Which budget should AI search visibility come out of?

AI search visibility should come out of your search and content budgets, because that is where the people and the pages already are, and it should be governed with the patience you give brand rather than the monthly scrutiny you give paid media. Both halves of that sentence matter. Where the money comes from is an organisational question. How it is judged is the one that decides whether the programme survives its second quarter.

The work itself is page work. Somebody rewrites a page so it answers a buyer's question in the buyer's words, somebody checks that engines can fetch it, somebody runs prompts monthly and reports what changed. All of that sits with whoever owns the website, which in most B2B companies is the SEO or content owner. Putting the budget anywhere else creates a line item whose owner cannot approve the only deliverable.

The judgement question runs the other way. If the line sits inside a performance budget, it inherits performance metrics, and within two quarters somebody will ask for cost per acquisition from ChatGPT. That number either does not exist or is built on attribution nobody can defend, and the programme gets cut for failing a test it was never going to pass.

So the practical answer most companies land on is a ring-fenced allocation inside search or content, with a reporting line that reads like brand reporting: presence, share of answers, which competitors are named instead of you. Our page on getting budget approved for AI search visibility covers the internal case; this page is about which pocket it comes from.

What are other companies actually funding it from?

Other companies are funding it mostly from acquisition budgets, and the split is documented. The Fractl survey reported by Digiday on 7 August 2026 put performance marketing teams at 32% of their budgets going to AI search visibility and SEO teams at 31%, with brand at 21% and content at 20%. Across the sample, about 24% of search or content budgets were going to AI visibility, 82% of marketers had allocated at least some budget, 43% were spending more than 20%, and 18% had allocated nothing at all.

Read that against what the channel does and a mismatch appears. The two teams spending the largest share are the two whose success is defined by measurable acquisition, and this is a channel whose output is being named in an answer. Fractl is an agency rather than a research body and Digiday does not publish the sample size, so treat the percentages as a picture of practice rather than a measurement of what works.

The 18% who have allocated nothing are worth a second look too, because they are not obviously wrong. A company with no indexable content, or in a category where engines refuse to name vendors, is better off waiting. We argue the case for waiting in is AEO worth it, or should you wait.

What the survey does establish is that the money is moving, which changes the internal conversation. The question in most companies is no longer whether to fund any of this. It is which existing line gets smaller, and that is a harder conversation than the one about whether AI search matters.

How much should you set aside in the first year?

Set aside enough to rewrite the pages that answer your buying questions and to re-measure monthly, which in practice means a programme rather than a project. The published percentage guidance is agency guidance and should be read that way: Brainlabs, in an undated post by Jeremy Hull read on 27 October 2026, suggests around 7 to 10% of digital marketing budget for generative engine work at a large brand, or adding roughly 20% to existing SEO investment where a separate allocation is not possible.

No independent study supports a particular percentage, and the absence is worth saying out loud on a page about budgets. What exists is practice, which the Fractl numbers above describe, and vendor recommendation, which is what the 7 to 10% is. Anyone quoting you a benchmark allocation for your industry is extrapolating.

What actually moves the number is scope rather than category. How many pages need rewriting, how much original research the category needs before you have anything to say, whether the engines are also stating wrong facts about you, and how much of the work your own team can absorb. Our current figures live on the pricing page rather than in an article, because a number copied into a guide goes stale the day it changes.

One budgeting detail catches teams out in month four. The work is maintenance, not construction: Seer Interactive, studying 7,683 cited pages in July 2026, found 75% of them updated within the last year. A budget that funds twenty rewrites and nothing afterwards buys a set of pages that will age out, which is why we treat this as a retainer question in is AEO a project or a retainer.

Why does funding it from the performance budget go wrong?

Funding it from the performance budget goes wrong because performance budgets are judged on attributed clicks, and this channel produces very few clicks and almost no clean attribution. The work can be going well and the performance report will show nothing.

Start with volume. SE Ranking's study of 101,574 websites, published 18 June 2026, found AI platforms accounting for 0.32% of all website traffic in 2026, against 0.24% in 2025, with organic search at 42.75% and Google sending about 134 times more visitors than AI platforms combined. Growth is fast and the base is tiny. A line item expected to deliver traffic at that scale will be cancelled, correctly, on its own terms.

Then attribution, which is worse than most plans assume. Kevin Indig, publishing on 19 October 2026, compared GA4 data across 12 sites and ran 632 timing tests across four brands against Direct traffic and branded search. Total Direct traffic rose 140.9% between January and September while engaged Direct moved by -1.0% on monthly counts, and identifiable AI-traced Direct sessions averaged 0.003% of Direct traffic. His conclusion was that he could not detect hidden AI traffic without self-reported attribution.

The practical consequence is a reporting decision taken before the first invoice. Agree what will count as evidence while nobody is under pressure: share of tracked buying prompts where you are named, which competitors are named instead, which of your pages get cited, and a question on your enquiry form asking where buyers first heard of you. Our page on how to tell whether your AEO is working covers what movement looks like on each of those.

Should you cut your SEO budget to pay for it?

Do not cut the SEO budget to pay for AI search visibility, because classic search is still the larger channel and is also part of how the engines find you. Cutting it to fund the smaller channel usually damages both.

The dependency is the part that gets missed. Chen and Koudas, in a study of 158,099 answers and 468,037 citations published 10 October 2026, found that which pages the engine's own search retrieves explains about 32% of the variation in what gets cited, against 2.8% for the properties of the pages. Being indexed, linked and findable is upstream of every page-level improvement, and that is the work an SEO budget pays for.

The overlap also means the marginal cost of this work is lower than a separate line suggests. A page rewritten to answer a buying question in the buyer's words is a better page for classic search as well, and the technical checks are the same checks. Where the two genuinely diverge is in the title and heading conventions, and our page on AEO versus SEO sets out the handful of places where optimising for one costs you something on the other.

What can reasonably be reallocated is the part of a content budget that funds volume for its own sake. Twenty pages chasing topics nobody asks about are worth less than five pages answering questions buyers type, and that swap funds a meaningful programme without touching anything that currently produces pipeline.

Which budget line should it sit in, and what should it be judged on?

Pick the budget line by asking who can approve a change to a page, then set the judgement criteria in the same meeting. The table sets out the four places companies put this work, what each buys, and the failure mode we see with each.

Budget line	What it funds well	What it should be judged on	Failure mode
SEO	Page rewrites, technical fixes, measurement, the whole core of the work	Share of tracked buying prompts where you are named or cited	Gets absorbed into ranking reports and the AI half quietly stops being measured
Content	New pages for questions you have never answered, original research	Coverage of the buying questions, not volume published	Funds publishing cadence rather than answering, so the library grows and citations do not
Paid media	Nothing in the work itself, though the two compete for the same attention	Not applicable; do not judge this work on cost per acquisition	Cancelled in two quarters for failing a click-based test it cannot pass
Brand or demand creation	The off-site half: being present where engines look, third-party coverage, research	Presence and competitive share of answers over quarters	No page-level owner, so recommendations arrive and nothing on the site changes

The combination that works in most companies is a split: the page work funded and owned inside search or content, the off-site half funded from brand, and one monthly report covering both. Our page on who should own AI search visibility internally goes into the ownership half of that arrangement. If you want the page-level scope priced from your own site before the budget conversation, our free visibility assessment shows which pages need work and which buyer questions you are absent from.

What else do people ask about budgeting for AI search visibility?

The five questions below about budget come up when a marketing lead has to name a number in a planning cycle and defend where it came from.