Over the past few months, six organizations have published new research, models, and perspectives on measuring marketing performance. They come from different disciplines, including SEO, PR, analyst relations, and media measurement, and they don’t always agree. Together, though, they point to a broader shift: marketing success can no longer be measured through website traffic alone.
Rather than competing ideas, these perspectives describe different dimensions of the same problem. Comparing them side by side reveals where they overlap, where they diverge, and what marketers can learn from each as they rethink demand generation in an AI-driven, zero-click world.
Six perspectives on the same problem
According to an ancient Indian parable, a group of blind men who had never encountered an elephant decided to learn what it was like by touch.
Each touched a different part of the animal and came away with a different conclusion:
- The side was a wall.
- The tusk was a spear.
- The trunk was a snake.
- The leg was a tree.
- The ear was a fan.
- The tail was a rope.
Because each believed only his own experience, they argued rather than recognizing that they were describing the same animal.
The six perspectives in this article work much the same way. Each captures a different aspect of measuring marketing performance in AI-driven search. Together, they offer a more complete picture of how AI is changing marketing measurement.
See where your brand appears in AI search, where competitors are winning, and what it takes to become the answer AI recommends.
1. Zero-click marketing
Most Search Engine Land readers have already seen Rand Fishkin’s SparkToro post, “In 2026, Less than One Third of Google Searches Still Send a Click.” In the first four months of 2026, 68.01% of Google searches ended without a click — up from 60.45% in 2024 — and Fishkin attributes much of that acceleration to AI Overviews, now present on over 20% of searches and cutting CTR by nearly 60% when they appear.
Fishkin’s recommendations boil down to six points:
- Replace traffic with a correlation dashboard tracking brand and demand signals over time.
- Do audience research to find where your ICP (Ideal Customer Profile) actually pays attention.
- Invest in channels you don’t own without obsessing over traffic back to your site.
- Keep publishing on-site content anyway, since it still shapes AI Overviews.
- Build short-form storytelling skills for the platforms where attention now lives.
- Remember that SEO still pays off for branded, local, and high-intent transactional searches — territory Cyrus Shepard recently mapped in “The Websites Still Winning In Google.”
SparkToro’s perspective naturally emphasizes where audience attention has shifted. That becomes important when compared with the perspectives that follow.
2. GEO tactics for AI visibility
The second perspective comes from research Fractl conducted with Search Engine Land, presented by cofounder Kelsey Libert at SMX Advanced in Boston on June 4. I covered the key findings for Search Engine Land.
One of the study’s most notable findings is a collapse in trust. In 2025, 82% of consumers found AI search more helpful than traditional search; by 2026, that had fallen to 54%, a 28-point drop in a year.
More useful, though, is the GEO tactic hierarchy Libert presented: high risk, table stakes, and the moat.
- FAQ optimization (49% adoption) is high risk because it’s trivially replicable. Brand mentions, topical authority, and structured data are table stakes.
- The moat is original data, proprietary research, and digital PR — the kind of content AI systems need but can’t replicate.
The hierarchy reflects the broader shift away from traffic-based metrics and toward influence, authority, and original information.
Fractl and Search Engine Land’s data give a tactical answer to which content actually travels. Branded web mentions and YouTube impressions correlate with AI visibility at 0.50-0.74, while backlink count and ad spend sit below 0.30 — a reallocation signal away from link building and paid tactics and toward earned placements and original research.
Additionally, the research found that buyers check an average of 2.4 platforms before validating a purchase. That’s a concrete, surveyable proxy for the “influence” Fishkin says should replace traffic as a KPI.
3. AI measurement through upstream evidence
On May 20, AMEC — the body behind the Barcelona Principles that have shaped PR measurement for over a decade — released its seven GEO Principles and a companion Practitioner’s Guide to GEO Measurement, developed with practitioners from FleishmanHillard, Ketchum, Hotwire Global, Converseon, Big Valley Marketing, and PR Agency One.
If that sounds like a PR trade story rather than an SEO one, that’s part of the problem. Many marketers have treated AI citations as the new rankings. AMEC takes a different view, arguing that visibility is only one part of a broader measurement model that connects AI discovery to awareness, trust, behavior, and business impact.
Although the SEO and PR/comms communities have operated in separate silos for years, both now depend on the same upstream content to shape what AI engines say.
AMEC organizes GEO measurement into three evidence domains:
- Upstream reputation (the earned, shared, and owned content AI models draw on).
- Search and content readiness (whether that information is structured and discoverable).
- Downstream AI output tracking (what stakeholders actually see — presence, framing, citations, accuracy).
Map this against Fishkin’s point that your site’s influence on AI Overviews persists even as clicks disappear, and the connection is clear. AMEC’s upstream and readiness domains are essentially a measurement protocol for the work Fishkin says still matters.
Where it gets more useful for demand gen is Principle 5: GEO measurement should distinguish visibility from outcomes and connect AI discovery to awareness, trust, behavior, and impact.
Appearing in an AI Overview is an output. Whether that appearance moved someone toward a purchase decision is an outcome — and AMEC is explicit that no single tool or score proves that connection.
The guide is candid that connecting any of it to pipeline requires “combined evidence” — a comfort with directional, triangulated evidence rather than a dashboard number that maps cleanly to MQLs (marketing-qualified leads).
For practitioners, the takeaway is a new minimum evidence bar: a governed query library tied to actual buyer questions, documented prompts and platforms, repeat testing with variation disclosed, and saved outputs as evidence.
4. Credibility and AI trust
The fourth perspective comes from Burson, one of the world’s largest PR and communications agencies, which released “The Credibility Paradox: Advancing Generative Engine Optimization from Visibility to Reputation” in June. While AMEC focuses on measuring AI visibility and its business impact, Burson shifts the conversation to whether audiences believe what AI says about a brand.
Burson partnered with AI marketing platform Profound to run thousands of reputation-related prompts across seven AI platforms, covering 85 companies in 10 industries against eight “reputation levers.” The agency’s Decipher tool then generated more than 55,000 “believability forecasts” for the resulting answers.
Interestingly, the headline finding sharpens AMEC’s fifth principle: A brand can be cited by an AI engine and still lose the reputation opportunity if the audience doesn’t believe what the AI says about it. Burson calls this the Credibility Paradox — seen, but not believed.
The most useful finding for demand gen is the proof-versus-posture divide. Levers backed by observable evidence — innovation, creativity, workplace, products — outperformed levers that depend on institutional self-description, like leadership, governance, and citizenship, by roughly a two-to-one margin.
AI engines are more willing to vouch for what your product does and what it’s like to work at your company than for what your leadership says about its own values — a direct signal for content prioritization.
Regarding the methodology, those 55,000 believability forecasts come from an AI system that predicts how human audiences would judge AI-generated answers — AI assessing AI at scale. That makes treating it as directional rather than definitive.
But on the underlying question of whether “credibility” can even be measured, my colleague Katie Paine — a measurement standards veteran who reviewed an earlier draft of this article — makes a useful point: Credibility isn’t as unmeasurable as it sounds, because for goals like this, you can define proxies.
If someone doesn’t find an AI-generated answer about your brand credible, they probably won’t follow you, share your content, or click through when a link is offered. Believability may be a precursor metric to behaviors GEO tools can already track.
Burson naturally emphasizes the credibility layer because reputation measurement is central to its work.
5. Analyst influence in B2B AI discovery
The fifth perspective comes from a different direction entirely. In a LinkedIn post, Jamin Spitzer — a former Microsoft communications insights leader now running his own measurement consultancy — argues that GEO belongs on the analyst relations desk, not the SEO desk.
Spitzer’s case: When a B2B buyer asks an AI platform who leads a category or what their shortlist should be, the answer is frequently a synthesis of analyst content — Gartner, Forrester, IDC, independent analysts — because that content is exactly the authoritative, comparative, taxonomy-rich material generative engines are built to reach for.
AR teams have spent decades trying to trace influence that “shapes a buyer’s mental model” long before it surfaces in a deal. Spitzer argues GEO tools now make that influence newly observable, such as:
- Which analysts’ framing a model is reproducing.
- Whether a brand is described using current or outdated positioning.
- Where gaps exist between a company’s priority analysts and what the AI is actually citing.
This is the B2B hand on the elephant that the first three perspectives largely miss. AMEC, Burson, and Fractl all gravitate toward consumer-facing or brand-reputation signals — workplace, innovation, earned media, and YouTube mentions.
None of them addresses the specific mechanism Spitzer describes: a multi-month enterprise sales cycle in which an AI-generated “consideration set,” built partly from analyst reports, can shape outcomes before a buyer ever opens a Magic Quadrant.
For B2B demand gen specifically, Spitzer’s perspective suggests the upstream content that matters most isn’t earned media or product pages, it’s analyst relationships and the content those relationships produce.
And the pattern repeats once more: An AR-focused measurement consultant is naturally positioned to see the analyst-influence layer of this problem, for the same reason a PR agency sees credibility and a digital PR agency sees entity authority.
Spitzer’s framing doesn’t compete with AMEC, Burson, or Fractl so much as identify a category of upstream source — analyst content that the others’ perspectives don’t name.
6. The credibility gap in AI citations
The sixth perspective connects to research from Angela Dwyer at Full Intel, which Paine also flagged.
Dwyer’s analysis of AI media citations and credible journalism examined which news sources AI platforms cite most often when answering questions. She found a gap between citation frequency and the outlets that audiences rate as most trustworthy.
That’s a publisher-side mirror of Burson’s brand-side paradox: Just as a brand can be visible but not believed, a publication can be heavily cited by AI engines while its own readers hold it in lower regard than less-cited competitors.
For demand gen, it’s a reminder that the upstream sources AMEC and Burson both point to aren’t a neutral pool. Some of the outlets AI leans on most are themselves fighting a credibility gap, which complicates the idea that “getting cited by a major publication” is a clean proxy for credibility transfer.
Track your visibility across AI search, uncover missed opportunities, and grow your presence where customers are asking questions.
Here’s the elephant
Read side by side, these perspectives resemble six different hands on the same elephant:
- SparkToro sees attention and correlation.
- Fractl sees entity authority and earned mentions.
- AMEC sees upstream evidence domains.
- Burson sees credibility and believability.
- Spitzer sees analyst influence in B2B buying cycles.
- Dwyer’s research highlights the credibility of the news sources AI relies on.
None of these perspectives replaces the others. Each measures a different dimension of how AI influences discovery, trust, and buying decisions. Together, they suggest marketing measurement is becoming multidimensional rather than website-centric.
Rather than searching for a single model, marketers may need to combine multiple perspectives. No single perspective captures the whole picture, but together they offer a more complete view of how AI shapes visibility, influence, and demand.
