How much should you budget for AI Search in 2026 is the wrong question until you know what number you can actually defend to a CFO. According to Conductor’s 2026 State of AEO/GEO CMO Investment Report — a vendor-sponsored survey of over 250 enterprise digital leaders (CMOs, VPs and Senior Directors) drawn from US enterprise organisations with 500+ employees across 12+ industries — enterprise AEO/GEO budgets concentrate on content restructuring, measurement and tooling, and 12% of digital budgets on average went to AEO in 2025 with 94% of respondents planning to increase that share in 2026. Treat those figures as vendor-attributed industry data, not neutral evidence, and size your own number from your own account facts.
Operating rule
A defensible AI-Search budget starts from three account numbers you can measure today: current AI-referral share, structured-data coverage, and the pages already eligible for AI Overview citation. Anything else is a benchmark you cannot audit.
Why an AI-Search budget line item is real work now
The peer-reviewed / preprint literature disagrees on activation rates, which itself is useful. Xu, Iqbal and Montgomery’s 2026 arXiv preprint “Measuring Google AI Overviews” (submitted 13 May 2026, under review) issued 55,393 trending queries across 19 topical categories over 40 days and measured overall AI Overview activation at 13.7%, rising to 64.7% for question-form queries. The same study decomposed AI Overview responses into 98,020 atomic claims and found 11.0% were unsupported by the cited pages. Both figures are preprint-status, and the authors describe the work as under peer review.
Grossman et al.’s “How Generative AI Disrupts Search” — submitted 30 April 2026 and accepted to ACM SIGIR 2026 — analysed 11,500 user queries and reported AI Overviews on 51.5% of queries with an average Jaccard similarity below 0.2 between engines’ retrieved sources. Do not average those numbers into one; the query sets are different and the studies asked different questions. The point is not the exact activation rate. The point is that AI Search now sits between your content and a large chunk of user queries, and that reality carries a budget line whether you name it or not.
Visual slot: Add a “where the 2026 AEO/GEO budget line item lives” map here — five buckets (content restructuring, structured-data audit, measurement, tooling, agency/hours) with the reader’s own approximate share ranges. Replace this block with the final visual later.
What Google itself says AI Search means for publishers
Google’s own eligibility statement is short and worth quoting. Google Search Central’s AI features documentation (last updated 10 December 2025) reads: “There are no additional requirements to appear in AI Overviews or AI Mode, nor other special optimizations necessary.” The same page adds: “You don’t need to create new machine readable files, AI text files, or markup to appear in these features. There’s also no special schema.org structured data that you need to add.” That eligibility statement is scoped: to be shown as a supporting link, a page must be indexed and eligible for a snippet under Google’s baseline Search rules.
The practical reading is not “AEO is unnecessary”. The practical reading is that Google is telling you not to buy a separate AEO-specific stack. Existing snippet controls apply (nosnippet, data-nosnippet, max-snippet, noindex). Baseline Search discipline — the “Introduction to structured data” page reminds publishers that “you should rely on the Google Search Central documentation as definitive for Google Search behavior, rather than the schema.org documentation” — is the AEO discipline.
What the “250+ CMO investment” report actually shows (and does not)
Reading rule
Every Conductor number is disclosed-methodology first-party industry data — not neutral evidence. Attribute the survey every time it appears, and do not stretch it beyond the reported figures.
Conductor’s 2026 investment report discloses its scope in one sentence: “CMOs and digital leaders were exclusively sourced from enterprise-level organizations (500+ employees) across 12+ industries in the U.S.” The headline figures on the page are worth reading carefully rather than in an aggregated summary. Conductor reports 97% saying AEO had a positive impact on the marketing funnel in 2025, 94% planning to increase AEO investments in 2026, 56% having made significant or high investments in 2025, 51% leveraging a fully integrated AEO platform, 12% on average allocated to AEO from digital budgets, and 73% classifying their AEO programs as advanced or very advanced.
What Conductor does not publish in that report is a specific “share of teams building AEO in-house rather than outsourcing”. Any secondary write-up that names a precise in-house-vs-outsource percentage from this study is repeating a number that is not in the primary source. This is the kind of drift a defensible enterprise budget cannot tolerate. If a stakeholder asks where the in-house percentage came from, the honest answer is that Conductor’s investment report does not name one.
| Conductor 2026 figure | What it measures | Attribution required | Do not extend into |
|---|---|---|---|
| 250+ enterprise digital leaders | Survey sample size | Conductor’s 2026 investment report | A universal enterprise average |
| 12% average of digital budget on AEO | Self-reported 2025 allocation | Same report, dated 2026-04-14 | A per-industry benchmark |
| 94% plan to increase in 2026 | Self-reported intent | Same report | A guaranteed increase |
| 97% positive marketing-funnel impact | Self-reported outcome | Same report | Neutral third-party outcome data |
| 73% advanced / very advanced maturity | Self-classification of program | Same report | Verified capability rating |
AI referral traffic today: the number your board will ask about
The commercial context comes from a separate Conductor artefact. Conductor’s 2026 AEO/GEO Benchmarks Report (updated 6 July 2026, drawn from 13,770 domains and 3.5M unique prompts between May and September 2025) reports that AI referral traffic accounts for 1.08% of all website traffic across ten industries, with ChatGPT accounting for 87.4% of that AI referral traffic and IT sitting highest at 2.80%. Those are aggregate cohort numbers from Conductor’s own dataset. Boards ask about them because “AI referral share” is easier to remember than “AI Overview activation rate”, but it is not the same measurement.
A defensible budget conversation is honest about the size of the current channel. If your AI referral share is below Conductor’s 1.08% aggregate, that does not mean you are behind; it means your board should not expect AEO/GEO investment to move top-line revenue tomorrow. If your share is above it, especially in an IT-adjacent vertical, it means the channel is already contributing something you can measure — which is exactly where a serious budget line item belongs.
Where the AEO/GEO budget actually goes
The five buckets that account for most enterprise AEO/GEO spend are: (1) content restructuring — rewriting existing pages so their answer to the top buyer questions is a paragraph a stranger can trust, (2) structured-data audit — closing the gap between what Google’s Search Central documentation calls definitive and what your CMS actually emits, (3) measurement — the new Search Console generative AI performance report plus dashboarding on top of GA4 and CRM, (4) tooling — LLM visibility monitoring, prompt-based citation tracking, and integrated AEO platforms, and (5) agency / analyst hours — someone senior enough to defend the budget line to finance.
Visual slot: Add a “before you name a number” checklist board here — current AI-referral share, structured-data coverage, GSC generative-AI report access, top-20-question audit, sales handoff, senior owner. Replace this block with the final visual later.
| Budget bucket | What is bought | Fair signal it is working | Do not claim |
|---|---|---|---|
| Content restructuring | Rewrites, decision tables, buyer-question audits | Long dwell, saves, GSC generative-AI impressions | Guaranteed AI citation |
| Structured data | Schema audit, Search Central-aligned markup | Rich-result eligibility per Google’s documentation | Rich results guaranteed |
| Measurement | GSC generative-AI report, dashboards, LLM prompts | Baseline + week-over-week AI referral share | A universal AI-referral benchmark |
| Tooling | Integrated platform or focused monitoring stack | Time saved on citation tracking | Tool ROI as evidence of channel ROI |
| Agency / hours | Senior AEO/GEO strategist and analyst time | A written monthly narrative to finance | Hours as evidence of business outcome |
Illustrative $50K annual scenario (Assume…)
Assume a mid-market B2B enterprise with an AI referral share around Conductor’s 1.08% aggregate, moderate structured-data debt, and no dedicated AEO analyst. Its US$50,000 annual scenario is illustrative — no case study, no real company, no fabricated result. Roughly a third of the budget goes to editorial rewrites on the twenty pages that receive the most impressions, another third funds a Search Central-aligned structured-data audit and remediation sprint, and the remaining third pays for one paid tool subscription plus a written monthly narrative that a marketing leader can defend upstream.
What that scenario does not include: an FTE hire, cross-market editorial workflow, or a Conductor-tier integrated platform. A budget of this size is a “measure and repair” year rather than a “scale and instrument” year. Its target output is a written baseline plus a documented top-20 question fix, not a percentage lift.
Visual slot: Add an illustrative $50K annual scenario breakdown here — content restructure, structured-data audit, monitoring tool, analyst hours, monthly narrative — with clearly labelled Assume… banner. Replace this block with the final visual later.
Illustrative $250K annual scenario (Assume…)
Assume a large enterprise with a US$250,000 annual scenario — same illustrative caveat, no case study, no fabricated result. The shape is different because the constraint is different. The budget covers a dedicated AI-visibility analyst FTE (or a senior consultant on ongoing retainer), an editorial workflow across two or three languages, an integrated AEO platform (Conductor-tier or the equivalent competitor), and a formal executive dashboard that lands in finance every month with a written narrative rather than a chart alone.
What that scenario does not buy: universal AI citation, a guaranteed ranking lift, or a promised percentage of AI-referral share. The most honest metric the budget can defend at year one is a controlled month-over-month change in AI-referral share, structured-data coverage percentage across the target templates, and comment-quality plus save-through-rate on the pages the team specifically restructured.
Visual slot: Add an illustrative $250K annual scenario breakdown here — FTE, editorial workflow, integrated platform, executive dashboard, monthly report — with clearly labelled Assume… banner. Replace this block with the final visual later.
| Line item | $50K scenario share | $250K scenario share | What it does not buy |
|---|---|---|---|
| Content restructuring | ~35% | ~25% | An AI citation guarantee |
| Structured-data audit + remediation | ~25% | ~15% | A rich-result guarantee |
| Measurement + dashboarding | ~10% | ~15% | A universal AI-referral benchmark |
| Tooling | ~15% | ~20% | Channel ROI as tool ROI |
| Analyst / FTE hours | ~15% | ~25% | A promised revenue outcome |
A sizing framework: org size × maturity × channel overlap
Instead of forcing a percentage onto a total marketing budget, size AI-Search work along three axes at once. Organisation size sets the range (small mid-market, mid-market, large enterprise, global enterprise). Maturity — Conductor’s report classifies its own respondents as “advanced or very advanced” 73% of the time, but that is self-classification of a program, not verified capability — sets the shape of the spend. Channel overlap with existing SEO, content, and paid search decides how much of the AEO/GEO number is genuinely new spend and how much is a reallocation.
A team that already has a strong SEO practice, a Search Central-aligned structured-data pipeline, and content editors who can be redirected to AI-answer-shaped pages will spend more on tooling and measurement than on content restructuring — because the restructuring is a redirection, not a new hire. A team without those foundations should not spend on tooling first. The pipeline that generates the tool’s inputs still has to be built.
Visual slot: Add a sizing framework matrix here — org-size rows, maturity columns, cells showing the primary budget bucket to prioritise first. Replace this block with the final visual later.
Measurement: what you can actually see in your own data
Google added a new measurement surface in mid-2026. Google’s Search Central Blog announced the Generative AI performance report in June 2026, and the Search Console Help page defines the report as data on “how your site performs in generative AI features on Google Search”, filterable by pages, countries, dates or devices, with impressions defined as the count of times “links to your site were shown to a user in a generative AI feature”. Rollout is gradual — the Help page notes not all properties have access yet.
The disciplined move for a budget owner is to bind the spend to that report from the beginning. If your property is not yet enrolled, the plan is to instrument the rest of the stack — content baseline, structured-data coverage, LLM visibility monitoring — so that when the report opens, the trend arrives with context rather than a blank chart. A budget that assumes the GSC report will hand you a number without prior baselines is a budget that will re-plan itself in the middle of the year.
Visual slot: Add a vendor-report attribution ladder here — academic vs Google-official vs vendor first-party research vs blog commentary — with the safe usage rule per row. Replace this block with the final visual later.
Risks: vendor attribution, single-quarter over-reads, generative-AI drift
Risk rule
Every AI-Search stat you cite has a source class — official, academic or vendor. Naming that class every time is how a budget line survives the second quarter of scrutiny.
- Vendor attribution: Conductor’s figures are useful and disclosed-methodology; they are not neutral. Treat them as first-party industry data and attribute every use.
- Single-quarter over-reads: A weekly AI-referral swing after one content sprint is not a channel trend. Wait for enough data before rebasing the plan.
- Generative-AI drift: LLM providers change retrieval behaviour without notice. A visibility number good today may not repeat next month; do not underwrite yearly ROI on a snapshot.
- Preprint over-generalisation: Xu et al. is under peer review; Grossman et al. is accepted at SIGIR 2026. Cite them with their status. Do not blend their query sets.
- Structured-data as ranking myth: Google’s own documentation frames structured data as eligibility for rich results, not a ranking guarantee. A budget claim built on “guaranteed rich results” fails a second-read.
Common mistakes when naming an AI-Search budget number
These patterns show up in almost every enterprise plan that tries to name an AEO/GEO number before doing the account work. None of them is unique to AI Search; the next channel disruption will surface the same list.
- Quoting a “93% build in-house” figure that a vendor report does not actually publish.
- Averaging different studies’ AI-Overview activation rates into a single number.
- Presenting a 1.08% AI-referral aggregate as a per-industry benchmark.
- Buying an integrated AEO platform before the structured-data pipeline exists.
- Reporting a single-week AI-referral spike as evidence of a channel trend.
- Assuming Google’s “no special markup” statement means AEO is not real work.
- Naming a $ number in a plan before instrumenting the GSC generative-AI performance report.
Choose the budget from what you can already measure
Rather than defending a fixed percentage of the marketing budget, defend a specific list: current AI-referral share as measured on your own property, structured-data coverage percentage on the templates that carry buyer questions, the number of pages already earning generative-AI impressions in Search Console (where available), and the write-up of the top-twenty questions your buyers actually ask. A budget line that answers to those four measurements survives the second quarter without a re-plan, because the four measurements will keep moving even when the vendor benchmarks do not.
Write the budget the same way you would write an ad-account review for finance: three numbers you can already measure, two numbers you promise to instrument this quarter, and one honest sentence about which vendor’s benchmark you are and are not extrapolating from.
A team that can hold that discipline for two quarters will have earned the right to increase the line — and the number to name in the next plan will come from the account, not from a chart on a vendor’s homepage.
Want a defensible AI-Search budget number written against your own account?
An AEO/GEO account review reads current AI-referral share, structured-data coverage, and the pages already earning generative-AI impressions — and returns a written budget plan tied to what your own analytics can measure. No fabricated percentages, no vendor extrapolation.
Sources and evidence
- Google Search Central — AI features and your website (updated 10 December 2025)
- Google Search Central Blog — Introducing Search Generative AI performance reports (June 2026)
- Google Search Console Help — Generative AI performance report
- Google Search Central — Introduction to structured data
- Xu, Iqbal, Montgomery — “Measuring Google AI Overviews”, arXiv:2605.14021 (submitted 13 May 2026, under peer review)
- Grossman, Liu, Chen, Smith, Borcea, Chen — “How Generative AI Disrupts Search”, arXiv:2604.27790 (submitted 30 April 2026, accepted ACM SIGIR 2026)
- Conductor — “The State of AEO / GEO in 2026: CMO Investment Report” (vendor first-party research, updated 14 April 2026)
- Conductor — “2026 AEO / GEO Benchmarks Report” (vendor first-party research, updated 6 July 2026)
FAQ
How much should we budget for AI Search in 2026?
There is no single defensible answer without your own account facts. Conductor’s 2026 investment report — a vendor-sponsored survey of over 250 enterprise digital leaders in the U.S. — publishes an average 12% AEO share of digital budget and 94% intending to increase in 2026. Use those as vendor-attributed reference points, and size your own line from current AI referral share, structured-data coverage, and pages already earning generative-AI impressions.
Where is the “93% build in-house” figure from?
Not from Conductor’s 2026 investment report itself. Conductor’s page publishes 97%, 94%, 56%, 51%, 12% and 73% figures. It does not name a specific in-house-vs-outsource percentage. Any secondary write-up quoting “93% build in-house” from that study is repeating a figure that does not appear in the primary source. Do not cite it.
Is Conductor a neutral source?
No. Conductor sells AEO/GEO software and services and has a commercial interest in the category. Its 2026 CMO Investment Report and 2026 AEO/GEO Benchmarks Report both carry disclosed methodology (250+ enterprise digital leaders; 13,770 domains and 3.5M prompts respectively). Treat their figures as first-party vendor industry data, cite the report every time you use one, and do not present a Conductor number as neutral evidence.
What does Google itself say we need to do for AI Overviews?
Google’s Search Central AI-features page states: “There are no additional requirements to appear in AI Overviews or AI Mode, nor other special optimizations necessary” and “You don’t need to create new machine readable files, AI text files, or markup to appear in these features.” Existing snippet controls apply. Baseline Search discipline — indexing, snippet eligibility, Search Central-aligned structured data — is the AEO discipline.
How much of my AI Overview activation rate can I predict?
Not much. The Xu et al. arXiv preprint measured overall AI Overview activation at 13.7% across 55,393 trending queries; Grossman et al. reported 51.5% across 11,500 different queries. Both figures are peer-review-status specific. Do not present either as your predicted rate; use them to inform the shape of the plan and instrument Google’s own Search Console generative AI performance report to see your property’s real activation and impressions.
Are the $50K and $250K scenarios accurate benchmarks?
No. Both scenarios are illustrative, clearly labelled Assume… and carry no case study or fabricated outcome. They exist to show the shape of a budget at two different constraints — a mid-market “measure and repair” year, and a larger “scale and instrument” year — not to serve as a market benchmark. Size your own budget from your own account facts.
Turn the framework into a number you can defend.
An AEO/GEO consulting engagement produces a budget plan tied to your current AI-referral share, structured-data coverage, and the pages that already earn generative-AI impressions — attribution intact, no vendor extrapolation, no guaranteed percentages.