Finding and Analyzing the Sources LLMs Cite with DataForSEO AI Optimization API

Search for “email marketing software” in ChatGPT, and you’ll probably see a list of linked sources included in the answer. Run that same query on Google, and its AI Overview may link to another set of pages. Checking one prompt is easy. The problem starts when you need citation data across an entire market rather than a single query. Copying links out of hundreds of answers by hand quickly becomes unmanageable. Building and maintaining a separate scraper for every AI platform is not a better option, as it takes a lot of time and resources. Instead, you can use three endpoints in our AI Optimization API to skip both problems and get a structured list of citations and related data for any topic or domain.

Why AI citations are hard to collect at scale

When an AI assistant answers a question, it usually lists the sources it drew from. You can see those citations one answer at a time. What you can’t see from inside the chat window is the aggregate: across every way people phrase “email marketing software,” which domains and pages get cited most, and how much AI search demand sits behind that topic.

That bigger picture is the part that matters for GEO. A single citation is only one data point. A list of citations across a whole topic is a map of sources from which AI systems currently pull their information. Our LLM Mentions API builds that map: it returns mentions and citations from Google’s AI Overviews and ChatGPT as structured JSON, so you make one request with a topic or domain and get all relevant AI answers instead of reading them one by one. Three endpoints will help you handle most of the workflow.

Step 1: See which domains AI cites most for your topic (Top Mentioned Domains)

Start broad. The Top Mentioned Domains endpoint takes a target (a keyword/topic, a domain, or a mix of up to 10 entities) and returns the domains cited most often in AI answers for it. Every domain comes back with a mentions count and an ai_search_volume metric, so you can sort the data by frequency or by search demand.

The example request below asks for the domains cited for “email marketing software” in the US, scoped to the sources the models actually linked to.

[
    {
        "target": [
            {
                "keyword": "email marketing software",
                "search_scope": [
                    "any"
                ],
                "search_filter": "include"
            }
        ],
        "platform": "google",
        "language_code": "en",
        "location_code": 2840,
        "limit": 5,
        "offset": 0,
        "links_scope": "sources"
    }
]

Endpoint used: Top Mentioned Domains Lite docs.

The response groups the dataset by domain. Here’s an illustrative slice of what comes back in the results of the Lite endpoint (a simplified version of the full Top Mentioned Domains endpoint).

{
    "version": "0.1.20260713",
    "status_code": 20000,
    "status_message": "Ok.",
    "time": "1.0997 sec.",
    "cost": 0.105,
    "tasks_count": 1,
    "tasks_error": 0,
    "tasks": [
        {
            "id": "07161807-8284-0658-0000-a445f0d37241",
            "status_code": 20000,
            "status_message": "Ok.",
            "time": "1.0282 sec.",
            "cost": 0.105,
            "result_count": 1,
            "path": [
                "v3",
                "ai_optimization",
                "llm_mentions",
                "top_mentioned_domains_lite",
                "live"
            ],
            "data": {
                "api": "ai_optimization",
                "function": "top_mentioned_domains_lite",
                "target": [
                    {
                        "keyword": "email marketing software",
                        "search_scope": [
                            "any"
                        ],
                        "search_filter": "include"
                    }
                ],
                "platform": "google",
                "language_code": "en",
                "location_code": 2840,
                "limit": 5,
                "offset": 0,
                "links_scope": "sources"
            },
            "result": [
                {
                    "total_count": 3281,
                    "offset": 0,
                    "items_count": 5,
                    "aggregated_metrics": null,
                    "items": [
                        {
                            "domain": "www.youtube.com",
                            "location": 2840,
                            "language": "en",
                            "platform": "google",
                            "metrics": {
                                "mentions": 15456,
                                "ai_search_volume": 5779200
                            }
                        },
                        {
                            "domain": "www.reddit.com",
                            "location": 2840,
                            "language": "en",
                            "platform": "google",
                            "metrics": {
                                "mentions": 8288,
                                "ai_search_volume": 8377600
                            }
                        },
                        {
                            "domain": "www.indeed.com",
                            "location": 2840,
                            "language": "en",
                            "platform": "google",
                            "metrics": {
                                "mentions": 4896,
                                "ai_search_volume": 2076480
                            }
                        },
                        {
                            "domain": "www.salesforce.com",
                            "location": 2840,
                            "language": "en",
                            "platform": "google",
                            "metrics": {
                                "mentions": 3232,
                                "ai_search_volume": 5817600
                            }
                        },
                        {
                            "domain": "www.linkedin.com",
                            "location": 2840,
                            "language": "en",
                            "platform": "google",
                            "metrics": {
                                "mentions": 1952,
                                "ai_search_volume": 636480
                            }
                        }
                    ]
                }
            ]
        }
    ]
}

This data is easy to organize into a neat table.

These are the domains AI cites most often for the topic. In practice, that list tells you two things at once: which publishers shape the AI answer, and which competitors already get pulled in as sources. Both matter for a GEO strategy. You go after the publishers for brand mentions, and you track the competitors to benchmark against them.

Step 2: Pinpoint the exact pages worth reviewing or earning a link from (Top Mentioned Pages)

The Top Mentioned Pages endpoint takes the same target options but groups the results by URL rather than by domain, so you get the exact pages AI systems cite. Keep the request identical and switch the endpoint path to api.dataforseo.com/v3/ai_optimization/llm_mentions/top_mentioned_pages/live.

The response returns page URLs with their citations and ai_search_volume metrics:

A domain-level view could tell you Salesforce or HubSpot are influential. The page-level view tells you which specific content is doing the work. These are the pages worth reviewing to see what AI systems reward, and which ones are worth earning a mention or a link on. For the full list of response parameters and their descriptions, check the Top Mentioned Pages docs.

Step 3: See the competitive context (brands and categories in AI answers)

Citation data shows where AI answers get their information, but GEO analysis also requires an additional view: which brands and categories AI systems mention most often. Instead of grouping results by URL or domain, Top Mentioned Brands and Top Mentioned Brand Categories rank the brands and product categories with the strongest visibility across your target topics.

For “email marketing software,” the Top Mentioned Brands endpoint returns names like Mailchimp, HubSpot, Klaviyo, Brevo, Constant Contact, and ActiveCampaign, each with its mention frequency and AI search volume. The Top Mentioned Brand Categories endpoint groups the data a level up, into labels like “email marketing platforms,” “marketing automation software,” and “CRM suites,” so you can see which product framings are more popular.

GEO teams can use it to check whether their brand is associated with the right categories, refine product positioning, and strengthen the signals that connect the brand to high-value topics.

Turning three lists into one GEO strategy

On their own, each endpoint answers a narrow question. Run all three against the same topic and they add up to one repeatable workflow:

• Top Mentioned Domains shows where AI gets its information, so you know which sites to pitch, monitor, or build a presence on.
• Top Mentioned Pages gives the exact URLs shaping the answer, so you know what to review and where a single link or mention could change what the model cites.
• Top Mentioned Brands and Brand Categories shows who AI already names, so you can measure topic demand and your share of voice and find the gaps.

Say you run this monthly for a client in the “email marketing software” space. A new roundup climbs the Top Mentioned Pages list. A competitor you weren’t tracking shows up among the most-mentioned brands. In just two requests, you have two actionable signals: a page worth targeting for outreach and a competitor worth investigating for content, positioning, or comparison opportunities. No manual review of multiple individual AI answers required. And since the results come back as structured JSON, you can easily integrate them straight into a dashboard, a client report, or a scheduled monitoring job. If you want to fine-tune topic scope, targeting and filtering, our guide to defining targets covers the options.

Start mapping the sources behind AI answers

Seeing the sources behind one AI answer is easy. The real challenge is identifying the domains, pages, brands, and categories that appear repeatedly across an entire topic. Our AI Optimization API turns that scattered data into a ranked, structured view with the Top Mentioned Domains, Pages, Brands, and Brand Categories data from Google’s AI Overviews and ChatGPT. Pricing is pay-as-you-go, starting at $1.1 per 1,000 data rows. Choose a topic, run the request, and see which sources and brands are earning visibility in AI answers.

Try for free