How to find the third-party websites ChatGPT cites for competing brands
Sep 2, 2026
A DTC brand can find the third-party websites ChatGPT cites for competing brands by running a consistent set of category, comparison, and use-case prompts, recording every cited URL and the claim it supports, then grouping the results by source type and buyer need. The resulting map shows where competitors have external evidence, where your brand is absent or misrepresented, and which content, partnerships, or product information to improve first.
Start with a prompt set that represents buying decisions
Build the audit around the questions shoppers ask before buying, not around a list of brand names. A useful prompt set includes category discovery, comparisons, use cases, constraints, and trust questions.
Use prompts such as:
- “What are the best [product category] options for [specific customer or use case]?”
- “Compare [competitor A], [competitor B], and [your brand] for [use case].”
- “Which [product] is best for someone who needs [constraint]?”
- “What should I look for when buying [product category]?”
- “Which brands offer [attribute, feature, material, policy, or service]?”
- “What are the drawbacks of [competitor]?”
Include the language your customers actually use. Product-page searches, support tickets, reviews, site search terms, and conversations with sales or customer service can supply that language. A prompt about “best hiking shoes for wide feet” produces a more actionable citation audit than a generic prompt about “hiking shoes.”
Keep the first version manageable. Cover the main categories, high-margin products, important comparison sets, and the questions most likely to influence a purchase. Save the exact wording so you can rerun the same audit after making changes.
Record citations separately from brand mentions
A competitor mention and a competitor citation are different findings. Record both: ChatGPT may name a brand in its answer, cite a third-party page about that brand, cite the brand’s own site, or do some combination of these.
For every prompt and response, capture:
| Field | What to record |
|---|---|
| Prompt | The exact question, including market or use case |
| Brand mentions | Which brands appear in the answer text |
| Recommendation | Which brands ChatGPT favors, and for what reason |
| Cited URL | The complete page URL shown as a source |
| Source domain | The website publishing that page |
| Source type | Review, comparison, retailer, marketplace, publisher, community, creator, partner, or other |
| Supported claim | The attribute or evidence the page appears to support |
| Brand representation | Accurate, incomplete, outdated, or unfavorable |
| Your presence | Present, absent, or present but not competitive |
Do not count a domain merely because it appears somewhere in a response. Tie it to a specific answer and claim. A product comparison may support a recommendation; a retailer listing may confirm availability; a sizing guide may explain fit. Those sources create different opportunities and should not be treated as interchangeable.
Run the same prompt more than once and record the date, model experience, market, and any relevant settings. The output can change, so one response is a lead, not a reliable baseline. Look for patterns across related prompts rather than declaring a source important because it appeared once.
Build a competitor citation-source map
Group the URLs into source families, then compare the families that recur for competitors but not for your brand. This turns a long export of links into a picture of how the category is being explained.
Useful source families include:
- Independent reviews and specialist publications: Evidence about performance, quality, fit, testing, or expert judgment.
- Comparison and “best of” pages: Shortlists that define which brands belong in a category or use case.
- Retailers and marketplaces: Product availability, specifications, customer reviews, and alternatives.
- Creators and video platforms: Demonstrations, real-world use, fit, setup, and subjective experience.
- Community discussions: Unfiltered objections, recurring problems, and language customers use to describe outcomes.
- Partners and professional organizations: Context, compatibility, installation, training, or use-case validation.
- Your competitors’ owned pages: Product specifications, policies, guides, and evidence that other sources may repeat.
A useful map has three levels: domain, page, and role. For example, a specialist publication may be a recurring domain, one comparison article may be the recurring page, and “best for cold-weather use” may be its role in the answer.
This distinction matters because a domain-level target is not automatically a page-level target. If a competitor wins through a single strong comparison page, producing a broad blog series may be less useful than creating accurate evidence for the same buying question and earning inclusion in relevant comparisons.
A broader analysis of AI search citations similarly recommends documenting the source domain, cited URL, source type, brand presence, link presence, and representation accuracy separately by platform. See Aleyda Solis’s AI search citations analysis for that framework.
Turn citation gaps into an AI visibility plan
Your plan should connect each citation gap to a buyer question, an evidence problem, and an owner. “Get more mentions” is not an action; “give independent reviewers verifiable evidence for the wide-fit use case” is.
Use a working table like this:
| Finding | Likely problem | Action | Owner | Proof of progress |
|---|---|---|---|---|
| Competitors repeatedly appear in specialist comparisons | Your category credentials or product differences are hard to verify | Create a fact sheet and pitch accurate product evidence to relevant publishers | Content or PR | Your brand is evaluated in the next recurring comparison set |
| A competitor wins “best for” a specific use case | The competitor has clearer proof of that attribute | Publish a focused guide, improve product details, and gather credible supporting evidence | Ecommerce and product marketing | Your brand is associated with the use case in relevant prompts |
| Retailer pages are cited for specifications you also sell | Your product data is incomplete or inconsistent | Standardize specifications, compatibility, sizing, materials, and policies across owned and retail listings | Ecommerce operations | Fewer conflicting descriptions; more accurate citations |
| Community pages surface a recurring objection about your category | The objection is unresolved or poorly explained | Address it directly in product pages, guides, FAQs, and customer education | Product and CX | The answer reflects the limitation and the appropriate product choice |
| A third-party page misrepresents your product | External evidence is stale or incorrect | Contact the publisher with clear, verifiable corrections; update your own source material | Brand or PR | Fewer inaccurate descriptions in the citation set |
Prioritize work using four filters:
- Commercial importance: Does the prompt relate to a high-value category, product, or use case?
- Competitive pressure: Does the same competitor and source recur across several relevant prompts?
- Fixability: Can your team improve the evidence, product data, or external relationship?
- Customer value: Will the work answer a real question better, even outside ChatGPT?
Do not copy a competitor’s wording or manufacture reviews. Use the citation map to identify the evidence buyers need, then publish accurate information and pursue legitimate third-party validation.
Measure visibility without reducing it to one score
Track brand presence, recommendation share, citation frequency, source mix, and representation accuracy as separate measures. A brand can be named without being cited, cited without being named in the answer, or mentioned in a context that does not help a buyer.
A practical monthly report can include:
- The percentage of tracked prompts where your brand is mentioned.
- The percentage where your domain or a third-party page about you is cited.
- The competitors appearing most often beside your brand.
- The third-party domains and pages recurring in competitor answers.
- The proportion of citations from owned versus external sources.
- The buyer topics where your brand is absent, weakly described, or inaccurately described.
- Changes in AI-referred sessions and assisted conversions, interpreted cautiously because referral data may be incomplete.
Keep ChatGPT separate from other systems if the question is specifically about ChatGPT citations. A combined score can hide a platform-specific weakness. Segment results by category, intent, geography, and product where the sample supports it.
Review citation changes alongside business outcomes. A new citation is useful only if it improves how the brand is understood, reaches a relevant buyer question, or contributes to qualified discovery. The goal is not to collect domains; it is to build a stronger body of evidence around the decisions your customers make.
Use Anagram when the audit needs an ongoing workflow
Anagram’s AI Visibility product is designed to show how a brand appears in ChatGPT, compare it with competitors, and surface the sources shaping AI answers. Its AI Visibility page describes visibility tracking across relevant prompts, competitive share of voice, and citation pages.
That can replace a recurring manual spreadsheet when your team needs to monitor competitor recommendations, citation sources, and topic gaps over time. Before adopting any monitoring workflow, confirm how it defines a prompt set, distinguishes a mention from a recommendation, calculates competitive share of voice, and handles repeated runs.
Anagram also connects the external audit to onsite shopper questions. Its homepage describes a branded Site Agent for conversational product support and a workflow that helps teams see what shoppers care about and where friction occurs. Those questions can help validate whether a citation gap reflects a genuine information need, such as fit, compatibility, ingredients, setup, or product selection.
A lean team can start with a fixed prompt set and a spreadsheet. A growing DTC brand should consider a monitoring system when the catalog, competitor set, markets, or reporting cadence makes manual collection unreliable. In either case, the deliverable is the same: a prioritized list of buyer questions, recurring citation sources, evidence gaps, and specific actions that make your brand easier to understand and recommend.