Before They Buy, They Ask AI: GEO Results From 70,000+ AI Answers

Shoppers increasingly turn to AI tools like ChatGPT, Google’s AI Overviews and AI Mode to decide what to buy and where to buy it. Brands named in those answers make the shortlist; brands that aren’t are easy to miss. Here is what our 2026 GEO results show about earning that spot, starting with an ecommerce brand.

The Short Answer

GEO is producing measurable results, and focused brands can outpace far larger competitors. We tracked three client programs across 70,457 AI answers in 2026, in consumer ecommerce and B2B services. The GEO results:

• • Every client named more often by AI: mention rates rose for all three clients, by as much as 63%.
  • Higher-value traffic: an ecommerce brand’s AI visitors were worth 4.5× its average visitor, and its AI-referred revenue reached 2.7× the prior year. A B2B firm’s program-driven traffic converted at 5.7% against a 1.8% site average.
  • Top-three most-cited company website in all three markets: #1 among tracked competitors for two clients, including the ecommerce brand, in every week we measured.
  • Share of voice up 47% and 48% for our two B2B clients, the largest gain of any firm holding at least 2% share in each category.
  • 15,022 AI citations to client content, across 85 earned media placements and 197 owned pages

Why AI Search Matters This Black Friday

More holiday shoppers now use AI to research purchases, and those who click through convert at higher rates. In the 2025 holiday season, AI-referred traffic to US retail sites rose 693.4% year over year, per Adobe Digital Insights. On Black Friday 2025, AI-referred shoppers converted 38% more often than non-AI traffic. For this season, Adobe forecasts AI traffic to US retail sites will grow another 130% year over year. In August 2026, AI-referred retail visitors already generated 43% more revenue per visit than non-AI traffic.

The same shift is happening in B2B. In a March 2026 survey, G2 found that 51% of B2B software buyers now start their research with an AI chatbot more often than with Google, and buyers named AI chatbots the top influence on which vendors make their shortlist.

Generative engine optimization (GEO) and answer engine optimization (AEO) are closely related names for one goal: earning a place in those answers. Two signals matter. A mention is when the AI names your brand. A citation is when it references your content as a source. Without either, you are unlikely to make the shortlist.

GEO is not repackaged SEO, even though some agencies are selling it that way. Our industry spent the last year arguing whether GEO and AEO were just SEO. Now that the difference is clear, many companies are simply repackaging SEO as GEO and AEO.

In my view, a few differences matter most:

1. 1. GEO calls for a human-first approach to prompt engineering and targeting. Business alignment is key, and if you don’t have a well-validated approach, you are likely to be targeting and measuring the wrong data.
   2. Earned media and off-page optimizations are paramount. You cannot win in AI search by only optimizing your own website. Owned media is your foundation, but if you don’t have a validated earned media approach, you are missing a fundamental component of your GEO/AEO strategy.

GEO Results at a Glance

Three clients, three very different markets, and measurable gains in each.

AI tracking windows: outfitter, May to September 2026; search firm, March to August 2026; ERP partner, March to September 2026.

Pattern 1: GEO-Driven Traffic Is Smaller, but Higher Quality

Most searches now end without a click: 68% of US Google searches ended without a click in early 2026, per SparkToro. In our data, the visitors who do click through are worth more:

• Rafting outfitter: AI-referred revenue reached 2.7× the prior year while AI sessions grew 56%. Each AI visitor was worth 4.5× the average site visitor.
• ERP partner: Program-driven traffic (optimized content, earned placements and AI referrals) converted to leads at 5.7%, against a 1.8% site average. In organic search, its optimized solution and insight pages converted at 9.9%, versus 0.55% for the largely branded homepage.
• Boutique search firm: AI-referred visitors showed a 58.5% engagement rate, more than twice the rate of direct traffic (25.3%).

That matches what I see across our programs: AI visitors arrive further down the buyer’s journey. They do less window shopping and more serious consideration, because they’ve already done the research.

Treat AI referral figures as a floor. Many assistants do not pass a referrer, and buyers who see a brand in an AI answer can also search for it or type the URL, so those visits record as organic or direct. For holiday sellers, the takeaway is simple: judge AI by revenue per visit, not session volume.

Pattern 2: Focus Beats Scale

In both B2B categories, a focused specialist out-gained global incumbents many times its size.

The boutique search firm competes with the largest global executive search firms, each built on decades of brand equity. Its AI share of voice rose 47%, the strongest gain of any firm holding at least 2% share, while five competitors lost share. On citation rate, it climbed from 7th of 14 to 3rd and is now cited more often than four of the five global incumbents.

The ERP partner competes with global consultancies and systems integrators for Dynamics 365 work. It posted the largest share-of-voice gain of any firm holding at least 2% share (+48%), while five of the ten other tracked firms lost share. It moved from 6th to 4th on mention rate and ranked first on owned-site citation rate on all 26 weekly readings, with more than double the nearest competitor’s citation rate across the full window.

The incumbents still lead on raw brand mentions, built over decades. But on citations, which show whose content an answer relies on, both specialists already rank ahead of most of the giants.

For smaller companies, owning their own backyard is often how they outperform larger ones: hyper-targeting what they do best and where their real strengths are.

Depth has to come before breadth.

Pattern 3: Citations and Mentions Rise Together

In all three programs, mentions rose alongside citation gains: in citation share for the outfitter, in citation rate for the search firm, and in citations to program content for the ERP partner. That is a correlation, not proof of cause, but the pattern held in every market.

• Rafting outfitter: Its share of outfitter citations rose 14% (36.8% to 41.9%), and its mention rate rose 21%. 89% of citations to its earned placements now come in answers that also name the brand, up from 68%.
• ERP partner: The share of AI answers citing its content, owned or earned, rose from 1.6% at the first reading to 20.9% at the latest. On the 35 prompts common to both periods, on the same engine, that was a 3.7× gain. Brand mentions arriving alongside its cited content rose from 79% to 89%.
• Boutique search firm: Citation rate rose 74% from April to August (8.1% to 14.1%), while its mention rate climbed 49% from March.

This held even as AI engines became pickier. In the outfitter’s market, the number of distinct domains cited fell by roughly half over the program. It stayed the most-cited outfitter on every weekly reading.

My view on why: buyers still move through the full journey, but now it happens inside a conversation, over multiple turns, instead of across a collection of web pages. Citations bring your brand and content into that conversation. Cover each stage, and you’re more likely to earn the mention, and ultimately the recommendation, at the end.

I’d caution against chasing citations for their own sake, though.

A citation on its own has very little value if it’s not tied to a distinct buyer’s journey. These GEO results came from content that mapped to how real buyers research and decide, not from chasing volume.

What It Took: The Human-First GEO Framework™

Every program started with the buyer, not the keyword: who they are, what is driving the decision, and what they need to trust before they choose. That understanding shaped what we tracked and what we published.

At the center of all of this is a human with a human need. Even when AI connects buyers and sellers, that need, and the solution that serves it, has to be the foundation of GEO and AEO strategy.

Activation ran on the same three pillars for all three clients:

You can’t win at GEO and AEO simply by optimizing your own website. You have to go out to the wider web.

The broader data agrees. Muck Rack’s May 2026 analysis of more than 25 million links cited by ChatGPT, Claude and Gemini found that 84% came from earned sources, while paid and advertorial content made up just 0.3%.

In each program, the single most-cited asset was built around a question buyers ask before they choose. A how-to-choose guide led for the outfitter (561 citations). A comparison page led for the search firm (1,078). For the ERP partner, an earned media placement aligned to buyer questions led (658).

What to Do Before (and After) the Holiday Rush

In our programs, gains are built over months, not days, so use this season to benchmark. Then build for the next one.

1. Ask the AI yourself. Run the 10 to 20 questions your buyers actually ask through ChatGPT, Google (check for AI Overviews), and Google AI Mode. Note who gets named and whose pages get cited.
2. Publish the answer, not the pitch. Comparison pages and how-to-choose guides were the most-cited owned pages for two of our three clients, and pages on cost and timing ranked among the top performers.
3. Earn visibility beyond your own site. Earned media aligned with the questions your customers ask produced 5,697 citations across our three clients, building credibility, earning AI mentions, and driving recommendations.
4. Open the door for AI crawlers. Confirm your robots.txt allows them, check server-level access, keep pages clean and fast, and add structured data.
5. Track visibility and revenue together. Measure share of voice, mention rate, and citation rate against named competitors. Then make sure analytics records conversions, so AI visits connect to sales (keep in mind AI referral visits are likely a small sample of the total expected value).
6. Give it months, not weeks. Each program here ran about six months, and the latest figures above come from the end of that window. GEO results compound over time as content earns citations and those citations build brand authority in AI responses.

If your current program has no clear plan for prompt engineering, prompt tracking at scale, and earned media, it’s worth reevaluating your approach to GEO and AEO.

How We Measured

• Scope: Three CSP Agency GEO programs, March through September 2026, with 100 tracked buyer prompts per client.
• Volume: 70,457 monitored AI answers across engines including ChatGPT, Gemini, Copilot, Google AI Overviews, and Google AI Mode.
• Competitive metrics: Share of voice, mention rate, and citation rate are measured against every tracked competitor in each category. Share-of-voice gain comparisons cover firms holding at least 2% share. Definitions and comparison periods differ by client, as noted below.
• No averaging: Because windows and baselines differ, we report growth per client and never average it across clients.
• Baselines: The outfitter’s tracking began after its program was live, so its gains understate the full effect; its mention rate compares the first four weekly readings with the latest four. The search firm’s citation growth is anchored to April, after its prompt set expanded, and its mention rate is a share of answers naming any tracked firm, March vs. August. The ERP partner’s share of voice and mention rate compare the first four weeks of its final prompt panel with the latest four, on non-branded prompts.
• Citation counts: The ERP partner’s citations are normalized to one standard weekly reading, so they are lower than raw platform totals.
• Business outcomes: Traffic, conversion, and revenue come from each client’s Google Analytics 4 property. AI referral figures are a floor, because many assistants do not pass a referrer.
• Confidentiality: Client and competitor names are withheld.

See Where Your Brand Stands in AI Search

I started building our GEO program in July 2025 because I saw the impact generative engines were going to have on search. I believed search would move to a conversational interface, and that despite AI’s inaccuracies, people would trade some accuracy for the convenience of that experience.

Your customers are asking AI who to buy from. Find out whether it answers with you. No dog-and-pony pitch, just an honest analysis from a GEO expert. Start the conversation at csp.agency.

About the Author

Chris Rodgers is CEO of CSP Agency. He has worked in digital marketing and SEO since 2005 and began building CSP’s GEO program in July 2025. Since then, CSP has developed its own framework and proven it across GEO programs in B2B services and ecommerce.

Sources

1. AI traffic surges across industries, retail sees biggest gains | Adobe for Business
2. 2026 U.S. Holiday Forecast | Adobe Digital Insights
3. G2 2026 AI Search Insight Report: The Answer Economy | G2
4. In 2026, Less than One Third of Google Searches Still Send a Click | SparkToro
5. What Is AI Reading? | Muck Rack Generative Pulse
6. CSP Agency client tracking data and Google Analytics 4 records, 2026 (client names withheld)