Splitting a Client's Content Budget: SEO, Google's AI Overviews, and the Assistants

Three acronyms arrive in planning meetings and compete for the same content budget. They are not alternatives, and treating them as a choice is the fastest way to build a plan you cannot defend six months later.

They map to three different moments: traditional organic results (SEO), Google’s AI summary blocks (AIO), and standalone assistants like ChatGPT, Perplexity and Claude (GEO). McKinsey’s October 2025 research found half of US consumers surveyed now deliberately use AI-powered search, with $750 billion in US revenue projected to flow through it by 2028.¹

Treat the revenue and conversion figures throughout as market context for planning. They are category forecasts, not outcomes anyone can commit to on a client’s behalf, and client-specific attribution depends on the client’s own analytics. Every statistic on this page carries a footnote to its source; the budget splits further down are our starting positions, not measured data, and are labelled as such.

	SEO	AIO	GEO
Where it happens	Google results page	Google AI Overview block	ChatGPT, Perplexity, Claude, Gemini
Discovery layer	Ranked results	Summary above results	Conversational answer
What you measure	Rankings, organic sessions	Inclusion in the overview	Presence in answers, share of mentions
Content signal	Authority, links, keywords	E-E-A-T (Google’s experience-expertise-authoritativeness-trust standard), 50-70 word summaries	Answer capsules, attributed statistics, entity density
Volume	High (about 14 billion Google searches a day²)	Growing (AI Overviews on 13.14% of tracked queries by March 2025³)	Growing fast from a small base
Competition	Saturated	Moderate	Low: most competitors have published nothing that answers the question directly
Time to first signal	3-6 months	4-8 weeks	Days to weeks on freshness-weighted platforms

SEO: still the volume, with a shrinking yield

SEO is improving a site’s visibility in traditional results through keyword targeting, technical health, content relevance and links.

It remains the highest-volume channel. Google handled more than five trillion searches in 2024, about 14 billion a day,² and still holds just under 90% of global search share.⁴ Organic search drives 48.5% of website traffic against under 1% from AI assistants, in SE Ranking’s study of nearly 64,000 sites.⁵ Any plan that abandons it is wrong.

The problem is yield rather than volume. Where an AI Overview appears, organic click-through fell from 1.76% to 0.61% between June 2024 and September 2025, a 61% decline, with paid down 68%.⁶ Ahrefs, on a separate 300,000-keyword sample, puts the click-through loss for the top-ranking page at 58%.⁷ AI Overviews went from 6.49% of tracked queries in January 2025 to 13.14% by March.³ The client earns the ranking and receives progressively less for it.

Where SEO still wins: high-volume transactional queries, product and category pages, informational content with genuine click intent, and any context where buyers deliberately browse multiple options rather than accepting one recommendation.

AIO: inside Google, by different rules

AIO (AI Overviews optimisation) is structuring content to be cited inside Google’s AI summary blocks above organic results.

It sits between the other two: Google’s ecosystem, different selection logic. Domain authority alone does not predict inclusion. A site ranking fifth organically may be cited while the top-ranked site is not, which is the single most useful fact to have ready when a client asks why their number-one position is not appearing in the summary.

It rewards 50-70 word summary answers near the top of a page, FAQPage and HowTo schema, E-E-A-T signals, and longer queries: questions of eight words or more are 7x more likely to trigger an overview.³ Brands cited in AI Overviews earn about 35% more organic clicks than uncited brands on the same results page.⁶

Where AIO wins: comparison queries, definitional content, and “how do I choose” material where the buyer wants an authoritative answer before committing to a click.

GEO: the uncrowded one

GEO is structuring content so assistants outside Google’s ecosystem cite it in generated responses.

The structural difference is that a cited brand is not one of ten options. It is the recommendation. This is why Semrush’s June 2025 research found AI-referred visitors worth 4.4x organic:⁸ they arrive post-recommendation rather than mid-research.

The opportunity is that almost nobody is doing it deliberately. In one December 2025 vendor benchmark, 87% of US businesses ranking on Google’s first page did not appear in AI-generated answers at all.⁹ For an agency, that combination - high buyer intent, low competitive density - is unusual and will not persist.

The Princeton GEO study (KDD 2024) identified what actually moves citation rates: quotation addition (+41%), statistics addition (+33%), citing sources (+28%). Keyword stuffing decreased visibility by 9%.¹⁰ Every one of those is an editorial instruction, which is why this line item belongs in a content budget rather than a technical one.

Where GEO wins: consideration-stage questions, “best X for Y” requests, comparative research, and any category where buyers delegate shortlisting to an assistant.

How the three map to a buying journey

Stage	Primary layer	What the buyer does
Framing the problem	GEO	”What should we be looking at for [situation]?”
Comparing options	AIO / GEO	”How does [A] compare to [B] for [use case]?”
Validating a choice	SEO / paid	”[Vendor name] pricing / reviews / implementation”
Expanding after purchase	GEO	”How do we get more out of [product]?”

A ranking is not a citation. SurferSEO’s study of 173,902 URLs found that 68% of the pages Google’s AI Overviews cite are not in the top ten organic results for the same query.¹¹ Yext’s analysis of 6.8 million citations found 86% come from sources brands already control: 44% from their own websites and 42% from their business listings.¹²

Those two statistics reframe the work. Rankings alone do not put a client into the answer, but the material the assistants read is overwhelmingly material the client can commission.

Four allocation shapes

Optimal splits depend on how the client’s buyers actually behave, not on category convention. These are starting positions to argue from, expressed as proportions of the content and search budget rather than fixed sums. They are our recommendations, not measured data.

B2B software

Forrester’s 2024 Buyers’ Journey Survey found 89% of B2B buyers had adopted generative AI in their research, roughly three times the rate seen in consumer markets;¹³ by the 2025 survey it was 94%.¹⁴

Layer	Share	What it buys
SEO / content	30%	Pillar content, comparison pages, integration and use-case pages
GEO	25%	Answer capsules for category questions, attributed thought leadership, community presence
AIO	20%	Feature comparison schema, FAQ content per use case, credentialed bylines
Paid	15%	Bottom-funnel branded and competitor terms
Experimental	10%	Podcast and newsletter citations, emerging surfaces

Professional and specialist services

Buyers here use assistants to build a shortlist before making any contact. The purchase is high-stakes, infrequent and research-intensive - the exact pattern where assistant recommendations carry most weight.

Layer	Share	What it buys
GEO	35%	Answer-shaped thought leadership, proprietary data, named-expert entity building
SEO / content	30%	Practice and specialism pages, case studies
AIO	20%	FAQ schema per service area, comparison content, expert quotes
PR / third-party	15%	Trade coverage, industry association content, conference presence

Considered-purchase ecommerce and manufacturing

In these categories a large share of what assistants cite is review sites, publishers and listings rather than the brand’s own pages, which is why the split leans on schema, buying guides and third-party presence.

Layer	Share	What it buys
SEO / paid	35%	Product and category SEO, shopping, brand terms
AIO	25%	Product schema, buying guides, 50-70 word summary blocks
GEO	20%	Coverage in AI-indexed publications, review-site and community presence
Social / other	20%	Platform-native discovery

Multi-brand or portfolio clients

Where one client owns several brands or product lines, the constraint is usually editorial capacity rather than budget. The allocation matters less than the sequencing: instrument every brand, then concentrate production on the two or three where the assistants currently return a competitor’s name for high-value questions. Spreading production evenly across a portfolio is the most common way this budget gets wasted.

The reallocation argument, honestly stated

For B2B and professional services specifically, moving 15-20% of an SEO budget toward GEO is defensible on four grounds.

Competitive density. A client competing for a saturated commercial keyword faces hundreds of well-resourced rivals. The same client competing to be named in an assistant’s answer to the same buying question faces a category where most competitors have published nothing addressing it directly.
Buyer stage. AI-referred visitors arrive post-recommendation. The 4.4x differential⁸ means a smaller volume can be worth modelling carefully against a larger one - with the client’s own numbers, not a benchmark.
Feedback speed. Freshness-weighted platforms surface new material quickly, so a programme produces evidence in weeks rather than quarters. That matters more for retaining a client than it does for the client.
Accumulation. A brand cited consistently across multiple AI-read sources builds a position that is progressively harder to displace.

It is a portfolio adjustment, not a replacement. The most practical version is not new spend at all: it is restructuring content already being commissioned - answer capsules, attributed statistics, comparison sections, direct claims - so the same budget produces material that works in both channels.

Measuring each layer

Measure	What it tells you
Presence in answers	Share of target buying questions where the client is named
Share of mentions	Client versus competitors across the same question set
Source attribution	Which pages assistants actually draw from
Framing	How the client is characterised when named
Published-work performance	Readers, click-throughs to the client’s site, last assistant fetch

The last row is the one most measurement frameworks omit and the one that most changes what gets commissioned. Knowing which published pieces are being read, which are sending people through to the client’s site, and when the assistants last fetched them tells you what the category actually responds to. Combined with the gap data, it means next month’s brief is decided by two kinds of evidence rather than by a planning instinct. A named editor - yours or your client’s - still makes the call, but they make it holding something.

Frequently asked questions

What is the difference between the three? SEO targets rankings on Google’s traditional results. AIO targets Google’s AI summary blocks above them. GEO targets citations inside standalone assistants like ChatGPT, Perplexity and Claude.

Which comes first? Maintain the SEO investment and add a 15-20% GEO allocation. It is the least contested high-intent channel currently available.

Does GEO work damage SEO performance? No. Content structured for AI citation also improves E-E-A-T signals and snippet eligibility. The techniques are additive.

How does AI search affect click-through rates? Where a Google AI Overview appears, organic click-through fell from 1.76% to 0.61% in Seer Interactive’s 15-month study,⁶ and by 58% for the top-ranking page in Ahrefs’ 300,000-keyword sample.⁷

Do AI-referred visitors convert better? Semrush found the average AI-referred visitor worth 4.4x an organic one.⁸ Treat it as a market benchmark; client-specific conversion has to come from the client’s analytics.

Which clients benefit most? B2B and professional services, because the competitive bar for being named in an assistant’s answer is dramatically lower than the bar for ranking on a contested commercial keyword.

Key takeaways

The three are not interchangeable. Different platforms, different content, different moments in a buying decision.
SEO is not dying; its yield is falling. A 61% click-through drop on AI Overview queries⁶ means clients must be cited in summaries to retain traffic they already earned.
GEO is the uncrowded layer. Most businesses that rank on Google are absent from AI answers.⁹ That will not last.
AI-referred buyers are qualitatively different. They arrive post-recommendation, which is why benchmark conversion rates are so much higher.
Allocation follows buyer behaviour, not category convention. B2B software and professional services justify the largest GEO share.
Measurement needs two signals, not one. What the assistants say now, and what the published work earned. Either alone produces a worse brief.
The strongest version of the reallocation costs nothing new. Restructure content already being commissioned so it works in both channels.

Found by AI monitors how leading AI assistants answer buying questions, produces the content that addresses the gaps, and reports on what it earns - under our partners’ brands, never in front of their clients. See how the partner programme works.

Sources

Every figure above links to the page it was taken from. Checked on 10 September 2026.

McKinsey & Company, “New front door to the internet: Winning in the age of AI search”, AI Discovery Survey of a representative US consumer panel, August 2025, published October 2025. ↩
SparkToro, “New Research: Google Search Grew 20%+ in 2024; receives ~373X more searches than ChatGPT”, 2025, drawing on Google’s own disclosure of more than 5 trillion searches in 2024. ↩ ↩²
SE Ranking, “Google AI Overviews: New Research Study”, 2025. ↩ ↩² ↩³
StatCounter, Search Engine Market Share Worldwide; reported by Search Engine Land, “Google’s search market share drops below 90% for first time since 2015”, January 2025. ↩
SE Ranking, “How Does Social Media Traffic Compare to Organic and AI in 2025?”, 63,987 websites, 2025. ↩
Seer Interactive, “AIO Impact on Google CTR: September 2025 Update”, 3,119 queries across 42 organisations, June 2024 to September 2025. ↩ ↩² ↩³ ↩⁴
Ahrefs, “Update: AI Overviews Reduce Clicks by 58%”, 300,000 keywords, December 2023 against December 2025. ↩ ↩²
Semrush, “AI Search Traffic Study”, 9 June 2025. ↩ ↩² ↩³
NEWMEDIA.COM, “New RankOS Benchmark Finds 87% of U.S. Businesses Are Invisible in AI Search Despite Ranking on Google”, press release, 22 December 2025. A vendor benchmark. ↩ ↩²
Aggarwal et al., “GEO: Generative Engine Optimization”, Proceedings of the 30th ACM SIGKDD Conference, 2024, Table 1. ↩
Surfer, “Query Fan-Out Impact”, 173,902 URLs across 10,000 keywords, 2025. ↩
Yext, “AI Doesn’t Rank, It Cites. And 86% of Its Sources Are Brand-Managed”, 6.8 million citations, July to August 2025, published October 2025. ↩
Forrester, “B2B Buyer Adoption Of Generative AI”, Buyers’ Journey Survey, 2024. ↩
Forrester, “B2B Buyers Make Zero-Click Buying Number One”, January 2026. ↩