> ## Documentation Index > Fetch the complete documentation index at: https://docs.qwairy.co/llms.txt > Use this file to discover all available pages before exploring further. # Do different AI providers describe my brand consistently? > Detect where ChatGPT, Claude, Perplexity, and other AI models converge or diverge in how they portray your brand — and understand why the differences exist. **TL;DR** — Your brand has a different AI voice on every model, and the divergence between ChatGPT, Claude, Perplexity, and Gemini can be dramatic. Go to **Analyze > Perception** and toggle the **Provider** filter to compare attribute maps side by side, then use **Cockpit > GEO Matrix** to spot prompts where your score swings wildly between provider columns. Check **Monitor > Response Analysis** for the same prompt across providers to read the exact differences in framing. Pro tip: a provider where you have both low visibility and divergent attributes is the highest-priority target for remediation. ## The Question > **"Do different AI providers describe my brand consistently?"** Your brand does not have a single AI voice — it has one per model. ChatGPT's training data, Perplexity's live retrieval, Claude's constitutional approach, and Gemini's corpus each produce a slightly different (and sometimes dramatically different) portrait of your brand. A user on Perplexity may encounter a description that emphasizes your pricing. A user on ChatGPT may encounter one that highlights your founding story. A user on Claude may encounter factual claims that differ from both. Understanding where providers converge gives you confidence in what AI has reliably absorbed about your brand. Understanding where they diverge shows you either where training data quality varies, or where different sources — some accurate, some outdated — are being retrieved by different models. **You might also be wondering:** * "Why does Perplexity describe my brand so differently from ChatGPT?" * "Which AI provider gives the most accurate description of my brand?" * "If I fix a claim on one provider, does it automatically fix on others?" *** ## Where to Go in Qwairy Navigate to **Analyze > Perception** — the primary view for convergence and divergence analysis. Apply the **Provider** filter one at a time (ChatGPT, Claude, Perplexity, Gemini) and compare the resulting attribute maps. Note which themes appear consistently across all providers (convergence) and which appear in only one or two (divergence). Cross-reference with **Cockpit > GEO Matrix** and examine the provider columns. Each cell shows your visibility score for a given prompt and provider combination. Scan for prompts where your score is high on one provider and low on another — these are the inconsistency hotspots worth investigating further. Open **Monitor > Response Analysis** and filter by a specific prompt. Then compare the response text side-by-side across providers. Read the full text of what ChatGPT says versus what Perplexity says for the same question about your brand. The divergence in language, claims, and framing will be immediately visible. Use **Workspace > Shared Views** to create a provider-specific perception snapshot. Lock the filters to a single provider so that different stakeholders (e.g., the SEO team focused on Perplexity, the comms team focused on ChatGPT) can each see their relevant view. *** ## What to Look For ### Brand Perception — Provider Comparison By switching the provider filter in Brand Perception, you can generate separate attribute maps for each AI model. The comparison reveals structural differences in how each model has "learned" your brand. | Element | What it tells you | | --------------------------------- | ------------------------------------------------------------------------------------------------------- | | **Shared attributes** | Claims that all providers agree on — these are stable and well-sourced | | **Provider-exclusive attributes** | Claims that only one model makes — often reflect unique training data or retrieval sources | | **Divergent strength scores** | An attribute that scores 80 on Perplexity and 20 on ChatGPT — signals a retrieval vs. training data gap | | **Narrative tone differences** | One provider may describe you confidently; another may hedge with "reportedly" or "claims to" | ### GEO Matrix — Provider Columns The GEO Matrix shows your visibility score per prompt and per provider in a grid. It is the fastest way to spot systematic divergence: if your scores are consistently low in one provider's column, that provider's representation of your brand is weaker or different from others. | Element | What it tells you | | ------------------------ | ------------------------------------------------------------------------------- | | **Column score average** | Your overall standing with a given provider across all monitored prompts | | **Row outliers** | Prompts where one provider scores you significantly higher or lower than others | | **Blank cells** | Prompts where a provider did not mention your brand at all — a hard absence | > **Pro Tip**: Combine **Brand Perception** attribute maps with **GEO Matrix** column averages. A provider where you have low visibility AND divergent attributes is the highest-priority provider to focus remediation efforts on — it both mentions you less and describes you differently when it does. ### Filters That Help | Filter | How to use it for this question | | --------------- | -------------------------------------------------------------------------------------------------------------------------------------- | | **Provider** | The primary filter for this analysis — switch between providers to compare attribute maps | | **Topic / Tag** | Focus the comparison on a specific product area or use case to avoid mixing signals from different positioning contexts | | **Period** | Check whether divergence has narrowed or widened over time — a provider that recently updated its model may have converged with others | *** ## How to Interpret the Results ### Good result All monitored providers describe your brand using the same 3–5 core attributes, even if the phrasing differs. Sentiment scores vary by at most 10–15 points between providers. No single provider makes unique factual claims about your brand that others do not. The GEO Matrix shows consistent visibility scores across provider columns for your priority prompts. ### Needs attention One provider consistently describes your brand using outdated information (old product names, pre-acquisition details, former leadership). A retrieval-based provider like Perplexity shows dramatically different attributes from training-based providers, suggesting that what is being indexed and retrieved about your brand differs significantly from what was in training data. Sentiment is positive on one provider and neutral or negative on another, creating inconsistent experiences for users depending on which AI tool they use. Differences between retrieval-based providers (Perplexity, which uses live web search) and training-based providers (ChatGPT, Claude, which rely on training cutoffs) are expected and do not necessarily indicate a problem. Retrieval-based providers reflect what is currently published about your brand; training-based providers reflect a historical snapshot. Treat them as two different audiences requiring two different content strategies. *** ## Example > **Scenario**: A mid-size gaming studio known for its narrative-driven RPGs wants to understand why Perplexity users seem familiar with their recent titles while ChatGPT users only reference a game they released five years ago. 1. Open **Cockpit > GEO Matrix**. Compare the Perplexity column versus the ChatGPT column across all monitored prompts. Perplexity average visibility: 72. ChatGPT average visibility: 29. The gap is systematic — not limited to a single prompt or game title. 2. Open **Analyze > Perception**, filter by Provider = "Perplexity". The attribute map shows strong signals for "narrative RPG", "award-winning writing", "open-world exploration", and "modding community". These match the studio's current portfolio and recent releases. 3. Switch Provider filter to "ChatGPT". The attribute map only shows "indie RPG" with significant strength (score: 39) and a reference to their 2021 debut title. Their two newer, critically acclaimed titles are absent from ChatGPT's brand perception entirely. 4. Open **Monitor > Response Analysis**, filter by Provider = "ChatGPT" and a high-priority prompt like "best story-driven RPGs 2025". Read the responses — the studio appears in 2 of 14 responses, both times as a footnote mentioning only the older game. 5. Conclusion: ChatGPT's training data predates the studio's breakout period, and the studio lacks presence in the high-authority gaming publications and wikis that were well-indexed before the training cutoff. Action: pursue features in long-standing gaming outlets (IGN, PC Gamer, Giant Bomb), contribute to community wikis, and create evergreen developer blog content that will be picked up in future model updates. *** ## Go Further Export how each AI provider describes your brand as CSV to identify narrative inconsistencies Build a cross-provider narrative consistency tracker in Looker Studio using the answer-details data source Use the answers endpoint to programmatically compare how different AI providers describe your brand ## Related Questions Prioritize which provider inconsistencies to address first based on traffic and reach. Read the verbatim responses before analyzing cross-provider patterns. Take action on the providers where consistency and visibility are lowest. Once you understand cross-provider divergence, assess overall alignment with intended positioning.