Is SEO Alone Enough for Business in 2027? SEO vs GEO, and What I Would Actually Budget For

Every few months a client asks me some version of the same question: "People are asking ChatGPT instead of Googling. Should we stop paying for SEO?"

It is a fair question, and I used to give a lazy answer. Then I spent a few weeks going through the actual referral data, the citation studies, and the academic research, and I came away with a more uncomfortable conclusion than either side of the debate wants to hear. SEO is not dying. But SEO on its own is now solving only half the problem, and the half it no longer solves is the half where buying decisions get made.

Here is what I found, laid out as a straight comparison, followed by the seven things we are actually doing about it.

First, the myth: "Everyone left Google for AI chatbots"

They did not. Not yet, anyway.

According to Cloudflare Radar data for May 2026, Google sent 87.63% of all search referral traffic. Every AI search engine combined, including ChatGPT, Gemini, Claude, and Perplexity, accounted for just 0.29% of search referrals. To put that in perspective, the four big AI chatbots together send roughly one fifth of what DuckDuckGo alone sends.

So if you are measuring in clicks, the migration story is wildly oversold. Rand Fishkin of SparkToro has been blunt about why that narrative travels so far. As summarized by Lunio, media outlets gain clicks from bold predictions, agencies use fear to sell new AI solutions, and investors benefit from inflated valuations for AI-first startups.

Now, the part that should worry you

Clicks are the wrong scoreboard, and that is exactly the problem.

SparkToro's clickstream analysis, reported by Launchcodex, found that 68 percent of US Google searches ended without any click to a website in the first four months of 2026, up from about 60 percent in 2024. Put in raw terms, only 276 clicks per 1,000 US searches now reach the open web, down from 374 two years earlier.

Pew Research found the gap plainly, as compiled here: an 8% click rate when an AI summary appeared, versus 15% without it. A randomized 2026 field experiment from ISB and Carnegie Mellon researchers measured a 38% drop in organic clicks.

Meanwhile the AI side is small but compounding fast. SE Ranking's study found that website traffic from AI search engines grew 16x between 2024 and 2026, and that visitors referred by AI engines spend 68% more time on websites than those arriving from traditional organic search. Their explanation is that AI tools work as intent filters, sending people who are already deep into a decision.

Omniscient Digital's buyer research puts a number on the commercial stakes: 42% of B2B decision-makers now use an LLM in the very first step of the buying process.

Fishkin's summary of where this leaves marketers is the line I keep repeating in meetings: "traffic is a vanity metric." The influence now happens before anyone reaches your site. (via Emarkable)

SEO vs GEO: what each one actually buys you

The honest framing came from Lily Ray, VP of SEO Strategy and Research at Amsive, who describes GEO not as an overhaul or abandonment of SEO but as "a new system for competing for, capturing, and measuring success across AI platforms." (via EMGI Group)

That matches what I see in client accounts. Ranking is still the input. It just stopped being the finish line.

One more wrinkle that broke my assumptions: you cannot optimize for "AI" as if it were one thing. Orbit Media's study of 13,184 citations across four tools found that out of 1,792 query-and-domain combinations, all four models agreed on citing the same domain for the same question only 30 times. That is 1.7%. Four different engines, four different worlds.

The 7 things we are actually doing for 2027

1. We keep the SEO budget, and stop judging it by sessions

Ranking still feeds the machine. Retrieval-augmented systems pull from indexed pages, so the page that ranks is frequently the page that gets quoted. We moved our reporting to three columns: rankings, AI citations, and qualified conversions. Sessions became a diagnostic, not a KPI.

2. We rebuilt pages around the tactics the research actually validated

This is the part most "AI SEO" advice skips. The founding academic work here is the Princeton GEO study by Aggarwal and colleagues, published at KDD 2024 (arXiv:2311.09735). It found that GEO can boost visibility by up to 40% in generative engine responses, and it named which changes did the lifting.

The winners, in the authors' own testing: adding relevant statistics, incorporating credible quotes, and including citations from reliable sources produced a 30 to 40% improvement on the Position-Adjusted Word Count metric. Stylistic changes such as fluency optimization and easy-to-understand writing delivered a further 15 to 30% visibility boost.

Read that again. Cite sources. Quote experts. Add data. Write clearly. That is the whole playbook, and it is the same advice good editors have given for a century.

A useful caution before you go all in: C-SEO Bench (Puerto et al., 2025) found that most conversational SEO tactics do not help and several actively hurt, while plain source relevance keeps working. So we test rather than assume.

3. We moved money from link building to earned mentions

This one changed our budget the most. Ahrefs analyzed 75,000 brands and found that brand web mentions were the dominant predictor of AI visibility, correlating 0.664 with AI Overview visibility versus 0.218 for backlinks.

Distribution matters more than we assumed. A Stacker study found an 8% baseline citation rate for content sitting on a brand's own site, rising to 34% when the identical content was distributed through third-party news outlets. That is a 325% lift, with a March 2026 follow-up showing a 239% median lift across a broader sample. Same words. Different address. Four times the citations. (both findings compiled here)

4. We stopped pretending Reddit is beneath us

Tinuiti's Q1 2026 AI Citations Trends Report found that Reddit citation share grew at least 73% from October 2025 to January 2026 across commercial categories. For Perplexity specifically, 24% of all citations in January 2026 came from Reddit alone. (analysis here)

The catch is that you cannot fake it. AIclicks' breakdown notes that direct posting almost always backfires, and that brands earning Reddit citations legitimately do so by having team members participate under real names, sharing original research that gets organically reposted, and engaging in category-specific subreddits without link-dropping. Earned visibility takes six months or more. Manufactured visibility lasts weeks.

5. We split our content by query type, because the engines do

Omniscient's citation analysis found a clean split. When users ask what customers think, LLMs cite earned media 82% of the time, favoring TrustPilot, Reddit, and social platforms over the brand's own site. But for factual product information, owned content appears most frequently, at 50%.

So we treat our own site as the source of truth for specs, pricing, integrations, and documentation, and we treat third-party ecosystems as the source of truth for reputation. Trying to win reputation queries on our own domain is a losing fight.

6. We publish original data instead of summarizing other people's

Nothing earns a citation like a number nobody else has. Fishkin's advice on this is the cleanest version I have seen: "If you want AI tools to recommend you, be the brand they learn from." We now run one small proprietary study per quarter, put it on our site, and pitch it outward.

7. We monitor across four surfaces, not one

Because engine agreement is so low, we track prompts monthly across ChatGPT, Google AI Overviews, Perplexity, and Gemini, and we watch mentions separately from citations. A mention without a link still shapes the shortlist.

So, is SEO enough for 2027?

No. But not for the reason people usually give.

SEO is not failing because Google is collapsing. Google is not collapsing. SEO is insufficient because the decision now happens inside an answer, and answers get assembled from sources that mostly sit outside your domain. If your entire visibility strategy lives on your own website, you have optimized for the one place where the smallest share of the influence now happens.

There is also a trust story worth keeping in view. Fractl's data, compiled by Loudspeaker, shows the share of consumers who find AI more helpful than traditional search fell from 82% in 2025 to 54% in 2026, while outright skeptics grew from 3% to 17%. Adoption is running ahead of trust. That means 2027 will very likely be a hybrid year, not a replacement year.

My recommendation is unglamorous: keep the SEO, add the earned media, publish the data, write clearly enough that a machine can quote you accurately, and change what you measure. That is one system, done once, paying in two places.