We asked AI who the best health insurance brokers in Kansas City are. One name showed up three times. Is it yours?

Asked daily, word for word

Counts read "answers naming it / answers that returned text", per engine. Names are classified when they appear in at least two answers. Counting rule: a business counts for an engine when its full name, written with the same capital letters, appears in that engine's saved answer text, so a count means named in the answer text. The matching is the homepage explorer's name matching with an added capital-letter check on each word of the name, run over this page's own window and counted per answer. A business listed as an example counts, and a count does not mean the engine picked it. Rows overlap (a chain and its locations are separate rows), so rows are not added together. Until 2026-10-21 this page counted from the extracted business lists, which under-counted plain-text mentions by different amounts per engine. Details: Method, DQ-5.

Names in the earlier extracted lists that this rule leaves out: Medicare KC (generic phrase).

Names AI gave, 2026-08-09 to 2026-10-27

Sorted by total namings across the engines (used only to order the rows; no total is printed). This is a count, not a ranking and not an endorsement.

Names not yet classified

These names were not in the business-or-not labeling pass dated 2026-10-07. They are shown as written, in answers where at least two named them, and have not been classified, so one may be a place, a term or a tool rather than a business. They are not counted in the table above.

Which engines answered, and how

Each door is a different way of asking. Settings are listed, failures are counted, and a door that did not answer says so.

What Perplexity read

Perplexity is the one API door that returns its sources. For this question it cited these domains, counted as answers citing the domain out of 77 Perplexity answers.

OpenAI, Anthropic and Gemini answers carry no source list in this record.

Search results for the same question

The AI Overview reference field exists from 2026-10-03. It holds links in 16 of 18 searches since then. Domains most often linked: bbs-insurance.com (13/18), health-insurance-kansas-city.com (13/18), jeffdavidinsurance.com (13/18), aishealthadvisors.com (12/18), bluekc.com (4/18).

Whether Google showed an AI Overview is a best-effort flag and is not reported here as a measurement. Only the links captured are counted.

Is Namebeam cited here?

Namebeam pages in Perplexity sources for this question: 19 of 77 answers (this page: 19 of 77). Namebeam pages in Google top results: 0 of 47 searches. Counted from the same fields as everything above. A second question about the same market, "What is the best health insurance broker in Kansas City MO for someone who is self-employed?", was also asked daily: this page appeared in Perplexity sources in 18 of 85 of those answers.

This page is itself part of the test. On most days the same question is asked again, and the count above is updated from the record. Whether this page is ever cited is reported with its denominator, up or down.

Hand capture, 2026-07-31

Before the daily record began, a hand capture dated 2026-07-31 was logged for this question. It is kept here unchanged and is the earliest dated entry. Each block names the engine and account type used. Consumer apps and the API can answer differently.

Check it yourself

Every answer on this page is one JSON file in the public record. The index for this page lists each row with its file name and SHA-256: namebeam.ai/data/e1/kansas-city-health-insurance.json. The files live in the repository namebeam-visibility-index, corpus/e1. To check a hash, hash the file as served: curl -s https://raw.githubusercontent.com/mrcrtr1979-droid/namebeam-visibility-index/main/corpus/e1/<file> | sha256sum. 400 rows, 2026-08-09 to 2026-10-27.

How this was run. The prompt is the exact text above, sent unchanged once per day to each door, 2026-08-09 to 2026-10-27. Each row of the record keeps the date, the door, the run status and the answer text exactly as returned. Rows that failed are counted as failed in the table above and are not dropped. Business names are read from the answer text: bold names and names listed by the engine are collected, each must appear in the answer verbatim, and a language-model pass labels each as a business, an organization that is not a provider choice, or not a name. A name is classified only if it appears in at least two answers to this prompt, so a business named once is not in the table. Written variants of one name are merged under one label, and the data file lists every written form under each label. The label pass can miss or wrongly merge a name, so the raw files below are the record and this table is a reading of it. Counts are answers naming a business out of answers that returned text, per door and in total. The table is sorted by count, which is not a ranking of quality and not an endorsement of any business. API answers can differ from the logged-in consumer apps; the run channel is recorded on every row. Perplexity returned a median of 19 sources per answer. The question and the engines are the same every day; the engines themselves change over time. Namebeam is an educational and marketing service and does not guarantee rankings, traffic, or revenue.

Namebeam, a Carter Enterprise property. This is the first leg.

• All questions in the record