SEO Budget or AEO/GEO: Where Should You Invest Now?

The three-layer budget: keep, add, defund

This framework is Tideflow's editorial synthesis. Each line in it rests on the evidence cited in the sections below.

Why cutting SEO to fund GEO backfires

On Google, AI visibility and SEO share one foundation. Google states that SEO best practices "continue to be relevant because our generative AI features on Google Search are rooted in our core Search ranking and quality systems" (Google Search Central).

The eligibility rule is explicit. To appear as a supporting link in AI Overviews or AI Mode, a page "must be indexed and eligible to be shown in Google Search with a snippet," and "there are no additional technical requirements" (Google Search Central). If you pull budget from indexing, site health, or content quality to pay for a separate GEO program, you weaken the thing that makes you eligible for Google's AI answers in the first place.

This guidance has one limit. It covers Google only. How ChatGPT, Perplexity, and Gemini retrieve sources, and how closely their choices track organic rankings, isn't documented at the same level. You should measure that for your own queries rather than assume it. For a deeper look at where the two disciplines overlap, see Is AEO Really a New Discipline?

What has actually changed: clicks shrink, answers grow

The click is worth less on AI-summary queries, but being in the answer matters more. These are two separate facts, and your budget should respond to both.

Start with the clicks. Pew Research Center tracked real browsing by 900 US adults across 68,879 Google searches in March 2025. When an AI summary appeared, users clicked a traditional result in 8% of visits. When no summary appeared, they clicked in 15% of visits. They clicked a link inside the summary itself in just 1% of visits (Pew Research Center).

Google disputes this. It called the study "a flawed methodology and skewed queryset" and says it has "not observed significant drops in aggregate web traffic" (Ars Technica). The data is also correlational, US-only, and a one-month snapshot. Even so, it is the best independent behavioral evidence available. It also points in the direction a careful team should plan for.

Now the reach. Google reports that AI Overviews has more than 2.5 billion monthly active users and that AI Mode has passed one billion monthly users (Google). These figures are self-reported. They still make one conclusion hard to avoid: "SEO is dead" is wrong, and so is treating AI answers as a side channel.

The practical implication is that a brand can lose clicks without losing visibility, but only if it is named in the answer. Being included in the answer and getting the visit now need to be planned and measured as separate outcomes.

Estimate your exposure before you reallocate

How much AI answers matter to you depends on what your buyers search for. In Pew's sample, only 18% of Google searches produced an AI summary. That share rose to 53% for searches of 10 or more words and 60% for queries starting with who, what, when, or why. It fell to 8% for one- or two-word searches (Pew Research Center). Google has kept expanding these features since March 2025, so current prevalence is likely higher.

You can turn this into a quick test:

1. Export your top 100 or so priority queries from Search Console, meaning the ones tied to pipeline, not just traffic.
2. Tag each query as short or head-term (roughly 1–3 words), long (10 or more words), or question-led (who/what/why/how/best/vs).
3. Treat the share of long and question-led queries as your exposure estimate.

Here is how the result might look for two illustrative companies. A B2B software company whose buyers search "best tool for X for small teams" or "X vs Y for compliance" is heavily exposed and should run a real pilot. A retailer whose demand comes mostly from brand and short product-name searches is less exposed. It should keep SEO weighted heavily and run a lighter pilot.

Where the new money should go

The additions below target work that most SEO budgets underfund. Pay for them mainly from the defund list rather than from the foundations.

First-hand content that answers the question directly

Google says unique, non-commodity content "will likely influence your website's presence in generative AI search in the long run more than any of the other suggestions" in its guide (Google Search Central). Google's own example of commodity content is "7 Tips for First-Time Homebuyers." Its example of non-commodity content is "Why We Waived the Inspection & Saved Money: A Look Inside the Sewer Line."

In practice, this means publishing fewer, heavier pieces. Build them from something only you have: customer data, implementation lessons, pricing reasoning, or expert opinion. Put the answer in the first paragraph. If a planned article could be written by anyone with a search engine, cut it and put the budget toward one that couldn't.

There is some independent support for the idea that content changes can shift inclusion. The foundational GEO study found that content-level optimization could increase visibility in generative engine responses "by up to 40%," and that results varied by domain (Aggarwal et al., KDD 2024). Treat that figure as a ceiling measured on a research benchmark, not a lift you should expect. The domain variation is the useful lesson: test what works in your own category.

Authentic presence in the sources AI already cites

AI answers draw heavily on third-party sources, so off-site presence deserves its own budget line. In Pew's data, Wikipedia, YouTube, and Reddit together made up 15% of sources cited in AI summaries and 17% of sources in standard results. Government (.gov) sites made up 6% of AI-summary sources, compared with 2% of standard results (Pew Research Center). Google also notes that its AI features can show what's said about products in blogs, videos, and forum discussions.

There is a firm limit on this work. Google warns that "seeking inauthentic 'mentions' across the web isn't as helpful as it might seem" (Google Search Central). The budget should go to things that create real presence:

• product demos on YouTube
• genuine participation in communities where your buyers ask questions
• review programs
• well-reasoned pitches to the roundups that already rank for your category

It should not go to seeded posts. Keep in mind that the citation data shows what gets cited. It doesn't prove that off-site spending causes citations. For the tradeoffs of chasing roundup inclusion, read Are Third-Party Listicles Good GEO or Just AI-Era Link Building?

Measurement that separates the signals

A single "AI visibility score" can't tell you whether spending is working. You need to track each signal on its own, as shown in the table below.

The distinctions between these signals are ones Tideflow draws in its own product documentation. That documentation notes that "a crawler request does not prove indexing or citation," and that referrals depend on identifiable source information, so not every AI-originated visit is recognized (Tideflow AI; Tideflow AI developer docs). These are first-party statements, but they describe measurement limits any team will run into.

Some numbers you would want are not available from any credible public source: AI-versus-organic conversion rates, how much AI answers vary across repeated prompts, and how closely AI citations overlap with top organic rankings. Measure these in your own data instead of borrowing someone else's.

What to stop funding

For Google, several popular AEO/GEO line items can be dropped. Google says:

• You don't need "new machine readable files, AI text files, markup, or Markdown" to appear in its generative AI features, because Google Search doesn't use them. This includes llms.txt.
• Structured data "isn't required for generative AI search," and no special schema is needed. Google still recommends keeping existing structured data, so this doesn't mean you should rip it out.
• Chunking content or rewriting pages purely for AI isn't necessary.
• Creating pages for query fan-out variations "primarily to manipulate rankings or generative AI responses" violates its scaled content abuse policy.

All four points come from Google's optimization guide. Other engines may treat files like llms.txt differently, and there's no solid public evidence either way. That makes these files a cheap experiment at most, not a budget priority.

The largest source of reallocation money for most teams is commodity content volume. A monthly quota of generic "tips" articles is exactly what Google says it doesn't reward, and it is usually the biggest line in the content budget.

A staged reallocation playbook

The steps below are ordered by dependency and cost, not by proven impact.

1. Baseline Google for free. As of August 31, 2026, Search Console provides AI-feature impression metrics worldwide, along with which pages appear in AI responses and in which countries. AI Overviews and AI Mode clicks are counted within the "Web" performance data (Google Search Central). Record your current AI-feature impressions for your priority pages.
2. Baseline the other engines. Pick 20–50 prompts your buyers would realistically ask. Run them in ChatGPT, Perplexity, and Gemini. Record whether you are mentioned, whether you are cited, which competitors appear, and which sources the engines cite. Repeat each prompt more than once, because a single run can mislead you.
3. Fund a time-boxed pilot from the defund list. Redirect the budget from a few commodity articles into a small set of first-hand answer pieces and one off-site effort aimed at a source the engines already cite for your category.
4. Read the signals in order. Check mentions and citations first, then crawler visits, then referrals, then conversions. Early movement in mentions with no referrals is still informative. If nothing moves after a reasonable run, stop or change the tactic.
5. Scale only where evidence shows up. Increase spend where your query mix is question-heavy and the pilot moved the signals. Hold steady where it didn't.

How this plays out differs by audience. A startup can often run the whole pilot from reallocated content hours. An established brand will usually find its biggest gap is off-site: roundups and communities that describe it inaccurately or not at all. An agency should run the step 2 baseline per client before proposing any split, because exposure varies widely by category. For a hands-on version of steps 2–3 for ChatGPT, see How to Get Mentioned by ChatGPT.

Running the loop without five separate tools

The hard part of this playbook is keeping the steps connected. Monitoring prompts across engines, turning a missing mention into a specific piece of content or an off-site target, and then tracing whether that work produced crawls, referrals, and conversions usually means stitching together several tools and spreadsheets.

Tideflow AI is built to run that loop in one place. By its own description, it:

• monitors prompts across ChatGPT, Gemini, Perplexity, Google AI Mode, and Google AI Overviews for mentions, citations, competitors, and cited sources
• turns each gap into a plan to improve a page, create content, or pursue inclusion in sources AI already cites
• drafts content that you review and approve before publishing through MCP into your own stack
• logs AI crawler visits and ties AI referrals to conversion events

These are first-party claims. Tideflow is also explicit that neither monitoring nor publishing guarantees a citation, recommendation, or ranking. The platform is currently in an early-access stage; see Pricing & Early Access for current terms. To compare it with alternatives, see Tideflow AI vs Profound and Tideflow AI vs Peec AI.

Frequently Asked Questions

Should I opt out of Google's AI features to protect my clicks?

For most sites, no. Google's new Search Console control lets you opt out of generative AI features, but opted-out sites receive no AI-feature traffic or impressions (Google). Opting out trades away visibility in surfaces used by billions of people each month in exchange for a click-through gain that isn't guaranteed. Only consider it if your business model depends on the click itself, such as ad-supported publishing, and you've measured the tradeoff.

Does Google's guidance apply to ChatGPT and Perplexity?

Not directly. Google's statements cover Google Search only. Strong first-hand content and authentic third-party presence are reasonable bets across engines, but the retrieval behavior of ChatGPT, Perplexity, and Gemini isn't documented in the same way. That's why step 2 of the playbook baselines those engines separately.

Is there a benchmark percentage to move from SEO to AEO?

No credible source supports one. The most defensible approach is to fund the AEO pilot from low-value SEO spend, such as commodity content and AI "hacks," rather than from foundations. Then set future allocation based on your query exposure and what the pilot's signals show.

Your next step

This week, pull your priority queries from Search Console, tag the long and question-led ones, and record your AI-feature impressions. That baseline tells you how exposed you are before you move any money. Then run a small prompt audit across the other engines, fund one pilot from your commodity content budget, and judge it signal by signal rather than by a single score.

Sources

• Google Search Central, Google's Guide to Optimizing for Generative AI Features on Google Search; AI Features and Your Website
• Athena Chapekis, Pew Research Center, Do people click on links in Google AI summaries? (July 2025)
• Ars Technica, Surprising no one, new research says AI Overviews cause massive drop in search clicks (July 2025)
• Google, New opportunities, control and insights for website owners (June 2026)
• Aggarwal et al., ACM SIGKDD, GEO: Generative Engine Optimization (2024)
• Tideflow AI, Tideflow AI vs Profound: Which AI Visibility Platform Should You Choose?; Add analytics and track views and conversions; Tideflow AI — AI Visibility & Content Platform