A new initial visibility study by Taptwice Media, leading AEO agency based in India, has found substantial differences in how leading AI engines represent industrial quoting and configure-price-quote software vendors, with citation behavior varying dramatically even when the systems were given exactly the same questions.

The study tested ChatGPT, Perplexity, Claude, Google Gemini and Bing Copilot using 20 non-branded questions about industrial quoting, AI-native quoting software, configure-price-quote platforms and engineer-to-order workflows.

Across 100 total AI-generated answers, three companies showed the strongest cross-platform consistency: Tacton, Salesforce CPQ/Revenue Cloud and Oracle CPQ appeared among the most frequently named companies across all five engines.

Epicor CPQ, SAP CPQ and DealHub appeared among the leading names in four of the five.

But beyond the companies being mentioned, Taptwice Media found that the five AI systems behaved very differently in how they constructed and supported their answers.

ChatGPT and Perplexity cited far more sources

Citation volume showed one of the clearest differences between the engines.

Across the 20 questions, ChatGPT produced 791 citation occurrences spanning 322 unique domains. Perplexity generated 598 citation occurrences across 251 domains.

The numbers fell sharply after that.

Bing Copilot produced 97 citation occurrences from 56 domains, while Google Gemini produced 56 citations from 48 domains.

Claude produced no source citations across any of the 20 tested answers.

Claude also declined to name vendors in four questions where it said it did not have sufficiently reliable current information, particularly around venture-backed quoting startups and adoption in specialist industries such as rail and fire apparatus manufacturing.

The other four systems named at least one company in nearly every tested query.

The result illustrates an important distinction in AI visibility: appearing in an AI answer and being supported by visible sources are not necessarily the same thing.

The same questions produced very different vendor landscapes

Tacton was named in 10 of ChatGPT's 20 answers, nine Perplexity answers, eight Gemini answers and six answers each from Claude and Bing Copilot.

Salesforce CPQ or Revenue Cloud appeared particularly frequently in Claude, Gemini and Bing Copilot, while Oracle CPQ maintained relatively consistent visibility across the engines.

Other companies showed much greater variation.

Paperless Parts, for example, appeared frequently in Perplexity but much less often in several other engines. DealHub was prominent across ChatGPT, Perplexity, Gemini and Bing Copilot but did not appear among Claude's most frequently named vendors.

Claude also surfaced companies such as Apptio, Coupa, Determine and JAGGAER more often than the other engines.

The differences suggest that there is no single AI search landscape for a software category.

A company may have strong visibility in one AI engine and limited visibility in another even when users ask the same underlying question.

Case studies appear to play a major role in AI answers

The engines repeatedly supported vendor recommendations with customer examples containing measurable outcomes.

ChatGPT referenced examples including Yaskawa's reported 87% faster quoting with Tacton, WABTEC's reported 38% faster quotation generation through Conga CPQ, and E-ONE's reported reduction in quote creation time through Cincom CPQ.

Perplexity cited cases involving companies including Mitsubishi Electric, StepStone, ROBEL and Chiron Group.

Bing Copilot similarly relied on quantified implementation examples when discussing quoting efficiency and implementation outcomes.

The figures themselves were generally sourced from vendor case studies or customer-reported results rather than independent controlled studies, and several of the AI engines explicitly acknowledged that limitation.

Even so, the frequency with which these cases appeared suggests that specific, attributable proof points can become important retrieval material when AI systems construct commercial software recommendations.

A generic claim that a platform "improves quoting efficiency" provides considerably less information for an AI system to work with than a documented example identifying a customer, starting point and measurable result.

Vendor-owned websites remain important AI sources

Another notable finding was how heavily several AI engines relied on vendor-controlled content.

Tacton's own website was ChatGPT's most frequently cited domain in the study, appearing 35 times.

Paperless Parts was Perplexity's most frequently cited domain, with 41 citation occurrences, followed by DealHub, LinkedIn, Tacton and aPriori.

Vendor documentation, product pages, implementation stories and technical materials appeared repeatedly throughout both engines' source sets.

Google Gemini displayed a much more fragmented citation pattern. Its most frequently cited domain appeared only three times, while 43 of its 48 cited domains appeared just once.

Bing Copilot looked different again.

Its most frequently cited sources were largely third-party statistics or research aggregation websites, including WorldMetrics, ZipDo, Gitnux and WiFiTalents, rather than the vendor websites that dominated ChatGPT and Perplexity.

That difference matters because AI visibility strategies built around a single search engine or citation pattern may fail to account for how differently other systems retrieve information.

AI engines do not simply repeat positive vendor messaging

The study also tracked how frequently AI systems introduced criticism, limitations or cautionary language alongside vendor recommendations.

Bing Copilot attached some form of criticism or caveat to a named company in all 20 tested answers.

ChatGPT did so in 17 of 20.

Gemini was considerably less critical, producing nine answers in which the vendors it named received no negative framing.

Common concerns raised across the engines included implementation complexity, ERP integration, dependence on rule-based systems, AI hallucination risk, opaque pricing logic, expensive professional services and the difficulty of converting engineering knowledge into reliable configuration rules.

This creates another dimension of AI visibility beyond simply being mentioned.

A company may appear frequently but consistently alongside concerns about implementation difficulty, integration limitations or product maturity. Another company may appear less frequently but receive stronger positioning when it is mentioned.

For companies monitoring their presence in generative search, mention volume therefore provides only part of the picture. Sentiment and context matter as well.

AI visibility is becoming a source-distribution problem

The cross-engine findings point toward a broader change in how software companies may need to think about discoverability.

Traditional search visibility has generally focused on whether a company's own pages rank for commercially important queries.

AI systems assemble answers differently.

They may draw from product documentation, customer case studies, industry publications, community discussions, technical documentation, company websites and third-party datasets simultaneously.

The study suggests that companies with a broad footprint of structured, specific and independently discoverable information have more opportunities to become part of those answers.

That does not mean simply publishing more marketing content.

AI engines repeatedly surfaced material containing identifiable customers, implementation details, technical capabilities, integrations and numerical outcomes.

In several cases, they also challenged unsupported claims or explicitly noted when performance figures originated from vendor-produced case studies.

For industrial software companies, the emerging visibility question may therefore be less about whether a website contains the right sales language and more about whether enough verifiable information about the company exists across the sources AI systems actually retrieve.

Five AI engines increasingly represent five different discovery environments

Perhaps the clearest finding from the study is that AI visibility cannot reliably be measured through one model.

ChatGPT and Perplexity produced dense citation networks.

Gemini cited far fewer sources and spread those citations widely.

Bing Copilot relied unusually heavily on third-party statistics sites.

Claude was considerably more willing to say that it lacked enough current information rather than name a company and supplied no visible citations in the tested answers.

Despite those differences, Tacton, Salesforce CPQ and Oracle CPQ maintained enough visibility to appear prominently across all five systems.

That cross-model consistency is relatively uncommon in the wider dataset and provides one indication of what durable category visibility can look like as software discovery shifts from conventional search results toward AI-generated answers.

The research was conducted by Taptwice Media on September 19, 2026, using 20 non-branded queries across ChatGPT, Perplexity, Claude, Google Gemini and Bing Copilot. The analysis covered company mentions, citations, customer examples and positive and negative contextual framing across 100 generated answers.

About Taptwice Media

Taptwice Media is a Delhi NCR-based growth and AI visibility agency founded by Shubham Kumar Agrawal. Its work includes Answer Engine Optimization, Generative Engine Optimization, content distribution, AI brand sentiment analysis and brand mention tracking across major AI platforms. Taptwice helps brands and enterprises increase awareness and improve how they appear in AI answers produced by ChatGPT, Google AI Mode and Gemini, Perplexity, Claude etc.

Media Contact :
Shubham Kumar Agrawal
shubham@taptwicemedia.com
+91-8506085059 / +91-8506085039