How much of my marketing budget should go to AEO vs traditional SEO

There is no universal percentage of marketing budget that should go to Answer Engine Optimization versus traditional SEO. Split the organic-search program from what you actually get searched for — informational, brand, and commercial — then fund citation work on the queries answer engines already answer, without starving the crawl, technical, and money-page work that still fills the candidate pool. A 70/30 “industry split” on a slide is not a statistic. It is a guess with a serif font.

This spoke sits under the Answer Engine Optimization playbook. Whether to write the check at all lives in when AEO is worth it. The first tickets live in the AEO audit checklist. This page is the mix table, not a fake ROAS.

The short answer

• There is no published, defensible universal AEO vs SEO percentage. Anyone selling one as a market fact is inventing a statistic.
• Classify the last 90 days of Search Console into informational, brand, and commercial. The mix picks the planning band. The band is a starting range, not a survey result.
• Informational-heavy programs tip toward AEO. Brand- and commercial-heavy programs stay SEO-weighted and keep a smaller citation / truth-layer line.
• Keep a technical-SEO floor. Google’s own docs still require indexed, snippet-eligible pages for supporting links in AI Overviews and AI Mode.
• Do not judge the AEO column on restored 2022 informational CTR, and do not invent a return-on-ad-spend number for citations.

If this is most of your non-brand demand	Tip the organic program	Do not
Informational (what-is, how-to, comparison)	Toward AEO: answer-first pages, prompt panel, entity facts	Cut crawl health to “free up AI money”
Brand / navigational	Toward classic SEO: protect brand URLs, NAP, speed	Rewrite the homepage into a glossary
Commercial / transactional	Toward SEO on money pages; AEO on comparison queries	Pause demo/pricing work for blog volume

A fake 70/30 is a slide. A mix table is a method.

Why is there no universal AEO vs SEO percentage?

Because the question hides three different businesses. A glossary-heavy software blog, a brand-navigational artist site, and a local service company do not share a click curve, an Overview rate, or a conversion path. Averaging them into one “right split” is how you get a number you cannot defend on a sales call.

The public studies explain the pressure. They do not give you a budget formula.

Source	What it actually measured	What you may cite	What you may not
Pew Research, July 22, 2025	March 2025 U.S. panel; 900 adults; 68,879 Google searches	Traditional-result clicks on 8% of visits with an AI summary vs 15% without; 1% clicked a link inside the summary	Your category’s CTR, or a budget %
Seer Interactive, April 2026	53-brand cohort; informational slice	Overviews on about 36% of informational queries vs about 8% commercial and 5% transactional in that set	A universal Overview rate, or a universal spend split
Ahrefs, March 2026	863,000 keyword SERPs; 4 million Overview URLs	38% of cited URLs also sat in the first 10 blocks, down from 76% in their July 2025 pass	“Rank is dead” as a reason to zero SEO
SparkToro, 2026	Similarweb clickstream; U.S. Google, Jan–Apr 2026	68.01% of searches ended without a click in that panel	Proof that SEO should be frozen
Google AI features	Product eligibility	Supporting links must be indexed and snippet-eligible; no special Overview markup	Permission to keep a rank-only dashboard

Semrush’s AEO vs SEO writeup is the operator translation: SEO aims to rank in traditional results and Google’s AI-powered SERP features to drive traffic; AEO aims to get cited or mentioned inside generated answers. Shared toolkit. Different acceptance test. That is a job split, not a pie chart of the company.

Google’s product PDF says AI Overviews reach more than 1.5 billion users. That is surface size. It is not “spend 40% of marketing on AEO.” Check your queries. Then move money.

Operator rule: quote a named study next to a click claim. Never quote a named study next to a budget percentage you invented.

Are we splitting total marketing, or the organic program?

The organic-search program. Ads, events, creative, and sales enablement are different envelopes. If you take “40% of marketing” for citations while paid search and the close motion starve, you did not do AEO. You raided the company.

This page splits the money already labeled SEO, content, technical search, and the new AEO line. Paid search is a cousin problem — informational themes compressing is real — but it is not this mix table.

Envelope	This post decides?	What it is for
Organic search / content / technical SEO	Yes	Rank, eligibility, citations, brand SERP
Paid search	No (adjacent)	Brand defense and high-intent clicks
Creative / brand campaigns	No	Demand you already know how to buy
Sales enablement	No	The conversation after the answer
“AEO” as a share of company opex	No	A slide that gets people fired

Checklist before you touch percentages:

• You named the envelope: organic-search program, not “all of marketing”
• You have 90 days of Search Console, not a vendor screenshot of someone else’s niche
• You can name brand vs non-brand without arguing for twenty minutes
• You will not fund AEO by deleting the crawl budget
• You will not sell the split as ROAS

If the envelope is still “the whole marketing budget,” stop. Label the search line first. Then this page has a job.

How do you classify informational vs brand vs commercial mix?

From Search Console, on a frozen query set, not from vibes. Pull the last 90 days. Tag the queries that actually move impressions. The mix is a count you can repeat next quarter, not a personality test.

Procedure:

1. Export queries from the Web performance report for 90 days. Keep clicks, impressions, CTR, and position as separate columns.
2. Take the top 50–80 queries by impressions (not by your favorite blog post).
3. Tag each row: informational, brand, commercial, or local. One primary tag. Mixed queries get a secondary note, not a second percentage.
4. Screenshot 20 of those queries: Overview or AI Mode present / absent; cited domains; your blue-link position.
5. Sum impressions and clicks by tag. Those two shares are your mix. They will not match. That is the point.
6. Write the dated mix on one page: “As of [date], non-brand informational is X% of impressions and Y% of clicks.”

Tag	Typical query shape	What the page is for	Mix implication
Informational	what is, how to, vs, best, checklist	Citation + brand handoff	AEO-weighted if this is the bulk of non-brand demand
Brand	your name, product name, “login,” nav	Protect accuracy and speed	SEO-weighted; small AEO truth-layer line
Commercial	pricing, demo, near me, buy, hire	Rank + convert on a protectable URL	SEO-weighted on the URL; AEO on the comparison cluster
Local	city + service, maps-adjacent	NAP + GBP + service page	Fix Maps before you buy a citation retainer

Seer’s cohort is why the informational row exists: in their informational slice, comparison and question queries showed Overviews at very high rates, while commercial and transactional queries showed them far less often. That is Seer’s classification, not a Google transparency report. Confirm on your 20-query SERP log.

Failure mode: tagging from the content calendar (“we publish how-tos, so we are informational”) while Search Console is mostly brand plus a handful of money terms. You will overfund AEO and under-protect the URLs that pay rent.

Worked mix shapes — same method, different bands. These are illustrations of the tagging step, not client results and not a recommended company-wide percentage.

Shape	What GSC usually shows	Planning-band row	First AEO dollars	Keep in SEO
Glossary-heavy software / studio blog	Non-brand impressions pile up on what-is, how-to, vs	Informational-heavy	Five comparison/definition URLs + panel	Pricing, demo, technical floor
Local services (AllCity HVAC shape)	City + service, brand, a few how-tos	Commercial-heavy or local-first	Light: category prompts that already name competitors	GBP, NAP, service pages
Brand-navigational (artist, merch, tour)	Name queries dominate clicks	Brand-heavy	Hallucination cleanup on facts the model already states	Brand URLs, speed, store/tour pages
B2B with a comparison cluster	Mix of “best X for Y” plus demo/pricing	Balanced or commercial-heavy	Criteria tables on the comparison URLs	Demo, pricing, case pages you can stand behind

If two rows feel true, pick the one that matches impression share, not the one that matches the content calendar. Re-tag next quarter. Do not average the four rows into a fake “typical split.”

What starting ranges should you use for each mix?

Planning bands for one quarter of the organic-search program. Not a survey. Not a claimed industry average. Not Spurlock Studios pricing. Recalibrate 10 points next quarter if citations, brand search, or money-page conversions disagree.

I will not publish a fake “the market spends X% on AEO.” I have been SEO certified since 2021; the honest artifact is a mix table with a date, not a percentage with a press citation it does not have.

Mix (from your 90-day GSC tags)	Classic SEO share of the organic program	AEO share of the organic program	What you are actually buying
Informational-heavy (informational is the majority of non-brand impressions)	30–45%	55–70%	Answer-first rewrites, prompt panel, entity/schema truth layer, corroboration
Balanced (no tag above ~50% of non-brand impressions)	45–55%	45–55%	Same two columns; swing 10 points next quarter from the scoreboard
Commercial-heavy (money queries dominate non-brand)	55–70%	30–45%	Protect pricing/demo/service URLs; AEO on comparison / “best” clusters only
Brand-heavy (brand queries dominate clicks)	65–80%	20–35%	Brand SERP, NAP, speed, hallucination cleanup — not a new blog religion
Foundation on fire (crawl, index, entity facts broken)	80–100% until eligibility passes	0–20% (panel + fact sheet only)	Do not buy citations you cannot enter the candidate pool for

How to read the band:

• The two columns are shares of the organic-search program, and they add to 100% of that envelope.
• “AEO share” includes people time on the prompt panel, citeable page work, entity/schema cleanup, and corroboration. It is not a software SKU.
• If you cannot staff a weekly log, you do not get the high AEO band. You get the worth-it gate: wait, or buy an audit, or stay SEO-weighted.

Do not set the band from	Set it from
A vendor one-pager with a 70/30 pie	Your tagged query mix
Someone else’s Overview rate	Your 20-query SERP log
Informational CTR panic	Clicks vs impressions as two series, plus citations
Agency org chart (SEO team vs “AI team”)	Jobs: eligibility vs inclusion
A fabricated ROAS	Pipeline on protectable URLs

If leadership demands a single number in the first meeting, give them the row matching today’s mix and the date you will re-pick the row. That is a decision. A fake average of Pew and Seer is not.

Worked arithmetic — a planning example, not a measured result. Suppose the last 90 days of non-brand impressions split 62% informational, 23% commercial, 15% other, and brand still carries most of the clicks. That is informational-heavy on reach and brand-heavy on clicks. You do not average those into 50/50. You put the non-brand informational row in the AEO column (start in the 55–70% AEO band of the organic program) and you keep a hard SEO floor on the brand and commercial URLs that actually convert. If you instead funded AEO at 20% because “brand clicks look healthy,” you starved the cluster Overviews are already answering.

Input you must show	Output you write on the one-pager
Impression share by tag	Which planning-band row you are in
Click share by tag	Whether brand/commercial still pay rent
20-query Overview log	Whether the informational row is actually an Overview problem
Named SEO floor (hours)	What cannot be raided
Dated band (two percentages)	Starting range for this quarter only

Checklist before the percentages leave the room:

• Envelope is organic-search, not company opex
• Both shares are labeled “planning band, dated [date]”
• SEO floor is hours or tickets, not a vibe
• No ROAS cell
• Re-pick date is on the page (end of quarter)

What does AEO spend buy that SEO spend does not?

Inclusion in the answer, and accurate brand facts inside it. It does not buy a restored informational CTR. It does not buy a ChatGPT “rank.” ChatGPT Search gives you named, cited, absent, or wrong — inline citations plus a Sources panel when search runs. Log those four outcomes.

Google’s May 2026 line is the shared floor: SEO best practices continue to be relevant because AI Overviews and AI Mode retrieve from the Search index. Eligibility is still indexed and snippet-eligible. AEO dollars after that floor buy a different acceptance test.

Dollar	Classic SEO is buying	AEO is buying	Shared (do not double-pay)
Technical	Crawl, canonicals, speed, indexation	Same eligibility — do not relabel it “AEO”	Robots, rendering, URL Inspection
Content	Cluster coverage, commercial intent pages	Quote-ready leads, tables, dated claims	The URL itself
Measurement	Rank, clicks, GSC Web performance	Prompt panel, citation/mention/accuracy log	Generative AI impressions report if the property has it
PR / links	Referring domains, classic E-E-A-T	Off-site facts engines can agree on	A real corroborating page, not a fake newsroom
Tools	Semrush-class SERP views	Human-reviewed answer log; dashboards are optional	Do not buy two retainers for one ritual

Ahrefs’ overlap drop — 76% of Overview cites also in the top 10 in July 2025, 38% in March 2026 — is why rank-only spend is a weaker proxy than it was a year earlier. Rank still helps you enter the pool. It is a worse guarantee that you will be the named source.

Pew’s 1% in-unit click rate is why you should not fund AEO as if the Overview were a new #1 referrer. Citation is brand-plus-residual-clicks. If the only success definition is “the Overview becomes our traffic source,” you will hate the spend and you will be right to.

I will cite receipts I can say out loud: SEO certified since 2021, 500+ automations built, 20,000+ hours on agentic systems, 35,000+ hours saved for clients, hundreds of production sites. I will not invent an AEO ROAS for your deck.

What must stay in the traditional SEO column?

Everything that keeps you in the candidate pool, plus the URLs that still close. If you starve this column because informational CTR fell, you will fund citations for pages Google cannot retrieve.

Google’s optimization guide is blunt on two points operators skip: you do not need a page per synonym to chase query fan-out, and manufacturing a page for every variation is scaled content abuse, not AEO. Keep the commercial cluster deep. Do not spawn thin AEO variants to “cover AI.”

Keep funding	Why it stays SEO-column	Kill criterion
Indexation, robots, canonicals, rendering	No snippet, no Overview link, no ChatGPT source chip	Open crawl bugs for more than two weeks
Money pages (pricing, demo, services, packages)	These still convert when someone leaves the answer	“Optimize” them into glossaries
Brand SERP + NAP + GBP	Brand queries are the control group	Entity facts that contradict schema / About
Intent mapping and internal links to protectable URLs	AEO pages need a next step that is not “read more blog”	Orphan citation pages with no handoff
Classic PR that creates real corroborating URLs	Engines agree with the open web	Paying for a dashboard logo instead of a page
Semrush-class SERP tooling	You still need to see the SERP	Using share-of-voice as a citation rate

Protect vs starve checklist:

• Pricing / demo / contact URLs have owners and are not in the “rewrite for AI” pile
• Technical SEO has a floor (hours or dollars) that cannot be raided for blog volume
• Brand queries are reported separately so informational CTR cannot hide a brand problem
• Internal links from reinvented explainers point at protectable URLs
• You are not buying an llms.txt as an Overview ticket — Google Search does not use it as a ranking or inclusion file

The SEO column is not nostalgia. It is the retrieval door. AEO is what you do after you walk through it.

What should move into the AEO column first?

Truth layer, then five money questions, then the log. Not a 40-post calendar. Not a tool bake-off. The audit checklist is the field sheet; this is the budget translation.

First-dollar order for the AEO share:

1. Freeze 25–40 prompts (and the same 3–6 competitors) before you change a page.
2. Baseline what ChatGPT, Perplexity, and Google AI Overviews already say. Dated screenshots.
3. Fix fact conflicts: price, city, offer, founding, NAP — site, schema, directories.
4. Rewrite five buyer questions answer-first: lead in the first 60 words, one table or procedure, a handoff to a protectable URL.
5. Assign a named owner to a weekly prompt panel. If the name is empty, the AEO share is decoration.

First AEO spend	Pass looks like	Waste looks like
Prompt panel	25–40 prompts, weekly owner, four outcomes logged	A dashboard with no frozen set
Truth layer	One fact sheet; site/schema/directories agree	Three About pages with three prices
Five answer-first URLs	Shipped HTML, not outlines	Twenty “AI content” tickets, zero fetches
Corroboration	One real off-site fact that matches the site	A newswire blast with no follow-up URL
Measurement	Panel + GSC gen-AI impressions if available	“AEO score” as a vanity average

Which pages to fund first inside that five:

Page type	AEO-column yes	AEO-column wait
Comparison / vs / best that already triggers Overviews	Yes — criteria table	Another listicle with ten near-duplicates
Category definition buyers ask in chat	Yes — dated definition	Glossary spam for every synonym
Pricing / demo	Only for fact accuracy and schema	Turning it into a 2,000-word explainer
Thin how-to with collapsing CTR, no citation, no leads	Reinvent or merge	“More posts” as the strategy
Case page with named outcomes you can stand behind	Protect; do not invent numbers	Attaching a fake lift % to a portfolio name

Local service companies (the AllCity HVAC shape) usually belong in the commercial/local row: Maps and service pages first, light AEO when category prompts already name competitors. A brand-heavy artist site usually belongs in the brand row: protect the URLs that sell, do not blow the quarter rewriting bios for a model that is not sending the tour traffic. I will not attach invented conversion rates to either.

If the question is “what pages should I fix first for AI search?” rank tickets in this order. It is the AEO-column queue, not a reason to skip the SEO floor.

Priority	Page	Why it is first	Done this month
1	Any URL whose facts contradict schema, About, or directories	Wrong citations are worse than absence	One fact sheet; conflicts closed
2	Comparison / vs / best URLs that already show Overviews	Seer’s informational slice is where Overviews cluster; confirm on your SERP log	Criteria table + dated lead
3	Category definitions buyers ask in ChatGPT / Perplexity	Inclusion on the shortlist question	Quote-ready definition, one table
4	How-tos with the impressions-up CTR-down fingerprint	Page job change, not more volume	Lead + procedure + handoff
5	Everything else	Not first	Park it

Procedure for the five:

1. List URLs that match rows 1–4.
2. Drop any URL that is a pricing/demo/contact page — those stay in the protect column except for fact accuracy.
3. Ship the first two as HTML this month. Index request only after the fetched document contains the lead answer.
4. Re-run the same prompts. If the answer did not move, do not “add more posts.” Change the object on the page (table, steps, fact).

How do you split people, tools, and content inside one envelope?

By job, not by invoice. Two agencies with two roadmaps is how you pay twice for the same crawl and still have no prompt panel. One owner, two KPI columns. Split teams only when volume and risk justify a visibility pod.

Cost center	Classic SEO hours	AEO hours	Default (small team)
Technical SEO	Floor every sprint	Eligibility checks only	Same person; do not hire “AI technical” as a second religion
Content	Commercial pages + cluster hygiene	Answer-first rewrites of the five money questions	Same writers; change the brief and the acceptance test
Measurement	GSC Web + rank views	Weekly panel + SERP log	Same analyst; two tabs, not two retainers
PR	Links you can defend	Corroborating facts	One outreach list with a fact-sheet brief
Tools	SERP + crawl	Optional mention trackers	Buy the ritual first; the SKU is optional

Default staffing rule I will actually say on a call:

• One named owner of the frozen mix table and the prompt panel (usually the SEO or content lead).
• Technical SEO keeps a floor even in an AEO-weighted band.
• Writers ship tables and lead answers, not word-count.
• Nobody is “the AEO person” whose job is slides.

Team size	Do this	Do not
One marketer	2–4 hours/week on the panel + two reinvented URLs this month	A six-month “AI content” retainer
SEO + content	Split the two columns on one weekly doc	Two vendors, two decks, no shared prompts
In-house + freelancer	Freelancer ships the five URLs against the audit tickets	Freelancer “does AEO” with no access to Search Console
Multi-brand / agency	Isolation of panels per property	One blended “AI SOV” across clients

Tool spend is the easiest way to fake a split. Semrush-class views still help you see SERP shape. They do not replace a human-reviewed Overview and chat log. If the AEO share is 90% software and 10% pages, you bought a logo.

I have shipped hundreds of production sites and 500+ automations. The failure shape is identical across lanes: buying a program nobody will operate.

Sample 40-hour week inside an informational-heavy band — a staffing sketch, not a timesheet you must photocopy:

Hours	Job	Column
8	Crawl, index, speed, URL Inspection	SEO floor
8	Money-page and internal-link hygiene	SEO floor
8	Prompt panel + SERP log + one-pager	AEO
12	Answer-first rewrites (tables, leads, fact alignment)	AEO
4	Corroboration / PR against the fact sheet	Shared

If you only have 10 hours, keep 4 on the SEO floor and 6 on the panel plus one URL. Do not spend the 10 on a tool evaluation.

What failure modes wreck the split?

The freeze, the 100% AEO swing, and judging citations with a ROAS slide. Those three kill more programs than “we picked 45 instead of 55.”

Failure	What breaks	What it costs	What you do instead
SEO freeze	“AI killed clicks, pause the blog / pause technical”	Eligibility decay; brand search decays months later	Protect money URLs; reinvent the informational cluster; keep the crawl floor
100% AEO swing	Crawl, commercial pages, and internal links starve	Pages never enter the candidate pool you are paying to be cited from	Keep the SEO floor; AEO is a layer
Two religions	SEO agency + AEO vendor, two roadmaps	Duplicate discovery, zero shipped HTML	One audit, one ticket list, two KPI columns
CTR as the AEO KPI	Informational CTR cannot return to 2019	You cancel the only work that can win the citation	Citation rate + brand + protectable conversions
Fabricated ROAS	“Citations return X:1” with no primary study	You will be asked to repeat a number you cannot source	Refuse the number; show the panel vs week 1
Mix from the org chart	Finance splits by vendor invoice	The demand mix never enters the room	Tag queries first; invoices second
Calendar as strategy	Volume to “outrank the AI”	Thin pages Google already told you not to mass-produce	Five URLs, then expand

Worked freeze: impressions hold, CTR falls on how-to queries, leadership cuts content and technical in the same week. Six months later brand queries sag because you stopped being the cited explainer, and the money pages are slower and less internally linked. You saved a line item and bought a demand hole.

Worked swing: the team relabels every ticket “AEO,” ships llms.txt, and stops fixing canonicals. Google still needs a snippet. ChatGPT Search still needs a crawlable page. You funded a file Google Search does not use as an inclusion ticket.

Pew’s 8% vs 15% and SparkToro’s 68.01% zero-click figure will get pasted into both failure decks. They are click-curve context. They are not permission to freeze, and they are not a ROAS.

Checklist that the split is already wrecked:

• Nobody can show a dated mix table
• The AEO vendor cannot diagnose a robots issue
• Success is defined as “get our old blog CTR back”
• A conversion multiplier appears in the SOW with no primary study attached
• Technical SEO hours went to zero in the same quarter AEO started

If three boxes are checked, stop adding budget. Repair the scoreboard.

How do you set the split for this quarter?

One working session, then ship HTML. Do not open a strategy offsite. The output is a dated row from the planning-band table plus ten tickets.

Procedure:

1. Classify mix (the six-step export above). Freeze the query set.
2. Run the worth-it gate from is AEO worth it: buyers already ask AI, and someone will log the panel. If not, stay SEO-weighted or wait.
3. Pick one row from the planning-band table. Write the two percentages as this quarter’s starting range, with the date and the mix shares next to them.
4. Lock the SEO floor: technical hours, money-page owners, brand SERP.
5. Lock the AEO first dollars: panel, truth layer, five URLs — in that order.
6. Translate into tickets with owners. If you need the field sheet, use the AEO audit checklist.
7. Define kill criteria for the quarter (next section). Put them on the same page as the percentages.
8. Re-pick the row at the end of the quarter from the scoreboard, not from a new think-piece.

Week	Artifact	Done looks like
1	Mix table + 20 SERP screenshots	Tags, Overview yes/no, cited domains
1–2	Prompt panel v1	25–40 prompts, owner named, week-1 answers stored
2–3	Truth-layer tickets	Fact conflicts listed with URLs
3–6	Two to five answer-first URLs shipped	Fetched HTML contains the lead answer
6–12	Weekly log + one-pager	Keep / swing 10 points / wait decision

Do not open twenty content tickets on day two. Overviews punish undifferentiated libraries. They do not reward volume for its own sake.

Quarterly swing rule:

Scoreboard after 12 weeks	Swing
Citations up or accurate more often; brand search stable; money-page conversions held	Hold the band, or tip 10 points toward AEO if informational demand is still the bulk
Citations stuck at zero on money prompts; eligibility is clean	Keep AEO hours; change the five URLs, do not buy a bigger dashboard
Eligibility broken (indexation, robots, snippet)	Swing back to SEO floor until URL Inspection is clean
Brand search in free-fall after you cut explainers	You cut demand. Restore the citation engines; do not call it “focus”
Money-page conversions down while you rewrote them as glossaries	Move those URLs back to the SEO/protect column this week

The swing is 10 points, not a new religion. If you need a 40-point jump, you classified the mix wrong in week 1.

How do you measure whether the split is working?

Two columns. No blended vanity score. No imported ROAS. If informational sessions fall while citation rate, brand search, and protectable conversions hold or rise, the split is doing the job. If all four fall, you have a demand or eligibility problem.

Inclusion language for the AEO column is named, cited, absent, or wrong — not a renamed rank report. This page only cares how that scoreboard tells you to move the band.

Column	KPI	Source	Review	Kill / swing criterion
SEO	Indexation / crawl errors	Search Console, URL Inspection	Weekly	Open blockers → steal hours back from AEO until closed
SEO	Clicks (not only CTR) on commercial + brand URLs	Search Console	Weekly	Both clicks and impressions down → demand, not “AEO failed”
SEO	Conversions on protectable URLs	Analytics / CRM	Weekly	The number that pays rent
AEO	Citation rate on the frozen prompt/query panel	Manual log (+ gen-AI impressions if available)	Weekly	0% for a quarter on money prompts, eligibility clean → change pages, not the acronym
AEO	Fact accuracy (wrong price/city/offer)	Same log	Weekly	A wrong citation is a truth-layer incident, not a win
Both	Brand queries	Search Console	Weekly	Free-fall after cutting explainers → you cut the AEO-shaped demand engine

What you will not put in the scoreboard:

Tempting metric	Why it lies here
Sitewide CTR	Mixes Overview zero-click with ranking loss and brand
A vendor “AEO score”	Undefined, not your frozen panel
Informational CTR vs 2022	The product changed; Pew and Seer already showed compression
ROAS on citations	No honest last-click path; Pew’s 1% in-unit click rate should humble that slide
Averaged Pew + Ahrefs + Seer into one “industry CTR”	Different objects, different methods

Google launched a Search generative AI performance report in June 2026 that isolates impressions in AI Overviews and AI Mode. The help-center writeup says it was rolling out to a subset of properties and that it is impression-led. If you do not have it yet, do not invent a workaround number. Use the Web report plus the manual log.

Measurement checklist:

• Frozen query set and frozen prompt panel (same IDs for 90 days)
• Brand vs non-brand split
• Overview-present vs Overview-absent on the SERP log
• Clicks and impressions as two series
• Citation / mention / accuracy as four outcomes, not a single score
• No ROAS cell on the AEO column

If someone asks “what’s the ROI,” answer with the panel delta and the pipeline on protectable URLs. If they need a multiplier to approve the band, they are asking you to fabricate. Say no.

What should you skip if you only have a week?

Skip the calendar, the bake-off, and the reorg. A week can produce a mix table and a band. It cannot ship a new editorial religion.

Do this week	Skip this week
Tag 50–80 GSC queries informational / brand / commercial / local	A new 40-post outline
Twenty SERP screenshots (Overview yes/no, cited domains)	A tool comparison matrix
Pick one planning-band row and write the date	A “70/30 because the industry” slide
Name the panel owner	Hiring an “AEO team”
List fact conflicts on price/location/offer	Schema theater with no fact sheet
Protect list: money URLs that may not be rewritten	Rewriting the homepage into a glossary

One-week procedure:

1. Export 90 days. Tag the top queries. Sum impression share by tag.
2. Spot-check 20 SERPs. Note who is cited.
3. Circle the planning-band row. Write “starting range for Q[n], dated [today].”
4. Write the SEO floor in hours (technical + money pages).
5. Write the AEO first dollars as: panel, truth layer, five URLs — even if they slip to week two.
6. Stop. Do not publish ten posts to “show momentum.”

If the week ends with a pie chart and no tagged queries, you performed finance, not visibility.

When is a visibility audit the right next step?

When the mix table is political, the data is muddy, or two vendors are selling two religions. DIY the classification if you have Search Console and one person who will own the log. Hire when the argument is the blocker.

Run or hire an audit when:

• You cannot agree whether you are informational-heavy or commercial-heavy after two weeks of GSC
• Competitors own the Overviews and chat answers on category terms, and you cannot prioritize ten tickets
• Entity conflicts or technical eligibility keep blocking inclusion
• Finance wants an external hammer before moving 10 points
• Someone is pushing a universal percentage or a fabricated ROAS as the decision

Situation	DIY	Audit
Clean GSC, one owner, obvious informational mix	Yes — pick the band, ship five URLs	Optional later
Two agencies, two decks	No	Yes — one ticket list
Crawl / snippet eligibility unclear	Maybe, if you can run URL Inspection	Yes if it has been open more than a sprint
Leadership wants “the number”	Give them a dated row, not a fake average	Audit if they will not accept a range
Worth-it still unknown	Read the worth-it spoke first	Audit as the 30/60/90 gate

The lane page is /visibility. The method is the playbook. I will not attach a fabricated lift percentage to that path. The acceptance test is the week-12 panel versus the week-1 panel on the same prompts, plus money-page conversions that did not fall off a cliff.

FAQ

How much of my marketing budget should go to AEO vs traditional SEO?

There is no universal percentage. Split the organic-search program from your informational vs brand vs commercial mix, keep a technical-SEO floor, and treat the numbers as a dated planning band — not a published industry statistic. Informational-heavy programs start AEO-weighted; brand- and commercial-heavy programs stay SEO-weighted with a smaller citation and truth-layer line.

How do I measure whether the AEO vs SEO split is working?

Run two columns on a frozen query and prompt panel: classic SEO (eligibility, clicks on commercial and brand URLs, conversions on protectable URLs) and AEO (citation rate, mention accuracy, brand-search trend). If informational CTR falls while those AEO and brand numbers hold, the split is doing its job. If both columns fall, you have a demand or eligibility problem — not proof that AEO “has no ROAS.”

What usually fails first when teams try this?

The scoreboard. Teams freeze SEO because informational CTR dropped, or they rename the team “AEO” and keep rank-only reports. The first operational break is usually no dated mix table and no named owner of the prompt panel.

How long does this take to show results?

A usable mix table and a prompt panel can exist in two weeks. Citeable pages can ship inside 30 days. Directional citation movement on some prompts is a quarterly object, not a week-two promise. Full recovery of pre-Overview informational CTR is not a 90-day guarantee and often not a guarantee at all.

What should I skip if I only have a week?

Skip a new content calendar, a tool bake-off, and a reorg. Classify 50–80 queries, screenshot twenty SERPs, and pick one planning-band row with a date. That is a split. A week of new posts is not.

When is this not worth doing yet?

When crawl, indexation, or entity facts are on fire, or when nobody will log a prompt panel. Fix eligibility first. If buyers never ask AI in your category and sales has never heard it, stay SEO-weighted and re-check quarterly instead of buying a second religion.

CTA

Stop shopping for a universal AEO vs SEO percentage. Classify the mix, pick a dated planning band, and keep a crawl floor.

Lane: /visibility · Book a visibility audit.