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

Why this feels like a binary choice, even though it isn't

The pressure is real. When Google shows an AI Overview above the organic results, Ahrefs found that clicks to the ranking page drop by 58% on average, and Seer Interactive's own tracking puts the decline in the 40-65% range depending on query type and month measured (Ryan Law, Xibeijia Guan, Ahrefs; Seer Interactive). That's not a rounding error. For a page that used to convert search visitors into leads, losing half its clicks is a direct revenue problem.

But zero-click search isn't new, and it didn't start with ChatGPT. SparkToro's clickstream analysis found that for every 1,000 Google searches in the US, only 360 end in a click to the open web; in the EU, it's 374 (SparkToro). Featured snippets, local packs, and knowledge panels had already been eating into clicks for years before generative answers arrived. What's different now isn't that clicks are disappearing; it's that the mechanism has changed from a snippet pointing at your page to a synthesized paragraph that may or may not cite your page at all.

The other detail that gets lost in "search traffic is collapsing" headlines: both the Ahrefs and Seer declines are concentrated in informational and educational queries, the kind where a summarized answer genuinely satisfies the searcher. Transactional, local, and branded searches are far less affected, because people comparison-shopping or looking for a specific business still tend to click through. If your query mix skews commercial, a blanket "cut SEO in half" reaction would overcorrect for a risk you don't actually carry at that scale.

The model that tells you which budget line is actually underfunded

The clearest way to think about the SEO/GEO split isn't "old tactic vs. new tactic." It's two different stages of the same pipeline, and each stage responds to a different kind of investment.

Academic research on generative engines describes the process in three steps: the engine first retrieves a set of candidate sources, much like a search ranking algorithm would; then it summarizes and selects from those candidates; then it responds with a synthesized answer, sometimes with citations attached (Aggarwal et al., GEO: Generative Engine Optimization, 2024). Classic SEO's job is getting you into that retrieved set in the first place. GEO's job is getting you selected, quoted, and cited once you're already a candidate.

This is why "we already do good SEO, so AI will notice us" doesn't hold up. Being retrieved doesn't guarantee being cited. In the same research, the authors tested specific edits to existing, already-ranking pages: adding statistics, direct quotations, and citations to sources. Those changes increased the page's visibility inside the generative engine's synthesized answer by up to 40% in controlled benchmark testing, without changing the page's actual Google ranking. That's a meaningful signal that GEO tactics are additive, not a rebrand of SEO wearing a new label.

Two qualifications matter here. First, that 40% figure comes from a benchmark test (the researchers' own "GEO-bench" evaluation), not a measured real-world traffic or revenue outcome. "Visibility inside an answer" is a distinct, engine-defined metric from clicks or conversions, and there's currently no single agreed way to measure it across the industry. Second, this only works on content that's already retrievable. If your page isn't well-structured, isn't authoritative, or isn't indexed, no amount of added statistics will get it selected for an answer it was never a candidate for.

What's actually proven, and what's still speculation

The GEO/AEO conversation right now has two failure modes: dismissing it as unproven hype, or treating it as a settled discipline with 15 years of best-practice research behind it, the way classic SEO has. Neither is accurate, and being precise about which claims have real evidence behind them matters more than picking a side.

Content-level tactics — adding statistics, quotes, and cited sources to existing pages — have experimental support from the Aggarwal et al. research above. That's a real, tested lever.

Earned-media bias is a separate, also-tested finding: one 2025 study comparing how generative engines and Google's traditional algorithm select sources found that AI answers systematically favor earned and third-party media over brand-owned content, more so than Google's classic results do (Generative Engine Optimization: How to Dominate AI Search). That means digital PR, third-party reviews, and earned citations belong in the GEO conversation even though they sit outside a classic on-page SEO budget entirely.

What isn't proven yet is the downstream revenue impact of any of this. Even Seer's data showing a citation click-through advantage, a 35-91% relative lift when your content is cited in an AI answer, starts from a tiny absolute base: 0.5% to 1.2% click-through, compared to what a historical top-three organic ranking used to deliver. Winning a citation is real progress. It is not the same as winning back the clicks organic search used to send you.

Check your own numbers before you move a dollar

Industry averages are a starting point, not a budget plan, because the CTR collapse and the citation opportunity both vary heavily by query type and vertical. Before reallocating anything, run this on your own traffic:

1. Pull your top 50-100 informational queries from Search Console and check how many currently trigger an AI Overview or a similar synthesized answer. This tells you what share of your content is actually exposed to the click-suppression effect, rather than assuming it applies uniformly.
2. Compare the CTR trend on those queries against your transactional and branded queries over the past 12 months. If the decline is concentrated where the research says it should be, you have a specific, defensible number to act on instead of a borrowed statistic.
3. For your best-ranking informational pages, check whether you're being cited, not just ranked. Run the actual query through ChatGPT, Perplexity, or Google's AI Overview and see whether your domain is quoted, linked, or absent entirely, even when you rank on page one of classic search. That gap, ranking without being cited, is exactly where GEO-specific work has room to help.

That third check is the one most teams skip, and it's the one that turns "SEO is fine" into "SEO is fine, but we're invisible in the layer that's replacing half our informational clicks."

What a phased reallocation actually looks like

Once you know which of your queries are affected and how much, the budget move itself is straightforward in structure, even if the numbers are specific to your business.

Protect your core technical and authority SEO spend as a floor. Crawlability, site architecture, backlink and domain authority work, and strong on-page fundamentals aren't legacy costs to trim; they're the retrieval-stage prerequisite for GEO too. A page that isn't a good SEO candidate was never going to be a good GEO candidate either.

On top of that floor, add a GEO line sized to what your diagnostic found: budget for restructuring your highest-traffic informational pages to include stronger sourcing, statistics, and quotable claims; budget for digital PR and earned-mention outreach, since the earned-media bias finding means some of this work sits outside your existing content team's remit; and budget for engine-aware monitoring, since AI engines don't behave identically. Research comparing ChatGPT, Perplexity, and Gemini found real differences in how each weighs source diversity, content freshness, and exact phrasing, which means a single AEO checklist won't perform the same way across platforms.

Phase this over two to three quarters, moving the ratio as your own citation and CTR data confirms or corrects your initial diagnosis, rather than locking in a fixed percentage split on day one.

Measuring GEO without fooling yourself

The hardest part of this shift isn't deciding to invest. It's knowing whether the investment worked, because AI answer engines don't hand you a report like Search Console does. Their answers vary by phrasing, session, and engine, and there's no stable, comparable number the way there is for organic rank.

The fix is treating "AI visibility" as several distinct signals instead of one score:

• Mentioned — the AI names your brand in an answer, with or without a link.
• Cited — the AI directly quotes or links to your specific content as a source.
• Crawled — an AI bot has visited your page, which is necessary but not sufficient for being cited.
• Converted — a visitor who arrived via an AI answer actually took a meaningful action on your site.

Collapsing these into a single "AI visibility score" is how teams end up celebrating a metric that never touches revenue. This is the specific gap Tideflow AI is built to close: it monitors how your brand actually shows up across AI answer engines, distinguishes crawler visits from real citations using its bot detection, analyzes where your content has gaps versus what's actually getting cited, and connects that gap analysis to content you can publish directly into your existing stack through MCP, then measures the chain through to tracked traffic and conversions. It's a closed loop rather than a monitoring dashboard you still have to act on manually, and it won't tell you it can guarantee a citation or a ranking, because no honest tool can.

Should smaller brands run GEO differently than large ones?

Yes. The same research that found earned-media bias in AI search also points to a "big brand bias," where established names get cited more readily regardless of content quality, partly because AI training data and citation patterns favor domains that already have a large footprint of third-party mentions. For a niche or smaller brand, that means the earned-media line of your GEO budget matters more, not less, than it would for a market leader. Digital PR, guest contributions, review-site presence, and third-party comparisons carry more weight for a smaller brand precisely because you can't out-publish a bigger competitor's owned-content volume. Doubling down on owned blog output alone, without pursuing citations from outlets AI engines already trust, is the misstep most likely to leave a smaller brand stuck ranking well on Google while remaining invisible in AI answers.

Frequently Asked Questions

Is paid search on informational keywords still worth funding given the CTR collapse?

It depends on where the collapse is concentrated in your account. The documented CTR declines are heaviest on informational queries where AI Overviews fully satisfy the searcher's question, so paid spend chasing pure informational intent is increasingly working against a shrinking pool of clicks. Transactional and comparison-intent keywords are far less affected, since those searchers still need to click through to act. Audit your paid campaigns the same way you audit organic: separate informational spend from transactional spend before deciding where to cut.

Do ChatGPT, Perplexity, and Google AI Overviews need different tactics, or is there one unified strategy?

There's meaningful overlap, since all three still perform some version of retrieve-then-summarize and all reward well-sourced, well-structured content. But they aren't identical: research comparing the engines found differences in how much they value source diversity, how sensitive they are to content freshness, and how much exact phrasing affects whether you're cited. Build your content and technical foundation once, but expect to monitor and tune citation performance per engine rather than assuming a single checklist covers all of them.

If you're still working out where your own content and technical foundations stand before layering GEO tactics on top, the Tideflow AI platform is designed to run that diagnosis and act on it in one workflow, and the early access page has details on how design partners are using it today.

Sources

• Ryan Law, Xibeijia Guan, Ahrefs, Update: AI Overviews Reduce Clicks by 58% (February 2026)
• Seer Interactive, AIO Impact on Google CTR: September 2025 Update (November 2025)
• SparkToro, 2024 Zero-Click Search Study: For every 1,000 EU Google Searches, only 374 clicks go to the Open Web. In the US, it's 360.
• Aggarwal et al., GEO: Generative Engine Optimization (2024)
• Generative Engine Optimization: How to Dominate AI Search (2025)