The 2026 AI Search Visibility Study
Several interviews in, the pattern is not what anyone expected.
We have been interviewing marketing and digital leaders at large enterprises about how their brands show up in AI answers. Same questions every time. The answers have been sharp, and they have contradicted each other more often than they have agreed.
30-minute conversation · Anonymous option available
Several interviews in, there is still no consensus on where to spend. Two of the sharpest examples sit at opposite ends of it, and the public citation data has started to take a side.
Fortune 500 industrial brand
Every model names Reddit as its largest source, so they went after it. With a dozen product lines, showing up authentically was not workable, and the platform does not tolerate corporate presence. What actually moved their visibility was the thing they said they never would have predicted: traditional PR. Press releases. Conversations with editors. Getting articles placed.
Fortune Global 500 CPG brand
A deliberate move away from broad backlink tactics toward reviews, forums, recommendation engines, and user-generated content. Exactly the channels the other brand walked away from, and channels their category had never taken seriously before.
Neither of them can tell you who is right.
One built their own tracking. Hundreds of test queries across five categories, measured against four different AI models. They still cannot connect a change they made to a result they saw. Another set a baseline in Q1 and is waiting on Q3 numbers, because the feedback loop runs three to six months.
So the most measured team in the study still cannot measure the thing that matters. Which means these bets are being placed on instinct, and the instincts point in opposite directions.
What the public data says
Between August and December 2025, YouTube's share of social citations in AI answers rose from roughly 19% to 39%. Reddit's fell from roughly 44% to 20%. A separate analysis of about 5.35 million citations across eight models found earned and news sources accounted for 39.5% of all citations, and growing.
Which means the team that abandoned Reddit and moved to earned media moved toward the channel that is gaining. The team scaling into Reddit is scaling into one that has lost more than half its citation share in under a year.
We are not calling the question on a handful of interviews. Channel-level citation share is not the same as category-level visibility. But it does reframe the disagreement, and it is the kind of thing you would want to know before your next planning cycle.
The ground moved about six months ago.
Major social platforms opened up to be indexed by generative search far more deeply than before. One team's recent audits found YouTube surfacing at or near the top close to 30% of the time in their categories. Most channel strategies were set against a different index.
A number-one ranked page can be invisible in an AI answer.
One participant's flagship brand ranks extremely highly in traditional search in its home market and did not surface at all in an adjacent one. A competitor did. The cause was internal inconsistency in how the brand describes itself across regions, not competitive pressure.
Nobody can prove it is working.
Every leader we have spoken with is investing without being able to connect an action to a result. The measurement gap has been unanimous, including for the team with the most rigorous tracking in the study.
Brand language built over decades is now working against large companies.
Models respond as though the exchange were between two people. Real prompts are blunt and plainly worded. The more distinctive a brand's vocabulary, the less likely it is to be the answer.
AEO, or answer engine optimization, is the question of how a brand stays visible as customers move from typing searches to asking AI assistants. Some teams call it GEO. Some just call it getting cited in ChatGPT. Almost nobody has a settled playbook for it yet.
Meridian Advisory is interviewing senior marketing, brand, and digital leaders at Fortune 2000 companies about what they are actually seeing, what they have tried, whether they can measure any of it, and what is funded for 2026. Every interview is 30 minutes and follows the same structure. Responses are aggregated and anonymized, then published as a single benchmark report.
Participants so far include senior leaders at a Fortune 500 industrial manufacturer, a global CPG brand, a global pharmaceutical manufacturer, the international division of a Fortune 100 healthcare and nutrition company, and a brand and content strategist who spent eight years leading owned content at a major US telecom.
Not vendor theory. Peer reality.
Prior to the group roundtable discussion, each participant sits for a 1:1 interview with Meridian Advisory to share your perspective on the same key themes around AEO. Once the study concludes and the interviews are finalized, everyone who contributed is invited to a closed-door virtual roundtable. Contributors only. We bring the aggregated findings to the group, anonymized, and everyone gets to compare notes directly with each other.
The roundtable is closed to anyone who did not contribute an interview. No vendors pitching, no audience.
Contributors receive the complete benchmark report ahead of public release.
Hear different perspectives, strategies, and bets to help shape your own internal AEO strategy.
How AI-driven discovery is showing up for your brand today, in your traffic and in what surfaces when someone asks about your category. Where it is real, and how much has changed in the last year.
The content, technical, and structural moves your team has made, and an honest read on what moved the needle versus what did not.
Whether you can measure your presence in AI answers yet or it is still a black box, and where this sits inside your organization.
What plays are you setting up for the back half of the year?
Meridian runs structured peer research with senior leaders and turns those conversations into published reports their peers actually read. The conferences and the analyst reports capture the polished version. What is actually being scoped, funded, and quietly killed inside Fortune 2000 marketing organizations rarely leaves the room. That is the gap we work in.
Gary is the founder of Meridian Advisory and runs every interview in this study personally. If you book a time below, you are booking it with him.
Connect on LinkedIn
Adobe and WPP are the joint sponsors of this study. Adobe builds the creative, content, and experience technology marketing teams use to reach and convert customers. WPP is the world's largest marketing and communications group. They sponsored this research because the question of how brands stay discoverable in AI answers is the question their clients are asking, and they want a peer-sourced view of it rather than a vendor narrative.
This follows the Forrester TEI and IDC InfoBrief model. The research is sponsored by technology partners and conducted independently by Meridian Advisory. Adobe and WPP do not write the questions, do not attend the interviews, and do not decide what the findings say.
Nothing you share is attributed to you, your name, or your company without your explicit written approval. You review any direct quote tied to your name or company before it appears anywhere.
Anonymized findings may appear in the published report and in related commentary, described only by role and industry, for example "a digital leader at a global pharmaceutical manufacturer." This follows the same convention as Forrester and IDC research.
If you choose anonymous participation, your name and company are removed from all materials and you appear only by role and industry descriptor, for example "VP of Marketing, Fortune 500 retailer". We do not sell, share, or repurpose your interview content beyond the published report and the anonymized findings shared with contributors.
30 minutes, max. A structured conversation about what is actually working in AI-driven discovery right now. Anonymous participation available.
Questions: garyburns@themeridianadvisory.com