Ahrefs briefly recovered in April, then fell harder in May. Claude hit its highest delta ever. The SEO vertical now has three AI Memory Brands. This is what platform divergence looks like at scale.
Run #3 of the GEO Brand Citation Index tracked 35 brands across 54 queries on ChatGPT, Perplexity, and Gemini. The headline numbers: Ahrefs dropped from Dominant Brand back to AI Memory Brand with a −39.13 delta — its worst Perplexity gap across three runs. Claude posted its highest delta ever at +60.87, now the definitive Live Search Brand signal in the AI tools vertical. The SEO vertical has three simultaneous AI Memory Brands for the first time: Ahrefs, Moz, and Google Search Console.
The two-mechanism model is now clear across three months of data: brands either win on model memory (ChatGPT) or on live retrieval (Perplexity) — very few win both simultaneously. Semrush and HubSpot remain the exceptions. Every other brand is drifting toward one pole or the other.
What Changed Between April and May in the GEO Brand Citation Index
In April’s run I wrote about Ahrefs graduating to Dominant Brand. Its Perplexity score had climbed to 75, its delta narrowed from −48.1 to −25.0, and the trend looked like a brand actively recovering ground on the live web. One month later, that reading looks premature.
May’s Perplexity score for Ahrefs dropped back to 60.87 — widening the gap to −39.13. The graduation was real in April’s data, but it didn’t hold. What April’s score may have captured was a short-lived burst of Perplexity retrieval (a product launch, a major piece of coverage, a ranking movement) rather than a durable shift in live web presence. That distinction — between a spike and a trend — is exactly what three months of data starts to reveal.
The more interesting inversion this month is Claude. Where Ahrefs scores 100% on ChatGPT and 60.87% on Perplexity — a training-data brand coasting on legacy — Claude does the opposite: 13.04% on ChatGPT, 73.91% on Perplexity, a +60.87pp gap running the other direction. That is the Live Search Brand signature: Anthropic is generating fresh content, press coverage, and practitioner discussion at a pace that Perplexity’s real-time retrieval picks up, while OpenAI’s own model unsurprisingly under-cites its direct competitor. The signal for any brand watching this index is that the two citation mechanisms — model memory versus live retrieval — are diverging faster than most teams realise. Optimising for one without the other is not a strategy; it is a bet on which type of AI query your prospects happen to use.
Two brands exited the panel this month: Copper (CRM) and Runway (AI tools). Both had consistently low scores across all three platforms with no meaningful trend signal. The contraction from 37 to 35 brands reflects a decision to track brands where citation patterns are informative — near-zero scores across all platforms for three consecutive runs provide no new data.
April vs May: The Full Comparison Table
The table below shows all 35 brands from May with their April delta for comparison. Sorted by delta shift, largest positive first. Brands that changed archetype are flagged in the rightmost column.
How to read the Δ Shift column: This tracks how the gap between Perplexity and ChatGPT changed month-on-month. ▲ means Perplexity gained ground relative to ChatGPT. ▼ means ChatGPT gained ground. The interactive full dataset — including all three months — is on the GEO Brand Citation Index.
| Brand | Vertical | Apr Δ | May Δ | Δ Shift | Archetype |
|---|---|---|---|---|---|
| Claude | AI | +43.5 | +60.9 | ▲ +17.4 | 🔍 Live Search |
| HubSpot | CRM | +16.4 | +25.9 | ▲ +9.5 | 👑 Dominant |
| Zoho CRM | CRM | +2.1 | +13.4 | ▲ +11.3 | ⭐ GEO Outlier |
| Gemini | AI | +8.7 | +17.4 | ▲ +8.7 | NEW: 🔍 Live Search |
| Surfer SEO | SEO | +16.1 | +39.1 | ▲ +23.0 | 🔍 Live Search |
| Monday.com | CRM | +1.1 | +12.5 | ▲ +11.4 | — |
| Copy.ai | AI | −13.0 | +4.4 | ▲ +17.4 | Fading → — |
| Cursor | AI | +17.4 | +13.0 | ▼ −4.4 | — |
| Copilot | AI | +17.4 | +8.7 | ▼ −8.7 | — |
| ChatGPT | AI | 0.0 | 0.0 | — flat | 👑 Dominant |
| Semrush | SEO | +4.3 | +4.4 | — flat | 👑 Dominant |
| Screaming Frog | SEO | +3.4 | 0.0 | ▼ −3.4 | — |
| Mangools | SEO | +3.8 | −4.4 | ▼ −8.2 | — |
| SE Ranking | SEO | +20.6 | +8.7 | ▼ −11.9 | — |
| Pipedrive | CRM | +48.1 | +10.7 | ▼ −37.5 | Live Search → — |
| Insightly | CRM | −7.4 | +0.9 | ▲ +8.3 | — |
| Notion AI | AI | +17.4 | +4.4 | ▼ −13.0 | — |
| Perplexity | AI | +8.7 | +8.7 | — flat | — |
| Salesforce | CRM | 0.0 | −20.8 | ▼ −20.8 | 👑 Dominant |
| Grammarly | AI | +4.3 | −8.7 | ▼ −13.0 | — |
| Writesonic | AI | 0.0 | −4.4 | ▼ −4.4 | — |
| Jasper | AI | −8.7 | −13.0 | ▼ −4.3 | — |
| Clearscope | SEO | +4.2 | +4.4 | — flat | — |
| Rank Math | SEO | −8.7 | 0.0 | ▲ +8.7 | — |
| Yoast SEO | SEO | −17.4 | −13.0 | ▲ +4.4 | — |
| Ubersuggest | SEO | −13.4 | −8.7 | ▲ +4.7 | — |
| Majestic | SEO | −8.7 | −8.7 | — flat | — |
| Google Search Console | SEO | +6.7 | −21.7 | ▼ −28.4 | GEO Outlier → 🧠 AI Memory |
| Moz | SEO | −43.7 | −34.8 | ▲ +8.9 | 🧠 AI Memory |
| Ahrefs | SEO | −25.0 | −39.1 | ▼ −14.1 | Dominant → 🧠 AI Memory |
| Keap | CRM | −7.4 | −7.4 | — flat | — |
| Close CRM | CRM | −2.6 | 0.0 | ▲ +2.6 | — |
| SugarCRM | CRM | −7.4 | −11.1 | ▼ −3.7 | NEW: 🫥 Fading |
| Freshsales | CRM | −10.1 | −14.8 | ▼ −4.7 | 🫥 Fading |
| Midjourney | AI | +4.4 | 0.0 | ▼ −4.4 | — |
Archetype Shifts: Twelve Changes in One Month
The full dataset records 12 archetype changes this month. Five brands moved into a new named classification; seven lost a classification and reverted to unclassified. Both types are real shifts — the distinction matters for how to read them.
Reclassified — gained or changed named archetype
| Brand | April Archetype | May Archetype | Driver |
|---|---|---|---|
| Ahrefs | 👑 Dominant Brand | 🧠 AI Memory Brand | Perplexity score fell from 75.0 to 60.87; delta widened to −39.13 |
| Google Search Console | ⭐ GEO Outlier | 🧠 AI Memory Brand | Perplexity score halved from 45.8 to 17.39; delta swung to −21.74 |
| Gemini | — Unclassified | 🔍 Live Search Brand | Perplexity score rose to 17.39 and Gemini itself scores 32.14; live web presence solidifying |
| SugarCRM | — Unclassified | 🫥 Fading Brand | Perplexity dropped to 0.0 for second consecutive run; consistent zero-citation pattern |
| Copy.ai | 🫥 Fading Brand | — Unclassified | Perplexity recovered from 0.0 to 17.39; no longer meets Fading threshold |
Dropped to unclassified — lost named archetype
| Brand | April Archetype | May Archetype | Driver |
|---|---|---|---|
| Pipedrive | 🔍 Live Search Brand | — Unclassified | Delta collapsed from +48.15 to +10.65; April reading now looks like a spike, not a trend |
| SE Ranking | 🔍 Live Search Brand | — Unclassified | Delta dropped from +20.6 to +8.7; fell below Live Search threshold |
| Notion AI | 🔍 Live Search Brand | — Unclassified | Delta dropped from +17.4 to +4.4; Perplexity advantage no longer consistent |
| Copilot | 🔍 Live Search Brand | — Unclassified | Delta dropped from +17.4 to +8.7; fell below Live Search threshold |
| Cursor | 🔍 Live Search Brand | — Unclassified | Delta dropped from +17.4 to +13.0; narrowly below threshold |
| Jasper | ⭐ GEO Outlier | — Unclassified | Delta widened from −8.7 to −13.0; crossed GEO Outlier boundary |
| Yoast SEO | 🫥 Fading Brand | — Unclassified | Delta improved from −17.4 to −13.0; no longer meets Fading threshold |
The Ahrefs reversal deserves a closer reading. In March, Ahrefs had a −48.1 delta (AI Memory). In April it recovered to −25.0 and graduated to Dominant Brand. In May it’s back at −39.1. This is not a trend — it’s oscillation. The brand is sitting on a structural boundary where a single good month of Perplexity coverage can temporarily push it into Dominant territory, but it cannot sustain that position. Until Ahrefs builds durable live web presence that Perplexity retrieves consistently, the ‘Dominant’ classification will remain unstable.
Google Search Console’s shift tells a different story. It moved from AI Memory to GEO Outlier in April (+6.7 delta, positive for the first time), then reversed hard to −21.74 in May. The April reading may have reflected Perplexity surfacing content around the new Search Console features and API changes announced that month. By May, that retrieval signal had faded back to baseline.
The SEO Vertical Has Three AI Memory Brands
This is a first across three months of tracking: Ahrefs, Moz, and Google Search Console are simultaneously classified as AI Memory Brands. All three show the same structural pattern — high ChatGPT scores (39–100%) reflecting historical training data, and Perplexity scores that fall well below (13–61%).
The only brand in the SEO vertical holding Dominant Brand status is Semrush (delta: +4.35, up marginally from +4.3 in April). Surfer SEO maintains Live Search Brand at +39.1. The rest of the vertical is trending toward AI Memory or holding at unclassified-but-negative deltas.
For practitioners advising SEO software brands: three concurrent AI Memory classifications in one vertical is a structural signal, not a coincidence. It suggests these brands built their market position during the period when current AI training data was gathered — predominantly 2021–2023 — and have not generated enough live web discussion since to match that historical footprint in Perplexity’s retrieval. The corrective action is not a single piece of content but a sustained programme: thought leadership, third-party coverage, structured data, and regular product updates that give Perplexity something current to retrieve. The GEO Stack framework covers the full set of levers available to a brand in this position.
Claude: The Highest Delta in Three Runs of the Index
Claude’s trajectory across three runs: +50.1 (March), +43.5 (April), +60.87 (May). The delta is not just large — it is increasing. Perplexity’s real-time retrieval is consistently surfacing Claude-related content at rates well above what OpenAI’s model cites.
The mechanism is structural. Perplexity retrieves from the live web, where Anthropic has generated substantial recent coverage: Claude 3.7, Claude Code announcements, model capability comparisons, and a growing practitioner community publishing results. ChatGPT’s model memory was trained primarily on data from before Claude reached its current market position, so the citation gap reflects a temporal mismatch as much as a competitive one.
The practical implication: if your target buyer uses ChatGPT for tool discovery, Anthropic has a problem. If they use Perplexity, Anthropic dominates. For brands in adjacent categories, this split is a preview of what happens when your category has both early-training-data incumbents and fast-moving challengers generating live content simultaneously. Understanding your brand’s retrieval probability across platforms is the first diagnostic step.
Three-Month View: What the Trend Lines Show
With three runs complete, some patterns are now stable enough to treat as structural rather than sampling noise. The March 2026 run established the initial baselines; the picture has sharpened considerably since.
Consistently stable Dominant Brands: ChatGPT and Semrush have held the top two positions by overall score across all three runs, with minimal delta variance. Both have what the index would call dual-track authority — deep in training data and active on the live web.
The CRM reversal: Salesforce moved from a flat delta (+0.0) in April to −20.83 in May. Its ChatGPT score held at 100%, but Perplexity dropped from 100% to 79.17%. This is the first negative signal for Salesforce in three runs and warrants watching in June.
Pipedrive’s correction: April’s +48.15 delta was the largest single positive reading in two runs. May’s +10.65 is a significant correction. The Live Search Brand classification is lost. Whether April was a genuine step-change or an anomaly will be clearer when June data arrives.
The Perplexity-only pattern is hardening: Surfer SEO (+39.1), Claude (+60.9), and Gemini (+17.4) now form a distinct cluster — brands with very low ChatGPT scores but meaningful Perplexity scores. All three are products where recent news, model updates, and practitioner reviews are generating the kind of fresh, crawlable content that Perplexity surfaces. All three are also products that compete with, or are adjacent to, OpenAI’s own offerings, which may introduce a citation suppression effect in ChatGPT responses.
Practitioner Perspectives
“The Ahrefs oscillation between Dominant and AI Memory across three months is exactly what I’d expect from a brand that had outsized training-data footprint but hasn’t kept pace on fresh content. The question isn’t whether they can recover — it’s whether they’re measuring the right things to know when they have.”
“The SEO vertical having three AI Memory Brands simultaneously is a wake-up call for the category. These are not obscure tools — Ahrefs, Moz, and GSC collectively define how most SEO teams work. If Perplexity is systematically under-citing them, the next generation of practitioners discovering tools via AI queries is going to get a very different shortlist.”
Frequently Asked Questions
Why did Ahrefs lose its Dominant Brand status after one month?
Archetype classification is based on each run’s data independently. In April, Ahrefs’ Perplexity score crossed the Dominant threshold. In May, it fell back below it. The index does not smooth or lag classifications — each month reflects current citation behaviour. Ahrefs’ pattern (strong ChatGPT, variable Perplexity) suggests its Dominant reading in April may have been a transient spike rather than a durable shift. Full classification methodology here.
Why does ChatGPT cite Claude so rarely if Claude is a major AI model?
Two factors. First, ChatGPT’s training data has a knowledge cutoff, and much of Claude’s market presence has grown after that cutoff. Second, OpenAI’s model has structural reasons to under-cite direct competitors — when asked about AI tools, ChatGPT is likely to mention alternatives that are complementary (Grammarly, Notion AI) rather than substitutes (Claude, Gemini). This is consistent with what we see in the data: ChatGPT scores for Claude at 13%, while Perplexity — which retrieves from the live web without the same competitive dynamic — scores Claude at 74%.
What does a negative delta mean for a Dominant Brand like Salesforce?
A brand can hold Dominant Brand classification while carrying a negative delta if its absolute scores remain high across all platforms. Salesforce’s May scores are 100% on ChatGPT and Gemini, 79.17% on Perplexity — the negative delta simply means Perplexity is slightly below ChatGPT. The classification threshold uses absolute score levels. The delta is a leading indicator worth watching: if Salesforce’s Perplexity score continues to slide while ChatGPT holds, it will eventually cross into AI Memory territory.
Three months of data now show two citation mechanisms diverging: model memory (ChatGPT) and live retrieval (Perplexity). Brands optimising for one without measuring the other are making a bet, not a strategy.