable	Acceptable proof	Theater
Panel	Redacted IDs + scoring rules	“We studied your keywords”
Baseline	Tables by engine and prompt bucket	A heat-map screenshot
Conflicts	URL vs URL with the winning fact	“Tighten your messaging”
30/60/90	Ranked tickets + non-goals	A 40-slide journey
Fetchability	robots.txt / render notes	“We’ll optimize for AI”

A first month that only produces a slide deck failed. A first month that produces a comparable week-1 vs week-4 log succeeded even if the citation rate is still ugly. Ugly and measured beats pretty and invented.

Google does not sell a citation slot. ChatGPT search does not have a submission portal. If a vendor guarantees “page-one AI Overviews in 30 days,” treat that as marketing. Google’s public AI Overviews explainer is about how the feature works for users — more than 1.5 billion people using Overviews, grounded in Search systems and the Knowledge Graph — not a placement product you can buy (Google: AI Overviews and AI Mode).

How DIY and agency scopes split

Keep the ritual. Rent the sight. That is the whole split.

Workstream	Keep DIY / in-house	Hire audit	Hire ongoing
Prompt panel ownership	Yes if hours exist	Design + first run	Optional retain
Fact reconciliation	Yes	Escalate conflicts	Rarely
Schema / llms.txt	Yes with light eng	Spec + review	Only if no eng
Answer-first rewrites	Yes for top pages	Sample + brief templates	When capacity is missing
Digital PR / roundups	Relationship-heavy — often in-house	Target list	Specialized PR partner
Weekly logging	Must stay owned	Train the ritual	Only if you refuse ownership
Competitive citation maps	Possible DIY	Faster externally	Quarterly

Rule: never outsource the ritual without a named internal owner. Audits that create orphan dashboards die in month two.

I will not invent a retainer price to make any split look smart. Public “AEO retainer” bands you see on agency blogs are market hearsay, not Spurlock Studios pricing and not a quality signal. Price the hours you will still spend internally, then the invoice. If nobody will own the log after the PDF lands, you bought education. Call it education.

Ongoing hire is rational when:

• The operator role does not exist and will not exist
• Multi-location entity work exceeds one marketer’s week
• You want a quarterly gap map you will not bias
• Legal needs an outside writer to force a fact freeze

Ongoing hire is waste when the only missing piece was courage to edit About.

When you need a developer for schema and crawlers

Call eng when the fact sheet is decided and the machine-readable layer is blocked. Do not call eng to “figure out our AEO strategy.” That is the owner’s job.

Task	Why marketing alone stalls	What to hand eng
Organization / LocalBusiness / FAQ JSON-LD in the template	Theme or framework ownership	Fact sheet + property list
llms.txt on the edge / host	Deploy pipeline	Final Markdown, cache rules
Robots / bot allow rules	Misconfig can block OAI-SearchBot or Googlebot	Written training vs search matrix
JS-rendered money copy with no SSR	Engines may miss the passage	Which passages must be HTML text
Multi-domain canonical / subdomain docs	Easy to publish two truths	Winning host + redirect list
WAF / bot-fight rules	Security defaults block the crawlers you want	Allowlist from vendor docs

OpenAI’s crawler page is the spec you hand security, not a Twitter screenshot. Allow OAI-SearchBot if ChatGPT search visibility is a goal; treat GPTBot as a separate training decision (OpenAI crawler docs). Googlebot remains the control for Search AI features; Google-Extended is the training / grounding lever for some other Google systems (Google AI features).

• Fact sheet signed before anyone opens the theme
• robots.txt change has an owner and a rollback
• JSON-LD matches visible text — Google’s own AI-features list calls this out (Google AI features)
• Staging checked with URL Inspection, not only “it looks fine in the CMS”
• A second human reads the bot matrix before production

How much time per week DIY takes

Honest ranges for a single-brand program after setup. These are operating estimates from running this ritual, not a vendor SLA.

Role task	Hours / week	What “done” means
Panel runs + logging	1–2	Rows exist; screenshots filed
Ticket grooming from fails	0.5–1	Each wrong fact has an owner
Content / truth-layer shipping	2–6 (bursty)	Pages published, not outlined
Stakeholder sync	0.5	One ranked list, not a debate

Setup week is heavier — one to three focused days to freeze the panel, write the fact sheet, and take the first baseline.

If you cannot find about four hours most weeks, DIY will stall. Hire for the audit and decide whether to buy execution or free an owner.

Capacity math that keeps founders honest:

1. Write the four-hour block on a real calendar for four weeks.
2. If week two gets eaten, you do not “catch up in week four.” You slipped.
3. Two slips → ritual collapse. That is a hire-or-stop decision, not a motivational poster.
4. Do not staff the panel with an intern who cannot change About.

A founder who will not give the hours should not announce a DIY AEO program in the Slack channel. Announce a hire, or announce that AI visibility is not this quarter’s job.

Tools: optional vs required

Required tools are access and a log. Everything else is optional until the ritual exists.

Required	Optional	Buy later if
Spreadsheet or Notion for the panel	Semrush / similar for SERP context	You need classic competitive maps
Browser access to ChatGPT and Google	On-page coverage aids	After money pages have answer-first leads
CMS + analytics access	“AI visibility” dashboards	You already log by hand for 30 days
Search Console	Agency-only platforms	An audit says the gap is fetchability
Written robots.txt / bot policy	Extra crawler-log products	Security will not take a Markdown matrix

Search Console is required because Google’s AI features report into the same Search property you should already have (Google AI features). It will not log ChatGPT citations. That is why the human panel exists.

• Log lives where the owner already works
• No tool login is a blocker for week-1 baseline
• Optional suite is not in the month-one budget unless you already pay for it
• Nobody is waiting on a procurement cycle to screenshot an answer

Buy the dashboard after two comparable baselines. If you cannot produce the first baseline without software, the software is not the problem.

How to evaluate an AEO agency that is just reselling SEO

Interview for a panel, not a rebrand of last year’s SEO deck.

Questions that surface theater:

1. Show last month’s prompt panel for a client (redacted) — not only rank charts.
2. What are your AEO KPIs in one sentence?
3. Walk a citation-gap finding to a ticket.
4. What do you deliberately not do in v1?
5. Who owns the weekly log after you leave?
6. How do you treat OAI-SearchBot vs GPTBot vs Googlebot?
7. What does a failed 30-day DIY look like, and would you have told us to wait?

Signal	Keep talking	Walk
KPIs	Citation, mention, accuracy, SOV	“Rank #1 in ChatGPT”
Panel	Frozen IDs, scoring rules	Keyword list with “AI” in the tab name
Bots	Training vs search, named agents	“We unblock all AI”
Non-goals	Written	“We do full-funnel AEO”
Aftercare	Named internal owner	Orphan login

Walk away if the deck only renames SEO deliverables, cannot describe ChatGPT search versus AI Overviews, or promises guaranteed citations on a calendar.

Google’s helpful-content and technical basics still apply to AI features. There is no special markup that buys an Overview chip (Google AI optimization guide). An agency that leads with a proprietary schema “for AEO” is selling you a file. Ask for the panel.

I collaborate with the n8n team and I have spent 20,000+ hours on agentic systems. The tell I trust in an AEO pitch is the same tell I trust in an automation pitch: can they show the test set, the failure, and the ticket? If not, you are buying adjectives.

When is a visibility audit the right hire?

Choose an audit when you need an outside pass that forces prioritization — not when you want someone else to “do AEO” indefinitely without an owner.

Good audit fits:

• Pre-budget justification after a real DIY month
• Post-rebrand or rename, when public facts still disagree
• Suspected hallucination or competitor citation dominance
• DIY day-30 go / no-go tripped a hire gate
• Leadership will not rank a backlog unless a third party writes it

Bad audit fits:

• No access granted
• No one will own the weekly panel afterward
• You already know the top five fixes and refuse to ship them
• You skipped the baseline because a webinar said to “get ahead of AI”
• You want a guarantee Google and OpenAI do not sell

Spurlock Studios’ visibility lane exists for the first list. The second list needs an internal decision, not a PDF.

If the only question you still have is “is any of this worth paying for,” read is AEO worth it before you book a call. That page is the spend gate. This page is the DIY-versus-hire gate. Do not mix them. A cheap month of logging can save you a retainer you were not ready to operate.

Day-30 decision table:

What the month showed	Next step
Citations or accuracy moved after truth-layer work	Keep DIY; widen the cluster
Panel ran; facts still conflict and nobody can force a winner	Hire the audit
Panel stalled; hours were never real	Hire execution or stop. Do not “try harder” in public
Baseline never happened	You did not DIY. Start, or hire for the scoreboard
Competitors own the recommendation prompts and you cannot see why	Hire the gap map

Run the 30 days. Hire when you are blind or stuck. That is the whole offer.

FAQ

What can a marketer ship without engineering?

Baselines, fact sheets, About and offer clarity, FAQ copy, internal links, directory consistency, and outreach lists. You can freeze a prompt panel and log citation versus mention without a deploy. Schema templates in the theme, robots rules, llms.txt on the host, and hard render issues need engineering once the facts are decided. Do not wait on eng to take the week-1 screenshots.

When do I need a developer for schema and crawlers?

When JSON-LD lives in the theme, llms.txt needs a deploy, bot rules are wrong, money copy is client-rendered only, or multiple hosts need one canonical truth. Marketers specify the fact sheet and the training-versus-search matrix. Developers ship the file. Call eng after the facts are frozen, not to invent the strategy.

How much time per week does DIY take?

After setup, plan roughly four hours most weeks for logging, grooming, and shipping — more during rewrite sprints. Setup week is one to three focused days. Below four hours most weeks, expect ritual collapse. That is a hire-or-stop signal, not a reason to buy another dashboard.

What tools are optional vs required?

Required: a panel log, the answer engines themselves, CMS, analytics, and Search Console. Optional: Semrush-class SEO suites, on-page helpers, and AI visibility dashboards. Ritual before logos. If you cannot baseline without software, ownership is the gap.

When is a visibility audit the right hire?

When two re-baselines stall after real fixes, stakeholders need an external backlog, entity conflicts are stuck, or complexity outgrows a solo operator. Also when you are blind to contradictions the model keeps quoting. Not when you refuse to run a single baseline, and not as a forever outsource with no internal owner.

How do I evaluate an AEO agency that is just reselling SEO?

Demand a real prompt panel, AEO KPIs in one sentence, a citation-gap example that became a ticket, explicit non-goals, a named post-engagement owner, and a coherent story on OAI-SearchBot versus GPTBot versus Googlebot. Rank-only decks with “AEO” in the title are a no. Guaranteed Overview placements are a no.

CTA

Run the 30-day DIY. Hire when the gates trip — not because a webinar scared you.

Lane overview: /visibility. When you need the outside pass: visibility audit.