{
  "format_version": 1,
  "canonical": "https://rotgar.com/medical/resources/chatgpt-medical-provider-shortlists",
  "path": "/medical/resources/chatgpt-medical-provider-shortlists",
  "title": "How ChatGPT Builds Medical Provider Shortlists | US Study | Rotgar",
  "h1": "ChatGPT Named Most Final-Shortlist Medical Providers in Its First Search Queries",
  "description": "Across 180 commercial medical-provider responses, 85% of final provider positions were already named in ChatGPT's first returned search-action queries.",
  "language": "en",
  "kind": "research",
  "source_html_sha256": "3d7b910cf71b48c72582ff3cb92d1fca4a37e00d916c75f40b6ed515b4d72e7f",
  "content_markdown_sha256": "08dcd41167fe2dd9d1f2505863f0428d994186d9defd6f0ffa7d5962848638f3",
  "alternatives": {
    "markdown": "https://rotgar.com/machine/pages/medical/resources/chatgpt-medical-provider-shortlists.md",
    "json": "https://rotgar.com/machine/pages/medical/resources/chatgpt-medical-provider-shortlists.json"
  },
  "content": {
    "markdown": "Rotgar Research · Medical\n\n# ChatGPT Named Most Final-Shortlist Medical Providers in Its First Search Queries\n\nAcross 180 commercial medical-provider responses, 85% of final provider positions were already named in ChatGPT's first returned search-action queries.\n\nBy **Evgeniy Yudin**, Founder and Strategy Lead Methodology reviewed by **Boris Teplyakov**, SEO Lead Published August 18, 2026 US English\n\nBoris Teplyakov reviewed the research methodology as SEO Lead. This was not a clinical subject-matter review or an assessment of the providers named in the responses.\n\n[Read the findings](https://rotgar.com/medical/resources/chatgpt-medical-provider-shortlists#answer) [Review the methodology](https://rotgar.com/medical/resources/chatgpt-medical-provider-shortlists#methodology)\n\nDirect answer\n\n## Most final provider positions were already visible in the first returned search action\n\nAcross 180 commercial medical provider-selection responses, **85.0%** of final provider positions were occupied by organizations already named in the query strings of ChatGPT’s first returned web-search action. The connection was strongest for broad comparisons and weakest for cost-and-insurance questions.\n\n**85.0%**612 of 720 final positions already named\n\n**179/180** responses with at least one provider in the first action\n\n**36.1%**cost-and-insurance positions absent from the first action\n\nStudy at a glance\n\n## A frozen US medical provider-selection panel\n\nResponses\n\n180\n\nPrompt cells\n\n72\n\nService lines\n\n6\n\nUS metros\n\n3\n\nPatient tasks\n\n3\n\nFieldwork\n\nAugust 18, 2026\n\nExplore the three patient tasks\n\n## The shortlist path changed with the commercial question\n\nBroad comparison **93.0% retained**\n\nResponses\n\n108\n\nFinal positions absent from first action\n\n7.4%\n\nFinal shortlist stability\n\n77.9%\n\nAppointment access **87.0% retained**\n\nResponses\n\n36\n\nFinal positions absent from first action\n\n16.7%\n\nFinal shortlist stability\n\n61.2%\n\nCost and insurance **72.4% retained**\n\nResponses\n\n36\n\nFinal positions absent from first action\n\n36.1%\n\nFinal shortlist stability\n\n50.4%\n\nHow to read this study\n\n## An observational shortlist study, not a provider rating\n\n**Observable stages only**\n\nThe study compares provider names in the first returned search action with providers in the final answer. It does not expose hidden reasoning or establish where initial candidates originated.\n\n**Visibility, not clinical quality**\n\nInclusion is a measured visibility outcome. It is not a clinical-quality assessment, a patient recommendation or evidence that one provider is better than another.\n\n**Dated API configuration**\n\nThe results describe one August 18, 2026 Responses API configuration. They are not a test of every model, consumer interface or personalized account state.\n\nThe answer in brief\n\n## Most final-shortlist providers were already present in the first observable query set\n\nAcross the 180 responses, 612 of 720 final provider positions, or 85.0%, were occupied by organizations already named in the query strings of the first returned web-search action.\n\nAt least one provider organization appeared at that stage in 179 of 180 responses. Each first action contained four query strings.\n\nThe relationship between that first observable candidate set and the final answer depended on the patient’s task. Broad comparison answers retained 93.0% of provider mentions from the first search action. Only 7.4% of their final provider positions contained an organization absent from that first action. For cost-and-insurance questions, retention fell to 72.4%, while 36.1% of final provider positions contained an organization absent from the first action.\n\nThe final shortlists also became less consistent across repeated runs. Mean provider-set similarity was 77.9% for broad comparisons, 61.2% for appointment-access questions and 50.4% for cost-and-insurance questions.\n\nFor a health system CMO, this creates two measurement questions: Did the organization enter the first observable candidate set? Did it remain in, or appear only in, the final answer? A single AI visibility score cannot answer both.\n\nWhat we tested\n\n## A frozen set of commercial provider-selection prompts\n\nWe created brand-free commercial questions asking ChatGPT to return exactly four local providers. The study covered:\n\n- six service lines: fertility, orthopedics, ophthalmology, urgent care, primary care and bariatric surgery;\n- three metros: New York, Chicago and Houston;\n- three patient tasks: broad provider comparison, appointment access and cost or insurance;\n- 180 fresh responses across 72 frozen prompt cells.\n\nEach response began without prior conversation state. The broad comparison used two neutral prompt phrasings and three repeats per service-line and metro cell. The appointment-access and cost-and-insurance extensions used two repeats per cell.\n\nWe captured the four query strings in the first returned web-search action and the four providers in the final answer. We then compared normalized provider entities at the two stages.\n\nFirst action\n\n## Provider names appeared in 179 of 180 first search actions\n\nThe user prompts contained no provider brands. Even so, at least one provider organization appeared in the first returned search action in:\n\n- 108 of 108 broad comparison responses;\n- 36 of 36 appointment-access responses;\n- 35 of 36 cost-and-insurance responses.\n\nThis is not a claim that 99.4% of providers were chosen before search. It is a response-level result: 179 of 180 first search actions contained at least one provider among their four query strings.\n\nThe query strings are observable API output. They show the names used at the first search step, but not where those names originated.\n\nPatient task\n\n## The patient’s task changed the shortlist path\n\n**Table 1. Shortlist-path metrics by patient task**\n\n| Patient task | Initial provider mentions retained in the final answer | Final provider positions absent from the first action | Final shortlist stability |\n| --- | --- | --- | --- |\n| Broad comparison | 93.0% (400 of 430) | 7.4% (32 of 432) | 77.9% |\n| Appointment access | 87.0% (120 of 138) | 16.7% (24 of 144) | 61.2% |\n| Cost and insurance | 72.4% (92 of 127) | 36.1% (52 of 144) | 50.4% |\n\nThe first two percentages in each row use different denominators. Retention starts with provider mentions in the first search action. Later addition starts with provider positions in the final answer. They are related measures, not complementary shares.\n\nBroad comparison prompts were mostly candidate-led in this collection: nearly all initial provider mentions survived into the final answer, and relatively few final positions were filled by providers absent from the first action.\n\nCost-and-insurance prompts produced a larger difference between the first observable set and the final answer. More than one-third of final provider positions were occupied by organizations absent from the first action. Appointment-access prompts sat between the two.\n\nFor reporting, do not combine these tasks into one score. A health system can look established in a general comparison and still be inconsistent when a patient asks about insurance, self-pay, financing, scheduling or locations.\n\n![Two-panel chart comparing three patient tasks. Initial provider mentions retained in final answers were 93.0% for broad comparison, 87.0% for appointment access and 72.4% for cost and insurance. Final providers absent from the first action were 7.4%, 16.7% and 36.1%, respectively. The panels use different denominators.](https://rotgar.com/images/research/chatgpt-medical-provider-shortlists/scenario-paths-1400x900.png)\n\nFigure 1. The relationship between the first returned search action and the final provider list varied by commercial task. Retention uses first-action provider mentions as its denominator; first-action-absent presence uses final provider positions.\n\nCost and insurance\n\n## Cost answers included first-action-absent providers across all six service lines\n\nIn the cost-and-insurance sample, the share of final provider positions absent from the first action ranged from 25.0% to 50.0%:\n\n**Table 2. Final cost-and-insurance provider positions absent from the first action**\n\n| Service line | Final providers absent from the first action |\n| --- | --- |\n| Bariatric surgery | 50.0% (12 of 24) |\n| Orthopedics | 41.7% (10 of 24) |\n| Fertility | 37.5% (9 of 24) |\n| Ophthalmology | 37.5% (9 of 24) |\n| Primary care | 25.0% (6 of 24) |\n| Urgent care | 25.0% (6 of 24) |\n\nEach service-line cut contains six responses across three prompt cells, so the ordering is exploratory. The practical point is broader: every service line tested produced final cost-and-insurance answers containing providers absent from the first action.\n\nFor a CMO, cost visibility needs its own audit. Review whether the public web gives a patient clear, current answers about accepted insurance, self-pay, financing, estimate requests and the next step for contacting the organization. Review both owned pages and external pages that answer the same commercial question.\n\n![Horizontal bars showing the share of final cost-and-insurance provider positions absent from the first search action: bariatric surgery 50.0%, orthopedics 41.7%, fertility 37.5%, ophthalmology 37.5%, primary care 25.0% and urgent care 25.0%.](https://rotgar.com/images/research/chatgpt-medical-provider-shortlists/cost-service-lines-1400x900.png)\n\nFigure 3. Every tested service line produced cost-and-insurance answers containing providers absent from the first action. Each service-line cut contains six responses across three prompt cells and is exploratory.\n\nRepeatability\n\n## One answer is not a stable benchmark\n\nWe measured stability with pairwise Jaccard similarity. A score of 100% means repeated responses returned the same provider set; 0% means they shared none.\n\nThe final provider set averaged 77.9% similarity for broad comparisons, 61.2% for appointment access and 50.4% for cost and insurance. The cost shortlist was therefore the least consistent of the three task types in this collection.\n\nOne screenshot can show whether a provider appeared once. It cannot show whether the result persists. A useful monitoring program freezes the exact question and location, repeats the run, records the configuration and reports appearance frequency alongside shortlist similarity.\n\n![Paired dot plot of mean provider-set similarity. First-action and final-answer similarity were 78.4% and 77.9% for broad comparison, 67.1% and 61.2% for appointment access, and 65.6% and 50.4% for cost and insurance.](https://rotgar.com/images/research/chatgpt-medical-provider-shortlists/shortlist-stability-1400x780.png)\n\nFigure 2. Final provider lists were less stable across identical repeated prompts for appointment-access and cost-and-insurance questions than for broad comparisons. Similarity is mean within-cell pairwise Jaccard.\n\nFor health system CMOs\n\n## A measurement framework for health system CMOs\n\nTrack the stages separately:\n\n1. **First-action presence.** Did the provider appear in any query string in the first returned search action?\n2. **Final inclusion.** Did the provider appear in the final four-provider answer?\n3. **Retention.** When the provider appeared in the first action, did it remain in the final answer?\n4. **Final-only presence.** When the provider appeared in the final answer, was it absent from the first action?\n5. **Stability.** How often did it appear across repeated runs of the same frozen prompt?\n\nBuild separate views for broad comparison, appointment access and cost or insurance. Then segment by service line and metro. This keeps a strong general-comparison result from masking a weak or unstable result for a high-intent patient task.\n\nUse the result to choose the next audit:\n\n- If a provider appears early and stays, document the commercial questions for which that pattern holds.\n- If it appears early but drops out, compare the final answer’s retained providers and their supporting public information.\n- If it appears only in the final answer, identify which public pages answer that specific patient task.\n- If it rarely appears across repeats, treat the result as unstable rather than as a fixed rank.\n\nThese are measurement paths, not a universal optimization recipe. The study shows where shortlist changes occurred; the page and source review explains what to investigate next.\n\nContext\n\n## How this extends earlier work\n\nThis study was prompted by [Suganthan Mohanadasan’s analysis of provider and product names in ChatGPT’s first search queries](https://suganthan.com/blog/chatgpt-decides-before-it-searches/). That work examined a logged-in ChatGPT account across a mixed, largely commercial query set and explicitly treated its percentages as directional.\n\nWe tested a narrower question in health care: whether provider organizations appeared in the first returned search action for brand-free medical provider-selection prompts, whether they persisted into the final answer and how the pattern changed across commercial tasks. The two studies use different surfaces and samples, so their percentages should not be pooled.\n\nMethodology\n\n## A dated observational snapshot\n\nThe analytical sample contains 180 completed responses from August 18, 2026. Collection used the OpenAI Responses API with the requested and returned model alias `chat-latest`, required `web_search`, low search context, fixed approximate metro location, no prior conversation state, `store=false` and a 1,200-token output ceiling. [OpenAI describes web search](https://developers.openai.com/api/docs/guides/tools-web-search) as a Responses API tool that lets models access current web information and return sourced answers.\n\nThe response was the observational unit. Provider names were normalized to organizations or consumer-facing provider brands. Rule-based extraction was followed by manual adjudication of every unmatched query string and all identified ambiguous positive matches. The analytical sample had complete coverage: four first-action query strings, four final providers and four citation URLs in every response.\n\nRetention equals retained initial provider mentions divided by all normalized provider mentions in the first action. First-action-absent final presence equals final provider entities absent from the first action divided by all final provider entities. Stability is the mean within-cell pairwise Jaccard similarity of normalized provider sets across repeated responses. Ninety-five percent prompt-cell cluster bootstrap intervals used 10,000 replications with the exact prompt cell as the cluster unit; the six fixed scenario seeds are published in the downloadable methodology and scenario data. The resulting intervals were 89.8%–95.8%, 81.2%–92.2% and 66.2%–79.4% for retention; and 4.4%–10.9%, 9.7%–24.3% and 28.5%–44.4% for first-action-absent final presence, in broad comparison, appointment access and cost or insurance respectively.\n\nLimits\n\n## What this study does and does not show\n\nThis is a dated snapshot of one `chat-latest` API configuration, not a test of the consumer ChatGPT interface, every model or personalized account behavior. Service-line and metro cuts are descriptive and exploratory. Two repeats in the commercial extensions and three in the broad comparison estimate short-run variation but do not create new prompt families.\n\nThe first returned search action is observable; hidden reasoning and the origin of initial provider candidates are not. Provider inclusion is a visibility outcome, not a clinical-quality rating or a recommendation to patients.\n\nRaw API responses remain private. The page reports reviewed aggregate results and does not publish response identifiers, full answers or encrypted tool content.\n\n**Suggested citation:** Yudin, E. (2026). ChatGPT Named Most Final-Shortlist Medical Providers in Its First Search Queries (Version 1.0) \\[Data set\\]. Rotgar Research. https://doi.org/10.5281/zenodo.22003037\n\nDownloadable research materials\n\n## Download the public v1.0 research package\n\nThe public package contains aggregate and anonymous response-level metrics, frozen prompts, documentation and integrity records under [CC BY 4.0](https://creativecommons.org/licenses/by/4.0/). It excludes raw API responses, returned query text, provider-level mention records, response IDs, credentials and internal collection identifiers.\n\n[Data **Primary CSV** Overall and scenario summary metrics in a compact tabular file.text/csv · 1.6 KB](https://rotgar.com/research-data/chatgpt-medical-provider-shortlists-v1.0/chatgpt-medical-provider-shortlists-v1.0.csv) [Data **Complete public JSON** Machine-readable public tables, metadata, metrics and interpretation boundaries.application/json · 331 KB](https://rotgar.com/research-data/chatgpt-medical-provider-shortlists-v1.0/chatgpt-medical-provider-shortlists-v1.0.json) [Data **Research workbook** Formatted workbook with ten documented research sheets.application/vnd.openxmlformats-officedocument.spreadsheetml.sheet · 62 KB](https://rotgar.com/research-data/chatgpt-medical-provider-shortlists-v1.0/chatgpt-medical-provider-shortlists-v1.0.xlsx) [Reproducibility **Prompt manifest** The 72 frozen, brand-free commercial prompt cells used in the study.text/csv · 46 KB](https://rotgar.com/research-data/chatgpt-medical-provider-shortlists-v1.0/prompt-manifest.csv) [Reproducibility **Stability pairs** The 144 repeat-pair rows used to reproduce provider-set stability without exposing provider names.text/csv · 4.3 KB](https://rotgar.com/research-data/chatgpt-medical-provider-shortlists-v1.0/stability-pairs.csv) [Reproducibility **Data dictionary** Definitions for every field in the public package.text/csv · 7.7 KB](https://rotgar.com/research-data/chatgpt-medical-provider-shortlists-v1.0/data-dictionary.csv) [Documentation **Package README** Scope, direct result, package contents and public-data boundary.text/markdown · 2.6 KB](https://rotgar.com/research-data/chatgpt-medical-provider-shortlists-v1.0/README.md) [Documentation **Methodology** Research design, observable stages, metrics, entity handling and interpretation limits.text/markdown · 3.5 KB](https://rotgar.com/research-data/chatgpt-medical-provider-shortlists-v1.0/METHODOLOGY.md) [Documentation **Dataset metadata** Machine-readable identity, scope, configuration, counts and license metadata.application/json · 4.2 KB](https://rotgar.com/research-data/chatgpt-medical-provider-shortlists-v1.0/dataset-metadata.json) [Verification **Package manifest** Versioned file inventory with bytes, media types, row counts and SHA-256 values.text/plain · 5.4 KB](https://rotgar.com/research-data/chatgpt-medical-provider-shortlists-v1.0/manifest.json) [Verification **SHA-256 checksums** Integrity checksums for the complete public package.text/plain · 2.2 KB](https://rotgar.com/research-data/chatgpt-medical-provider-shortlists-v1.0/SHA256SUMS.txt) [Citation **BibTeX citation** Ready-to-import BibTeX citation metadata.text/plain · 600 B](https://rotgar.com/research-data/chatgpt-medical-provider-shortlists-v1.0/citation.bib) [Citation **Citation File Format** CFF 1.2 metadata for research and repository tools.text/plain · 883 B](https://rotgar.com/research-data/chatgpt-medical-provider-shortlists-v1.0/CITATION.cff)\n\n**Version 1.0** · Published August 18, 2026 · [Manifest](https://rotgar.com/research-data/chatgpt-medical-provider-shortlists-v1.0/manifest.json) · [SHA-256 checksums](https://rotgar.com/research-data/chatgpt-medical-provider-shortlists-v1.0/SHA256SUMS.txt)\n\nCitation and research record\n\n## A traceable, versioned publication record\n\nYudin, E. (2026). ChatGPT Named Most Final-Shortlist Medical Providers in Its First Search Queries (Version 1.0) \\[Data set\\]. Rotgar Research. https://doi.org/10.5281/zenodo.22003037\n\nAuthor\n\nEvgeniy Yudin, Founder and Strategy Lead\n\nMethodology reviewer\n\nBoris Teplyakov, SEO Lead\n\nVersion and license\n\nVersion 1.0; [CC BY 4.0](https://creativecommons.org/licenses/by/4.0/)\n\nVersion DOI\n\n[10.5281/zenodo.22003037](https://doi.org/10.5281/zenodo.22003037)\n\nArchive\n\n[Zenodo record 22003037](https://zenodo.org/records/22003037)\n\nFieldwork\n\nAugust 18, 2026\n\nSurface\n\nOpenAI Responses API with required web\\_search\n\nLanguage and market\n\nUS English; New York, Chicago and Houston\n\nBoris Teplyakov reviewed the research methodology as SEO Lead. This was not a clinical subject-matter review or an assessment of the providers named in the responses.\n\nArtifact provenance\n\n### Approved aggregate evidence\n\nThe page reads headline and segment metrics directly from the reviewed publication bundle and stops building if its approved count contract changes.\n\n**Artifact SHA-256** `11b12c97fa937e384f5006f560b50d4441bca5b907a8bd20bb84c774fdad656c`\n\nEvidence identities\n\n- `public-metrics.json` — approved aggregate metrics\n- `CHART-DATA.csv` — chart-level values and denominators\n- `SITE-PAGE.md` — approved article narrative and limitations\n- `PAGE-METADATA.json` — route, metadata and artifact identity\n\nContinue from the evidence\n\n## Related medical research and measurement pages\n\nAI Visibility Audit\n\n## See how your organization appears in medical AI answers\n\nThe AI visibility audit starts with one clinic or selected location, one priority market and one patient language.\n\n**Best fit** Clinics, hospitals and medical groups in:\n\n- Dental\n- Aesthetic medicine\n- Physiotherapy\n- Fertility & IVF\n- Dermatology\n- Hair transplant\n- Cosmetic plastic surgery\n- Mental health\n- Multi-specialty groups\n\n- Your current visibility and named competitors\n- The sources AI answers cite\n- A prioritized list of improvements to implement\n\nNeed to cover several clinics, locations, countries, markets or patient languages? [Book a 30-minute call ](https://calendly.com/z-rotgar/30min) We’ll discuss the scope for your organization.\n",
    "headings": [
      {
        "level": 1,
        "text": "ChatGPT Named Most Final-Shortlist Medical Providers in Its First Search Queries"
      },
      {
        "level": 2,
        "text": "Most final provider positions were already visible in the first returned search action",
        "id": "answer-summary-heading"
      },
      {
        "level": 2,
        "text": "A frozen US medical provider-selection panel",
        "id": "study-glance-heading"
      },
      {
        "level": 2,
        "text": "The shortlist path changed with the commercial question",
        "id": "scenario-explorer-heading"
      },
      {
        "level": 2,
        "text": "An observational shortlist study, not a provider rating",
        "id": "study-boundary-heading"
      },
      {
        "level": 2,
        "text": "Most final-shortlist providers were already present in the first observable query set",
        "id": "answer-heading"
      },
      {
        "level": 2,
        "text": "A frozen set of commercial provider-selection prompts",
        "id": "tested-heading"
      },
      {
        "level": 2,
        "text": "Provider names appeared in 179 of 180 first search actions",
        "id": "presence-heading"
      },
      {
        "level": 2,
        "text": "The patient’s task changed the shortlist path",
        "id": "task-heading"
      },
      {
        "level": 2,
        "text": "Cost answers included first-action-absent providers across all six service lines",
        "id": "cost-heading"
      },
      {
        "level": 2,
        "text": "One answer is not a stable benchmark",
        "id": "stability-heading"
      },
      {
        "level": 2,
        "text": "A measurement framework for health system CMOs",
        "id": "framework-heading"
      },
      {
        "level": 2,
        "text": "How this extends earlier work",
        "id": "prior-heading"
      },
      {
        "level": 2,
        "text": "A dated observational snapshot",
        "id": "method-heading"
      },
      {
        "level": 2,
        "text": "What this study does and does not show",
        "id": "limits-heading"
      },
      {
        "level": 2,
        "text": "Download the public v1.0 research package",
        "id": "study-downloads-heading"
      },
      {
        "level": 2,
        "text": "A traceable, versioned publication record",
        "id": "study-record-heading"
      },
      {
        "level": 3,
        "text": "Approved aggregate evidence",
        "id": "provenance-heading"
      },
      {
        "level": 2,
        "text": "Related medical research and measurement pages",
        "id": "study-related-heading"
      },
      {
        "level": 2,
        "text": "See how your organization appears in medical AI answers"
      }
    ],
    "tables": [
      {
        "caption": "Table 1. Shortlist-path metrics by patient task",
        "rows": [
          {
            "cells": [
              {
                "text": "Patient task",
                "markdown": "Patient task",
                "header": true,
                "colSpan": 1,
                "rowSpan": 1
              },
              {
                "text": "Initial provider mentions retained in the final answer",
                "markdown": "Initial provider mentions retained in the final answer",
                "header": true,
                "colSpan": 1,
                "rowSpan": 1
              },
              {
                "text": "Final provider positions absent from the first action",
                "markdown": "Final provider positions absent from the first action",
                "header": true,
                "colSpan": 1,
                "rowSpan": 1
              },
              {
                "text": "Final shortlist stability",
                "markdown": "Final shortlist stability",
                "header": true,
                "colSpan": 1,
                "rowSpan": 1
              }
            ]
          },
          {
            "cells": [
              {
                "text": "Broad comparison",
                "markdown": "Broad comparison",
                "header": true,
                "colSpan": 1,
                "rowSpan": 1
              },
              {
                "text": "93.0% (400 of 430)",
                "markdown": "93.0% (400 of 430)",
                "header": false,
                "colSpan": 1,
                "rowSpan": 1
              },
              {
                "text": "7.4% (32 of 432)",
                "markdown": "7.4% (32 of 432)",
                "header": false,
                "colSpan": 1,
                "rowSpan": 1
              },
              {
                "text": "77.9%",
                "markdown": "77.9%",
                "header": false,
                "colSpan": 1,
                "rowSpan": 1
              }
            ]
          },
          {
            "cells": [
              {
                "text": "Appointment access",
                "markdown": "Appointment access",
                "header": true,
                "colSpan": 1,
                "rowSpan": 1
              },
              {
                "text": "87.0% (120 of 138)",
                "markdown": "87.0% (120 of 138)",
                "header": false,
                "colSpan": 1,
                "rowSpan": 1
              },
              {
                "text": "16.7% (24 of 144)",
                "markdown": "16.7% (24 of 144)",
                "header": false,
                "colSpan": 1,
                "rowSpan": 1
              },
              {
                "text": "61.2%",
                "markdown": "61.2%",
                "header": false,
                "colSpan": 1,
                "rowSpan": 1
              }
            ]
          },
          {
            "cells": [
              {
                "text": "Cost and insurance",
                "markdown": "Cost and insurance",
                "header": true,
                "colSpan": 1,
                "rowSpan": 1
              },
              {
                "text": "72.4% (92 of 127)",
                "markdown": "72.4% (92 of 127)",
                "header": false,
                "colSpan": 1,
                "rowSpan": 1
              },
              {
                "text": "36.1% (52 of 144)",
                "markdown": "36.1% (52 of 144)",
                "header": false,
                "colSpan": 1,
                "rowSpan": 1
              },
              {
                "text": "50.4%",
                "markdown": "50.4%",
                "header": false,
                "colSpan": 1,
                "rowSpan": 1
              }
            ]
          }
        ]
      },
      {
        "caption": "Table 2. Final cost-and-insurance provider positions absent from the first action",
        "rows": [
          {
            "cells": [
              {
                "text": "Service line",
                "markdown": "Service line",
                "header": true,
                "colSpan": 1,
                "rowSpan": 1
              },
              {
                "text": "Final providers absent from the first action",
                "markdown": "Final providers absent from the first action",
                "header": true,
                "colSpan": 1,
                "rowSpan": 1
              }
            ]
          },
          {
            "cells": [
              {
                "text": "Bariatric surgery",
                "markdown": "Bariatric surgery",
                "header": true,
                "colSpan": 1,
                "rowSpan": 1
              },
              {
                "text": "50.0% (12 of 24)",
                "markdown": "50.0% (12 of 24)",
                "header": false,
                "colSpan": 1,
                "rowSpan": 1
              }
            ]
          },
          {
            "cells": [
              {
                "text": "Orthopedics",
                "markdown": "Orthopedics",
                "header": true,
                "colSpan": 1,
                "rowSpan": 1
              },
              {
                "text": "41.7% (10 of 24)",
                "markdown": "41.7% (10 of 24)",
                "header": false,
                "colSpan": 1,
                "rowSpan": 1
              }
            ]
          },
          {
            "cells": [
              {
                "text": "Fertility",
                "markdown": "Fertility",
                "header": true,
                "colSpan": 1,
                "rowSpan": 1
              },
              {
                "text": "37.5% (9 of 24)",
                "markdown": "37.5% (9 of 24)",
                "header": false,
                "colSpan": 1,
                "rowSpan": 1
              }
            ]
          },
          {
            "cells": [
              {
                "text": "Ophthalmology",
                "markdown": "Ophthalmology",
                "header": true,
                "colSpan": 1,
                "rowSpan": 1
              },
              {
                "text": "37.5% (9 of 24)",
                "markdown": "37.5% (9 of 24)",
                "header": false,
                "colSpan": 1,
                "rowSpan": 1
              }
            ]
          },
          {
            "cells": [
              {
                "text": "Primary care",
                "markdown": "Primary care",
                "header": true,
                "colSpan": 1,
                "rowSpan": 1
              },
              {
                "text": "25.0% (6 of 24)",
                "markdown": "25.0% (6 of 24)",
                "header": false,
                "colSpan": 1,
                "rowSpan": 1
              }
            ]
          },
          {
            "cells": [
              {
                "text": "Urgent care",
                "markdown": "Urgent care",
                "header": true,
                "colSpan": 1,
                "rowSpan": 1
              },
              {
                "text": "25.0% (6 of 24)",
                "markdown": "25.0% (6 of 24)",
                "header": false,
                "colSpan": 1,
                "rowSpan": 1
              }
            ]
          }
        ]
      }
    ],
    "links": [
      {
        "text": "Read the findings",
        "url": "https://rotgar.com/medical/resources/chatgpt-medical-provider-shortlists#answer",
        "kind": "internal",
        "download": false
      },
      {
        "text": "Review the methodology",
        "url": "https://rotgar.com/medical/resources/chatgpt-medical-provider-shortlists#methodology",
        "kind": "internal",
        "download": false
      },
      {
        "text": "Suganthan Mohanadasan’s analysis of provider and product names in ChatGPT’s first search queries",
        "url": "https://suganthan.com/blog/chatgpt-decides-before-it-searches/",
        "kind": "external",
        "download": false
      },
      {
        "text": "OpenAI describes web search",
        "url": "https://developers.openai.com/api/docs/guides/tools-web-search",
        "kind": "external",
        "download": false
      },
      {
        "text": "CC BY 4.0",
        "url": "https://creativecommons.org/licenses/by/4.0/",
        "kind": "external",
        "download": false
      },
      {
        "text": "Data Primary CSV Overall and scenario summary metrics in a compact tabular file. text/csv · 1.6 KB",
        "url": "https://rotgar.com/research-data/chatgpt-medical-provider-shortlists-v1.0/chatgpt-medical-provider-shortlists-v1.0.csv",
        "kind": "internal",
        "download": true
      },
      {
        "text": "Data Complete public JSON Machine-readable public tables, metadata, metrics and interpretation boundaries. application/json · 331 KB",
        "url": "https://rotgar.com/research-data/chatgpt-medical-provider-shortlists-v1.0/chatgpt-medical-provider-shortlists-v1.0.json",
        "kind": "internal",
        "download": true
      },
      {
        "text": "Data Research workbook Formatted workbook with ten documented research sheets. application/vnd.openxmlformats-officedocument.spreadsheetml.sheet · 62 KB",
        "url": "https://rotgar.com/research-data/chatgpt-medical-provider-shortlists-v1.0/chatgpt-medical-provider-shortlists-v1.0.xlsx",
        "kind": "internal",
        "download": true
      },
      {
        "text": "Reproducibility Prompt manifest The 72 frozen, brand-free commercial prompt cells used in the study. text/csv · 46 KB",
        "url": "https://rotgar.com/research-data/chatgpt-medical-provider-shortlists-v1.0/prompt-manifest.csv",
        "kind": "internal",
        "download": true
      },
      {
        "text": "Reproducibility Stability pairs The 144 repeat-pair rows used to reproduce provider-set stability without exposing provider names. text/csv · 4.3 KB",
        "url": "https://rotgar.com/research-data/chatgpt-medical-provider-shortlists-v1.0/stability-pairs.csv",
        "kind": "internal",
        "download": true
      },
      {
        "text": "Reproducibility Data dictionary Definitions for every field in the public package. text/csv · 7.7 KB",
        "url": "https://rotgar.com/research-data/chatgpt-medical-provider-shortlists-v1.0/data-dictionary.csv",
        "kind": "internal",
        "download": true
      },
      {
        "text": "Documentation Package README Scope, direct result, package contents and public-data boundary. text/markdown · 2.6 KB",
        "url": "https://rotgar.com/research-data/chatgpt-medical-provider-shortlists-v1.0/README.md",
        "kind": "internal",
        "download": true
      },
      {
        "text": "Documentation Methodology Research design, observable stages, metrics, entity handling and interpretation limits. text/markdown · 3.5 KB",
        "url": "https://rotgar.com/research-data/chatgpt-medical-provider-shortlists-v1.0/METHODOLOGY.md",
        "kind": "internal",
        "download": true
      },
      {
        "text": "Documentation Dataset metadata Machine-readable identity, scope, configuration, counts and license metadata. application/json · 4.2 KB",
        "url": "https://rotgar.com/research-data/chatgpt-medical-provider-shortlists-v1.0/dataset-metadata.json",
        "kind": "internal",
        "download": true
      },
      {
        "text": "Verification Package manifest Versioned file inventory with bytes, media types, row counts and SHA-256 values. text/plain · 5.4 KB",
        "url": "https://rotgar.com/research-data/chatgpt-medical-provider-shortlists-v1.0/manifest.json",
        "kind": "internal",
        "download": true
      },
      {
        "text": "Verification SHA-256 checksums Integrity checksums for the complete public package. text/plain · 2.2 KB",
        "url": "https://rotgar.com/research-data/chatgpt-medical-provider-shortlists-v1.0/SHA256SUMS.txt",
        "kind": "internal",
        "download": true
      },
      {
        "text": "Citation BibTeX citation Ready-to-import BibTeX citation metadata. text/plain · 600 B",
        "url": "https://rotgar.com/research-data/chatgpt-medical-provider-shortlists-v1.0/citation.bib",
        "kind": "internal",
        "download": true
      },
      {
        "text": "Citation Citation File Format CFF 1.2 metadata for research and repository tools. text/plain · 883 B",
        "url": "https://rotgar.com/research-data/chatgpt-medical-provider-shortlists-v1.0/CITATION.cff",
        "kind": "internal",
        "download": true
      },
      {
        "text": "Manifest",
        "url": "https://rotgar.com/research-data/chatgpt-medical-provider-shortlists-v1.0/manifest.json",
        "kind": "internal",
        "download": true
      },
      {
        "text": "SHA-256 checksums",
        "url": "https://rotgar.com/research-data/chatgpt-medical-provider-shortlists-v1.0/SHA256SUMS.txt",
        "kind": "internal",
        "download": true
      },
      {
        "text": "10.5281/zenodo.22003037",
        "url": "https://doi.org/10.5281/zenodo.22003037",
        "kind": "external",
        "download": false
      },
      {
        "text": "Zenodo record 22003037",
        "url": "https://zenodo.org/records/22003037",
        "kind": "external",
        "download": false
      },
      {
        "text": "Book a 30-minute call",
        "url": "https://calendly.com/z-rotgar/30min",
        "kind": "external",
        "download": false
      }
    ]
  },
  "structured_data": [
    {
      "@context": "https://schema.org",
      "@graph": [
        {
          "@type": "Organization",
          "@id": "https://rotgar.com/#organization",
          "name": "Rotgar",
          "url": "https://rotgar.com/",
          "foundingDate": "2011",
          "email": "evgeniy.yudin@rotgar.com",
          "telephone": "+34 605 303 439",
          "logo": {
            "@type": "ImageObject",
            "@id": "https://rotgar.com/#logo",
            "url": "https://rotgar.com/logo-square.png",
            "contentUrl": "https://rotgar.com/logo-square.png",
            "width": 112,
            "height": 112
          },
          "sameAs": [
            "https://clutch.co/profile/rotgar",
            "https://www.linkedin.com/company/rotgarseo-pro/"
          ]
        },
        {
          "@type": "BreadcrumbList",
          "@id": "https://rotgar.com/medical/resources/chatgpt-medical-provider-shortlists#breadcrumb",
          "itemListElement": [
            {
              "@type": "ListItem",
              "position": 1,
              "name": "Home",
              "item": "https://rotgar.com/"
            },
            {
              "@type": "ListItem",
              "position": 2,
              "name": "Resources",
              "item": "https://rotgar.com/resources"
            },
            {
              "@type": "ListItem",
              "position": 3,
              "name": "ChatGPT Medical Provider Shortlists",
              "item": "https://rotgar.com/medical/resources/chatgpt-medical-provider-shortlists"
            }
          ]
        },
        {
          "@type": "Organization",
          "@id": "https://rotgar.com/#rotgar-research",
          "name": "Rotgar Research",
          "parentOrganization": {
            "@id": "https://rotgar.com/#organization"
          }
        },
        {
          "@type": "DataCatalog",
          "@id": "https://zenodo.org/#datacatalog",
          "name": "Zenodo",
          "url": "https://zenodo.org/"
        },
        {
          "@type": "Person",
          "@id": "https://rotgar.com/about#evgeniy-yudin",
          "name": "Evgeniy Yudin",
          "url": "https://rotgar.com/about#evgeniy-yudin",
          "jobTitle": "Founder and Strategy Lead",
          "worksFor": {
            "@id": "https://rotgar.com/#organization"
          }
        },
        {
          "@type": "Person",
          "@id": "https://rotgar.com/medical/resources/chatgpt-medical-provider-shortlists#boris-teplyakov",
          "name": "Boris Teplyakov",
          "jobTitle": "SEO Lead",
          "description": "Boris Teplyakov reviewed the research methodology as SEO Lead. This was not a clinical subject-matter review or an assessment of the providers named in the responses.",
          "worksFor": {
            "@id": "https://rotgar.com/#organization"
          }
        },
        {
          "@type": "ImageObject",
          "@id": "https://rotgar.com/medical/resources/chatgpt-medical-provider-shortlists#primary-image",
          "url": "https://rotgar.com/images/research/chatgpt-medical-provider-shortlists/research-card-1200x630.png",
          "contentUrl": "https://rotgar.com/images/research/chatgpt-medical-provider-shortlists/research-card-1200x630.png",
          "caption": "Research card showing that 85.0% of 720 final medical-provider positions were occupied by organizations already named in the query strings of ChatGPT's first returned web-search action across 180 responses.",
          "width": 1200,
          "height": 630,
          "encodingFormat": "image/png"
        },
        {
          "@type": "WebPage",
          "@id": "https://rotgar.com/medical/resources/chatgpt-medical-provider-shortlists#webpage",
          "url": "https://rotgar.com/medical/resources/chatgpt-medical-provider-shortlists",
          "name": "ChatGPT Named Most Final-Shortlist Medical Providers in Its First Search Queries",
          "description": "Across 180 commercial medical-provider responses, 85% of final provider positions were already named in ChatGPT's first returned search-action queries.",
          "inLanguage": "en-US",
          "datePublished": "2026-08-18",
          "dateModified": "2026-08-19",
          "isPartOf": {
            "@id": "https://rotgar.com/#website"
          },
          "author": {
            "@id": "https://rotgar.com/about#evgeniy-yudin"
          },
          "reviewedBy": {
            "@id": "https://rotgar.com/medical/resources/chatgpt-medical-provider-shortlists#boris-teplyakov"
          },
          "publisher": {
            "@id": "https://rotgar.com/#organization"
          },
          "primaryImageOfPage": {
            "@id": "https://rotgar.com/medical/resources/chatgpt-medical-provider-shortlists#primary-image"
          },
          "mainEntity": {
            "@id": "https://rotgar.com/medical/resources/chatgpt-medical-provider-shortlists#report"
          },
          "breadcrumb": {
            "@id": "https://rotgar.com/medical/resources/chatgpt-medical-provider-shortlists#breadcrumb"
          },
          "encoding": [
            {
              "@type": "MediaObject",
              "contentUrl": "https://rotgar.com/machine/pages/medical/resources/chatgpt-medical-provider-shortlists.md",
              "encodingFormat": "text/markdown"
            },
            {
              "@type": "MediaObject",
              "contentUrl": "https://rotgar.com/machine/pages/medical/resources/chatgpt-medical-provider-shortlists.json",
              "encodingFormat": "application/json"
            }
          ]
        },
        {
          "@type": [
            "Article",
            "Report"
          ],
          "@id": "https://rotgar.com/medical/resources/chatgpt-medical-provider-shortlists#report",
          "url": "https://rotgar.com/medical/resources/chatgpt-medical-provider-shortlists",
          "headline": "ChatGPT Named Most Final-Shortlist Medical Providers in Its First Search Queries",
          "description": "Across 180 commercial medical-provider responses, 85% of final provider positions were already named in ChatGPT's first returned search-action queries.",
          "inLanguage": "en-US",
          "datePublished": "2026-08-18",
          "dateModified": "2026-08-19",
          "author": {
            "@id": "https://rotgar.com/about#evgeniy-yudin"
          },
          "contributor": [
            {
              "@id": "https://rotgar.com/#rotgar-research"
            },
            {
              "@id": "https://rotgar.com/medical/resources/chatgpt-medical-provider-shortlists#boris-teplyakov"
            }
          ],
          "image": {
            "@id": "https://rotgar.com/medical/resources/chatgpt-medical-provider-shortlists#primary-image"
          },
          "version": "1.0",
          "license": "https://creativecommons.org/licenses/by/4.0/",
          "hasPart": {
            "@id": "https://rotgar.com/medical/resources/chatgpt-medical-provider-shortlists#dataset"
          },
          "publisher": {
            "@id": "https://rotgar.com/#organization"
          },
          "mainEntityOfPage": {
            "@id": "https://rotgar.com/medical/resources/chatgpt-medical-provider-shortlists#webpage"
          },
          "identifier": "sha256:11b12c97fa937e384f5006f560b50d4441bca5b907a8bd20bb84c774fdad656c",
          "citation": "Yudin, E. (2026). ChatGPT Named Most Final-Shortlist Medical Providers in Its First Search Queries (Version 1.0) [Data set]. Rotgar Research. https://doi.org/10.5281/zenodo.22003037",
          "about": [
            {
              "@type": "Thing",
              "name": "ChatGPT web search"
            },
            {
              "@type": "Thing",
              "name": "Medical provider visibility"
            },
            {
              "@type": "Thing",
              "name": "AI search measurement"
            }
          ]
        },
        {
          "@type": "Dataset",
          "@id": "https://rotgar.com/medical/resources/chatgpt-medical-provider-shortlists#dataset",
          "name": "ChatGPT Named Most Final-Shortlist Medical Providers in Its First Search Queries",
          "description": "A descriptive US benchmark of medical-provider names in the first returned web-search action and the final four-provider answer from one chat-latest Responses API configuration.",
          "url": "https://rotgar.com/medical/resources/chatgpt-medical-provider-shortlists",
          "mainEntityOfPage": {
            "@id": "https://rotgar.com/medical/resources/chatgpt-medical-provider-shortlists#webpage"
          },
          "creator": {
            "@id": "https://rotgar.com/about#evgeniy-yudin"
          },
          "contributor": {
            "@id": "https://rotgar.com/medical/resources/chatgpt-medical-provider-shortlists#boris-teplyakov"
          },
          "publisher": {
            "@id": "https://rotgar.com/#rotgar-research"
          },
          "isPartOf": {
            "@id": "https://rotgar.com/medical/resources/chatgpt-medical-provider-shortlists#report"
          },
          "identifier": [
            "chatgpt-medical-provider-shortlists-v1.0",
            {
              "@type": "PropertyValue",
              "propertyID": "DOI",
              "value": "10.5281/zenodo.22003037",
              "url": "https://doi.org/10.5281/zenodo.22003037"
            }
          ],
          "sameAs": [
            "https://doi.org/10.5281/zenodo.22003037",
            "https://zenodo.org/records/22003037"
          ],
          "includedInDataCatalog": {
            "@id": "https://zenodo.org/#datacatalog"
          },
          "version": "1.0",
          "license": "https://creativecommons.org/licenses/by/4.0/",
          "isAccessibleForFree": true,
          "inLanguage": "en-US",
          "keywords": [
            "ChatGPT",
            "healthcare marketing",
            "medical providers",
            "brand visibility",
            "web search",
            "United States"
          ],
          "datePublished": "2026-08-18",
          "dateModified": "2026-08-18",
          "temporalCoverage": "2026-08-18T13:35:58Z/2026-08-18T14:10:01Z",
          "spatialCoverage": [
            {
              "@type": "Place",
              "name": "New York, New York"
            },
            {
              "@type": "Place",
              "name": "Chicago, Illinois"
            },
            {
              "@type": "Place",
              "name": "Houston, Texas"
            }
          ],
          "measurementTechnique": "Comparison of normalized medical-provider organizations in the query strings of the first returned web-search action and the final four-provider answer, with repeated exact prompt cells and prompt-cell-clustered bootstrap intervals.",
          "variableMeasured": [
            {
              "@type": "PropertyValue",
              "name": "Final provider positions already named in first-action query strings"
            },
            {
              "@type": "PropertyValue",
              "name": "First-action-to-final provider retention"
            },
            {
              "@type": "PropertyValue",
              "name": "Final provider positions absent from first-action query strings"
            },
            {
              "@type": "PropertyValue",
              "name": "Within-cell provider-set Jaccard stability"
            }
          ],
          "distribution": [
            {
              "@type": "DataDownload",
              "name": "Primary CSV",
              "description": "Overall and scenario summary metrics in a compact tabular file.",
              "contentUrl": "https://rotgar.com/research-data/chatgpt-medical-provider-shortlists-v1.0/chatgpt-medical-provider-shortlists-v1.0.csv",
              "encodingFormat": "text/csv",
              "contentSize": "1678 bytes",
              "isAccessibleForFree": true,
              "sha256": "b6299b98ac3ad30c4b93be6f91a3e4de2c1c9fd931719b70b7bc7ffa86b5440e"
            },
            {
              "@type": "DataDownload",
              "name": "Complete public JSON",
              "description": "Machine-readable public tables, metadata, metrics and interpretation boundaries.",
              "contentUrl": "https://rotgar.com/research-data/chatgpt-medical-provider-shortlists-v1.0/chatgpt-medical-provider-shortlists-v1.0.json",
              "encodingFormat": "application/json",
              "contentSize": "339287 bytes",
              "isAccessibleForFree": true,
              "sha256": "7c4371a2cc50fbb3b48236ff13beeea845a0f4ff19f8155ba5ddc1b6e5372abd"
            },
            {
              "@type": "DataDownload",
              "name": "Research workbook",
              "description": "Formatted workbook with ten documented research sheets.",
              "contentUrl": "https://rotgar.com/research-data/chatgpt-medical-provider-shortlists-v1.0/chatgpt-medical-provider-shortlists-v1.0.xlsx",
              "encodingFormat": "application/vnd.openxmlformats-officedocument.spreadsheetml.sheet",
              "contentSize": "63464 bytes",
              "isAccessibleForFree": true,
              "sha256": "6c7537354f934a91245c587c23529cd386c90ffa5bb3cf020b3e71adde528f39"
            },
            {
              "@type": "DataDownload",
              "name": "Prompt manifest",
              "description": "The 72 frozen, brand-free commercial prompt cells used in the study.",
              "contentUrl": "https://rotgar.com/research-data/chatgpt-medical-provider-shortlists-v1.0/prompt-manifest.csv",
              "encodingFormat": "text/csv",
              "contentSize": "47441 bytes",
              "isAccessibleForFree": true,
              "sha256": "cff507c483119a4d39e128080d23c8f71aef21a43d35ba438626e3a5526a6244"
            },
            {
              "@type": "DataDownload",
              "name": "Stability pairs",
              "description": "The 144 repeat-pair rows used to reproduce provider-set stability without exposing provider names.",
              "contentUrl": "https://rotgar.com/research-data/chatgpt-medical-provider-shortlists-v1.0/stability-pairs.csv",
              "encodingFormat": "text/csv",
              "contentSize": "4400 bytes",
              "isAccessibleForFree": true,
              "sha256": "4f7ceb813b153a24a44377515658f9f278bc79df0298b18bafa93898b78f3a8f"
            },
            {
              "@type": "DataDownload",
              "name": "Data dictionary",
              "description": "Definitions for every field in the public package.",
              "contentUrl": "https://rotgar.com/research-data/chatgpt-medical-provider-shortlists-v1.0/data-dictionary.csv",
              "encodingFormat": "text/csv",
              "contentSize": "7882 bytes",
              "isAccessibleForFree": true,
              "sha256": "f5733f2d60e2d23b5d243875e562f15be4f6736620295a26b9b8a8c7798c3864"
            }
          ]
        },
        {
          "@type": "WebSite",
          "@id": "https://rotgar.com/#website",
          "name": "Rotgar — AI Visibility for Clinics",
          "url": "https://rotgar.com/",
          "inLanguage": "en",
          "publisher": {
            "@id": "https://rotgar.com/#organization"
          }
        }
      ]
    }
  ]
}