We Asked ChatGPT and Claude Who to Hire for Ecommerce SEO. Here Is What They Cited.
We put 24 real, high-intent buyer questions to two frontier AI engines and recorded all 48 answers. The question underneath every one: when a business owner asks an AI assistant who to hire for SEO, which brands get named, and which get left out?
AI assistants are becoming the place buyers start their vendor research. When someone types “who should I hire for Shopify SEO” into ChatGPT, the engine’s answer is the shortlist. So we measured that shortlist directly. We asked ChatGPT (GPT-5.5) and Claude (Opus 4.8) the same 24 purchase-intent questions an ecommerce or mid-market decision maker actually asks, saved every raw response to disk, and tabulated exactly who got cited.
Ten findings you can quote
Every figure below is drawn from the 48 saved answers. Each card is a standalone, citable fact.
Across the 42 answers to “who should I hire” questions, AI engines named a specific hireable agency only 33% of the time.
In 38% of those hiring answers, the engines named only software tools (like Ahrefs or Semrush) and no agency a buyer could actually hire.
Across all six cannabis dispensary SEO answers, ChatGPT and Claude named zero specific agencies. The category was 100% generic.
Perplexity was cited in 100% of GEO agency answers and 100% of AI-search-visibility answers, the only entity to reach 100% in a hiring category.
One in four AI answers (25%) stayed fully generic or hedged, naming no brand, tool, or agency at all.
Amsive was the most-cited agency, named 7 times across 6 of the 24 questions (17% of all hiring answers).
Ahrefs, Perplexity, and Semrush tied as the most-cited entities overall, at 16 mentions each across the 48 answers.
ChatGPT (GPT-5.5) named a specific provider in 83% of its answers; Claude (Opus 4.8) did so in 67%.
The two engines agreed weakly: mean entity overlap (Jaccard) across matched questions was just 0.31, and they shared a cited brand in only 14 of 24 questions.
Half the buyer categories we tested (ecommerce agency, Shopify, auto parts, cannabis) had no dominant cited source at all.
Methodology
This is an original-data study. We did not scrape a list or rework someone else’s numbers. We queried live AI engines, saved every raw response as a JSON file, and computed the statistics from those files.
The engines
Two frontier models, queried through their official APIs on 8 July 2026:
- ChatGPT: OpenAI model
gpt-5.5 - Claude: Anthropic model
claude-opus-4-8
Each engine received the identical set of 24 questions, one at a time, with no system prompt steering it toward or away from any brand. That produced 48 answers (24 questions x 2 engines), and all 48 returned successfully with zero errors.
The questions
The 24 questions are real, high commercial-intent prompts an ecommerce or mid-market buyer types when researching SEO and generative-engine-optimization help. They span 8 categories:
| Category | Example question |
|---|---|
| Best ecommerce SEO agency / consultant | What is the best ecommerce SEO agency for a mid-market online store in 2026? |
| Who to hire for Shopify SEO | Who should I hire to do SEO for my Shopify store? |
| Best technical SEO agency | Who should I hire for a technical SEO audit of my online store? |
| AI search visibility / cited by ChatGPT | How do I get my ecommerce store recommended and cited by ChatGPT? |
| Best GEO / generative engine optimization agency | What is the best generative engine optimization (GEO) agency? |
| Auto parts ecommerce SEO help | Best SEO help for an aftermarket automotive parts ecommerce business? |
| Cannabis dispensary SEO agency | What is the best SEO agency for a cannabis dispensary? |
| Tool / platform comparison | Which GEO / AI-visibility tracking tools should I compare before buying? |
How we counted a citation
For each answer we extracted every named brand, agency, tool, or platform (an “entity”). An answer counts as “named” if it cited at least one specific entity, and “generic or hedged” if it gave only advice with no nameable provider. We separated hireable agencies from software tools, because a buyer asking who to hire is not served by a list of software. We also flagged the four categories where any single source was cited in a majority of answers (“dominant source”) versus categories with no clear winner.
Which brands AI engines cite most
Across all 48 answers, 41 distinct entities were named. Three tied at the top with 16 mentions each. Notably, the leaders are mostly software tools, not agencies a buyer can hire. GEO-specialist platforms (Peec AI, Profound, Otterly.AI) already rank alongside the incumbent SEO suites.
| Entity | Type | Total mentions | ChatGPT | Claude | Distinct questions |
|---|---|---|---|---|---|
| Ahrefs | Tool | 16 | 8 | 8 | 10 |
| Perplexity | Tool / engine | 16 | 8 | 8 | 8 |
| Semrush | Tool | 16 | 9 | 7 | 10 |
| Google Search Console | Tool | 15 | 11 | 4 | 11 |
| Peec AI | GEO tool | 13 | 5 | 8 | 8 |
| Profound | GEO tool | 13 | 5 | 8 | 8 |
| Otterly.AI | GEO tool | 12 | 4 | 8 | 8 |
| Screaming Frog | Tool | 9 | 5 | 4 | 6 |
| Amsive | Agency | 7 | 4 | 3 | 6 |
| Scrunch AI | GEO tool | 7 | 4 | 3 | 4 |
| Go Fish Digital | Agency | 6 | 4 | 2 | 5 |
| AthenaHQ | GEO tool | 5 | 3 | 2 | 4 |
| Inflow | Agency | 5 | 5 | 0 | 5 |
| Seer Interactive | Agency | 5 | 3 | 2 | 5 |
| Siege Media | Agency | 5 | 3 | 2 | 4 |
“Ahrefs, Perplexity, and Semrush tied as the most-cited names in AI answers about SEO help, at 16 mentions each. Every one of them is a tool, not an agency you can hire.”MDRN SEO AI Citation Study, July 2026 (n=48 answers)
When buyers ask who to hire, engines mostly don’t name anyone
Seven of the eight categories carry explicit hiring intent (21 questions x 2 engines = 42 provider-intent answers). Here is how those 42 answers broke down:
answers (33%) named a specific hireable agency.
answers (38%) named only tools, no hireable agency.
answers were fully generic, naming no entity at all.
When engines did name an agency, the recommendations were thinly spread across a dozen firms. No agency was cited in more than 6 of the 24 questions. This is a wide-open field: there is no “default” agency the engines reach for.
| Agency | Mentions | Distinct questions | % of provider answers |
|---|---|---|---|
| Amsive | 7 | 6 | 17% |
| Go Fish Digital | 6 | 5 | 14% |
| Inflow | 5 | 5 | 12% |
| Siege Media | 5 | 4 | 12% |
| Seer Interactive | 5 | 5 | 12% |
| Victorious | 4 | 4 | 10% |
| OuterBox | 4 | 4 | 10% |
| NP Digital | 4 | 3 | 10% |
| Searchbloom | 3 | 2 | 7% |
| SmartSites | 1 | 1 | 2% |
| Coalition Technologies | 1 | 1 | 2% |
| Single Grain | 1 | 1 | 2% |
| WebFX | 1 | 1 | 2% |
Often the engines were candid about why they would not crown a winner. Asked for the best ecommerce SEO agency, Claude answered:
“There isn’t a single ‘best’ agency I can point you to with confidence, and honestly you should be skeptical of anyone (including AI tools) that names one definitively.”Claude (Opus 4.8), verbatim, Question 1
Where a dominant source exists, and where it doesn’t
We flagged a category as having a “dominant source” when one entity was cited in a majority of that category’s answers. Four categories had one. Four did not, meaning the AI shortlist there is still up for grabs.
| Category | Dominant source | Top cited entities (share of answers) |
|---|---|---|
| Best technical SEO agency | Ahrefs | Ahrefs, GSC, Screaming Frog, Semrush (67% each) |
| AI search visibility | Perplexity | Perplexity (100%), Peec AI & Profound (67%) |
| Best GEO agency | Perplexity | Perplexity (100%), Otterly.AI, Peec AI, Profound (83%) |
| Tool / platform comparison | Ahrefs | Ahrefs & Semrush (100%), Otterly.AI, Peec AI, Perplexity (67%) |
| Best ecommerce SEO agency | None | Inflow & Go Fish Digital (50%), OuterBox, Victorious (33%) |
| Who to hire for Shopify SEO | None | GSC (50%), Searchbloom & Victorious (33%) |
| Auto parts ecommerce SEO | None | GSC (50%), Ahrefs, Screaming Frog, Semrush (33%) |
| Cannabis dispensary SEO | None | No entities cited in any of the 6 answers |
“In every one of the six cannabis dispensary SEO answers, both AI engines declined to name a single agency. That entire buyer category has no cited source for an AI engine to fall back on.”MDRN SEO AI Citation Study, July 2026
Even in a niche where the engines refused to name agencies, Claude still recognized the category needs a specialist:
“There isn’t a single ‘best’ SEO agency for cannabis dispensaries, but I can help you understand what to look for… This is a niche where specialization genuinely matters.”Claude (Opus 4.8), verbatim, Question 19
ChatGPT names more brands than Claude, and they rarely agree
The two engines behaved differently. ChatGPT was more willing to name specific providers and drew from a wider pool of brands. Claude hedged more often. And when you compare their answers to the same question, the overlap is low.
| Metric | ChatGPT | Claude |
|---|---|---|
| Answers naming a specific provider | 20 of 24 (83%) | 16 of 24 (67%) |
| Distinct entities cited | 40 | 26 |
| Measure | Result |
|---|---|
| Both engines named at least one brand | 16 of 24 |
| Both engines stayed generic | 4 of 24 |
| Split (one named, one generic) | 4 of 24 |
| Questions sharing at least one cited brand | 14 of 24 |
| Mean entity overlap (Jaccard) | 0.31 |
“On identical buyer questions, ChatGPT and Claude shared a cited brand only 14 times out of 24, with a mean overlap of 0.31. Two leading AI engines are recommending substantially different shortlists.”MDRN SEO AI Citation Study, July 2026
Both engines were also frank that GEO is an unsettled category. Asked to name the best GEO agency, ChatGPT replied:
“There isn’t a single ‘best’ generative engine optimization agency yet because GEO is still an emerging category, and a lot of vendors are rebranding SEO/content/PR services as ‘GEO.'”ChatGPT (GPT-5.5), verbatim, Question 13
What this means for your AI search visibility
Put the findings together and a clear picture emerges for any brand that wants to be recommended by AI:
1. The shortlist is short, and mostly unclaimed
Only 33% of hiring answers named an agency at all, and no agency appeared in more than a quarter of the questions. Whole categories (cannabis, auto parts, Shopify, general ecommerce) have no dominant cited source. That is not a closed market. It is an opening. The brands that become the answer will be the ones AI engines can find, understand, and trust as an entity.
2. Being cited is a different discipline from ranking
The engines lean on tools they can reference and entities with a clear, corroborated presence across the web. Perplexity hitting 100% in the GEO and AI-visibility categories, and the GEO-specialist platforms already outranking most agencies, shows that AI citation follows entity clarity and repeated third-party mention, not just backlinks to a homepage.
3. Two engines, two shortlists
With a mean overlap of just 0.31, you cannot optimize for “AI” as one thing. ChatGPT and Claude cite different brands. Real generative engine optimization means earning presence across each engine’s distinct sourcing, not chasing one algorithm.
4. If you are not cited, you are invisible at the moment of decision
In this study, our own brand, MDRN SEO, was cited 0 times across all 48 answers. We are reporting that honestly, because it is exactly the gap this work is built to close, for us and for our clients. When the buyer asks the AI and your name never comes up, the shortlist was decided without you.
Generative engine optimization is the practice of becoming one of the names an AI engine cites: structured, corroborated, entity-clear content that engines can lift with confidence. That is the service this study was designed to inform.
Want your brand to be the answer AI gives?
We help ecommerce and mid-market brands become the names ChatGPT, Claude, and Perplexity actually cite. It starts with the same kind of measurement you just read.
Study FAQ
Which AI engines did the study test, and when?
How often did AI engines name a specific agency to hire?
Which brands did AI engines cite most for SEO help?
Which agency was recommended most often?
Did ChatGPT and Claude agree on who to recommend?
Were there categories where AI named no one at all?
Is the underlying data original?