Query Fan-Out API: expand and rank the sub-queries behind AI answers

The Fan-Out API returns the sub-questions an AI engine derives from a prompt before it answers, with how often each one occurs across every stored response. Each fan-out comes with its share of the total and whether the brand appeared in the answers that used it, so a team can see which questions it already covers.

Updated 14 October 2026

Get an API key Try the endpoint

ChatGPT
Perplexity
Google AI Mode
Claude
Microsoft Copilot
Grok

Playground

One seed prompt expands into 8 fan-outs, and the brand covers 3 of them

The Fan-Out API groups the sub-questions behind every stored answer and ranks them by occurrence. For the sample prompt on ChatGPT, client approval workflows top the list with 284 occurrences, and the brand appears in answers to 3 of the 8 questions. Raise the minimum occurrence to trim the tail, or switch the model to see how each engine expands the same prompt.

POST /v1/fan-outs/top

seed prompt

model

Unique fan-outs: 8
Avg per response: 4.7
Brand covered: 3 / 8

curl -X POST https://api.apiforgeo.com/v1/fan-outs/top \
  -H "Authorization: Bearer $API_FOR_GEO_KEY" \
  -H "x-project-id: $PROJECT_ID" \
  -H "Content-Type: application/json" \
  -d '{
    "queryText": "best project management tool for creative agencies",
    "models": [
      "chatgpt"
    ]
  }'

import os, requests

API = "https://api.apiforgeo.com/v1"
H = {"Authorization": f"Bearer {os.environ['API_FOR_GEO_KEY']}",
     "x-project-id": os.environ["PROJECT_ID"]}

r = requests.post(f"{API}/fan-outs/top", headers=H, json={
  "queryText": "best project management tool for creative agencies",
  "models": [
    "chatgpt"
  ]
}, timeout=30)
print(r.json())

const API = "https://api.apiforgeo.com/v1";
const headers = {
  Authorization: `Bearer ${process.env.API_FOR_GEO_KEY}`,
  "x-project-id": process.env.PROJECT_ID,
  "Content-Type": "application/json",
};
const post = (path, body) =>
  fetch(API + path, { method: "POST", headers, body: JSON.stringify(body) }).then((r) => r.json());

console.log(await post("/fan-outs/top", {
  "queryText": "best project management tool for creative agencies",
  "models": [
    "chatgpt"
  ]
}));

200 OK · 84 ms · 8 rows

Fan-out question	Occurrences	Share	Brand present
which project management tools have client approval workflows	284	21%	yes
how much does project management software cost per seat	231	17.1%	yes
best pm tool for agencies with under 20 people	198	14.6%	no
does it integrate with design and proofing tools	176	13%	yes
project management software with built-in time tracking	154	11.4%	no
loopwork vs cadence for creative teams	141	10.4%	no
can clients comment without a paid seat	96	7.1%	no
which tools support retainer and utilisation reporting	74	5.5%	no

{
  "scope": {
    "queryText": "best project management tool for creative agencies",
    "models": [
      "chatgpt"
    ]
  },
  "overview": {
    "totalFanOuts": 1354,
    "uniqueFanOuts": 8,
    "avgPerResponse": 4.7
  },
  "results": [
    {
      "text": "which project management tools have client approval workflows",
      "occurrences": 284,
      "share": 21,
      "brandPresent": true,
      "queryIds": [
        "q_4f1a2b"
      ]
    },
    {
      "text": "how much does project management software cost per seat",
      "occurrences": 231,
      "share": 17.1,
      "brandPresent": true,
      "queryIds": [
        "q_4f1a2b"
      ]
    },
    {
      "text": "best pm tool for agencies with under 20 people",
      "occurrences": 198,
      "share": 14.6,
      "brandPresent": false,
      "queryIds": [
        "q_4f1a2b"
      ]
    },
    {
      "text": "does it integrate with design and proofing tools",
      "occurrences": 176,
      "share": 13,
      "brandPresent": true,
      "queryIds": [
        "q_4f1a2b"
      ]
    },
    {
      "text": "project management software with built-in time tracking",
      "occurrences": 154,
      "share": 11.4,
      "brandPresent": false,
      "queryIds": [
        "q_4f1a2b"
      ]
    },
    {
      "text": "loopwork vs cadence for creative teams",
      "occurrences": 141,
      "share": 10.4,
      "brandPresent": false,
      "queryIds": [
        "q_4f1a2b"
      ]
    },
    {
      "text": "can clients comment without a paid seat",
      "occurrences": 96,
      "share": 7.1,
      "brandPresent": false,
      "queryIds": [
        "q_4f1a2b"
      ]
    },
    {
      "text": "which tools support retainer and utilisation reporting",
      "occurrences": 74,
      "share": 5.5,
      "brandPresent": false,
      "queryIds": [
        "q_4f1a2b"
      ]
    }
  ]
}

Fields returned per fan-out by POST /v1/fan-outs/top
Field	Type	Description
`text`	string	The fan-out question, as the engine phrased it.
`occurrences`	int	How many in-scope responses generated this fan-out.
`share`	float	Occurrences ÷ total fan-outs in scope.
`brandPresent`	boolean	Whether the brand was mentioned in answers containing this fan-out.
`queryIds[]`	array	The tracked prompts that produced it, to join back to prompt tracking.

Totals returned for the scope by POST /v1/fan-outs/overview
Field	Type	Description
`totalFanOuts`	int	Total fan-outs in scope.
`uniqueFanOuts`	int	Distinct fan-out questions in scope.
`avgPerResponse`	float	Average fan-outs generated per AI response.

Per prompt

The per-prompt variant returns the 4 fan-outs behind one tracked prompt

The Fan-Out API also reads one tracked prompt at a time through POST /v1/fan-outs/by-query/{queryId}. The id in the path names the prompt, which suits a prompt detail page or a content brief built around a single question. The response carries the same fields as the full list, for that prompt only.

POST /v1/fan-outs/by-query/{queryId}

curl -X POST https://api.apiforgeo.com/v1/fan-outs/by-query/q_7d20e4b1 \
  -H "Authorization: Bearer $API_FOR_GEO_KEY" \
  -H "x-project-id: $PROJECT_ID" \
  -H "Content-Type: application/json" \
  -d '{
    "models": [
      "chatgpt"
    ]
  }'

import os, requests

API = "https://api.apiforgeo.com/v1"
H = {"Authorization": f"Bearer {os.environ['API_FOR_GEO_KEY']}",
     "x-project-id": os.environ["PROJECT_ID"]}

r = requests.post(f"{API}/fan-outs/by-query/q_7d20e4b1", headers=H, json={
  "models": [
    "chatgpt"
  ]
}, timeout=30)
print(r.json())

const API = "https://api.apiforgeo.com/v1";
const headers = {
  Authorization: `Bearer ${process.env.API_FOR_GEO_KEY}`,
  "x-project-id": process.env.PROJECT_ID,
  "Content-Type": "application/json",
};
const post = (path, body) =>
  fetch(API + path, { method: "POST", headers, body: JSON.stringify(body) }).then((r) => r.json());

console.log(await post("/fan-outs/by-query/q_7d20e4b1", {
  "models": [
    "chatgpt"
  ]
}));

200 OK

{
  "scope": {
    "queryId": "q_7d20e4b1",
    "models": [
      "chatgpt"
    ]
  },
  "results": [
    {
      "text": "which client portal tools allow comments without a seat",
      "occurrences": 96,
      "share": 31.1,
      "brandPresent": false
    },
    {
      "text": "how do client portals handle file approvals",
      "occurrences": 88,
      "share": 28.5,
      "brandPresent": true
    },
    {
      "text": "client portal project software for agencies",
      "occurrences": 71,
      "share": 23,
      "brandPresent": true
    },
    {
      "text": "can clients see project timelines in the portal",
      "occurrences": 54,
      "share": 17.5,
      "brandPresent": false
    }
  ]
}

Fields returned per fan-out for one prompt
Field	Type	Description
`text`	string	The fan-out question, as the engine phrased it.
`occurrences`	int	How many responses to this prompt generated it.
`share`	float	Occurrences ÷ all fan-outs of this prompt.
`brandPresent`	boolean	Whether the brand was mentioned in answers containing this fan-out.

Get started

Four setup steps put the first fan-outs in your project

The Fan-Out API reads questions captured during normal executions, so it needs no job of its own. Once the project exists, its key is generated in the app and its prompts are imported, the first execution fills the fan-out tables, and reading them is not billed.

Step 1

Create a project

Create an organisation and a project, one project per brand or per client.
Step 2

Generate a key

Generate an API key in the app; keys are managed there, never through the API.
Step 3

Import your prompts

Import your prompts and tags with a single POST /v1/queries.
Step 4

Run and read

Trigger an execution, poll the job, then read any analytics endpoint.

Comparison

Only 1 of 4 approaches returns fan-outs ranked and ready to use

The Fan-Out API returns the follow-up questions as ranked data, with occurrences and brand presence. An AI scraping API returns the final answer without them, a GEO dashboard tool rarely exposes them, and an in-house build has to capture and count them per engine, then keep up whenever an engine changes how it composes answers.

3 of the 15 capabilities on the full comparison
Capability	API for GEO	AI scraping API	GEO dashboard tool	In-house build
Fan-out follow-up questions	Yes	No	Rare	Build it
Raw answer text	Yes	Yes	Sometimes	Build it
Ongoing maintenance when engines change	Ours	Partly theirs	Theirs	Yours

See the full comparison on 15 capabilities

Use cases

Fan-outs feed 3 deliverables: briefs, prompt sets and gap reports

The Fan-Out API turns the questions engines ask themselves into work items. Agencies build content briefs from the top questions, platforms expand their customers' prompt sets with the ones that keep coming back, and brands look for questions where competitors are cited and they are not. Each card opens the matching pattern.

SEO & martech platforms

Prompt sets that grow themselves

Promote high-occurrence fan-outs into tracked prompts, so visibility is measured where the demand actually sits. The queryIds field links each question back to its source prompt.

Agencies

Content briefs in the engine's words

Turn the top 20 fan-outs of a topic into H2s and FAQ entries, phrased the way the engine already phrases them. Coverage of those questions is what gets a page retrieved.

E-commerce & brands

Gaps where competitors get cited

Cross fan-outs with the citation layer to find questions where a competitor is cited and the brand is not. Each gap becomes a page to write or a source to win.

Pricing

Query Fan-Out API pricing runs from $0.013 to $0.007 per credit

The Fan-Out API adds no charge of its own: fan-outs are captured while prompts run, and reading them is not billed. The credits are the executions themselves, one per prompt, per model, per locale, priced by the month's total volume with no commitment needed to reach a lower tier.

0 – 50k / month

$0.013

per credit

50k – 250k / month

$0.011

per credit

250k – 1M / month

$0.009

per credit

1M+ / month

$0.007

per credit

Price per credit, set by the month's total volume.

200 prompts × 3 models × 30 runs = 18,000 credits a month, $234 at $0.013

Get an API key Estimate my volume

FAQ

Fan-out data, answered

What does the ranked list return?

The Fan-Out API returns each sub-question as the engine phrased it, with its occurrences, its share of all fan-outs in scope, whether the brand appeared alongside it and the tracked prompts that produced it. The overview call adds the totals for the scope: total fan-outs, unique fan-outs and the average per response.

How stable are fan-outs between runs?

The Fan-Out API counts occurrences across every stored response because single expansions vary from one run to the next. The most frequent questions change slowly while the rare ones churn, so a monthly brief built on the twenty most frequent questions stays valid.

Do fan-outs cost extra credits?

The Fan-Out API bills nothing beyond the executions themselves. Fan-outs are captured as part of each enriched run, and every read of the ranked list, the overview or a single prompt's variant is unmetered, so exploring the data does not change the monthly bill.

Can I turn a fan-out into a tracked prompt?

The Fan-Out API returns the queryIds behind each fan-out, so a frequent question can join the prompt set with one POST /v1/queries. Tracking it from then on measures visibility on the question the engine actually asked itself, next to the prompt it came from.

Can I resell fan-out data?

The Fan-Out API allows resale and white-label use, like every endpoint family. The payload carries no branding and each client can sit in an isolated project, so an agency can ship fan-out briefs under its own name or a platform can show the ranked questions inside its product.

Related GEO APIs

API

Expand your prompts into the questions AI engines ask

Get an API key

Query Fan-Out API: expand and rank the sub-queries behind AI answers

One seed prompt expands into 8 fan-outs, and the brand covers 3 of them

The per-prompt variant returns the 4 fan-outs behind one tracked prompt

Four setup steps put the first fan-outs in your project

Create a project

Generate a key

Import your prompts

Run and read

Only 1 of 4 approaches returns fan-outs ranked and ready to use

Fan-outs feed 3 deliverables: briefs, prompt sets and gap reports

Prompt sets that grow themselves

Content briefs in the engine's words

Gaps where competitors get cited

Query Fan-Out API pricing runs from $0.013 to $0.007 per credit

Fan-out data, answered

What does the ranked list return?

How stable are fan-outs between runs?

Do fan-outs cost extra credits?

Can I turn a fan-out into a tracked prompt?

Can I resell fan-out data?

Related GEO APIs

Prompt Tracking API →

AI Visibility API →

AI Model Comparison API →

Expand your prompts into the questions AI engines ask