Google Search AI Overview

Request it with `include.aioverview`. The field carries the answer text, markdown, sources and citation pills when Google shows an overview, and `null` when it doesn't, which is a valid observation for presence tracking.

Returned in result

Request

curl -X POST "https://api.answerline.dev/v1/monitor/google" \
  -H "Authorization: Bearer $API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"query":"project management software","country":"US","include":{"aioverview":{"markdown":true}}}'

What the AI Overview is

The AI Overview is the generated answer Google places at the top of some results pages, with citation chips, a rail of sources, and sometimes videos and ads.

result.aioverview returns all of it as fields. Because the overview answers the query above the organic results, it is tracked for presence (which keywords trigger one, per market), citation share (which domains Google cites, yours included), attribution of sources to individual chips through citationPills, and the ads and videos placed inside it.

Field reference

result.aioverview

Object, or null when requested but not shown. Absent when not requested.

sources[]

citationPills[] and relatedLinks[]

videos[]

ads[]

Annotated excerpt

Values are illustrative; names and types follow the POST /v1/monitor/google result schema. The request set include.aioverview: {"markdown": true}.

{
  "result": {
    "aioverview": {                          // null if Google showed no overview
      "text": "A heat pump can cut heating costs when...",
      "markdown": "A heat pump can **cut heating costs** when...",
      "sources": [
        { "position": 1, "url": "https://energy.example.gov/heat-pumps", "label": "Heat Pump Systems", "description": "..." },
        { "position": 2, "url": "https://www.example-hvac.com/blog/heat-pump-cost", "label": "Heat pump running costs", "description": "..." }
      ],
      "citationPills": [
        // One chip ("Heat Pump Systems +1") citing two sources: two entries, same ID.
        { "label": "Heat Pump Systems", "citationPillId": 1, "url": "https://energy.example.gov/heat-pumps", "domain": "energy.example.gov", "position": 1 },
        { "label": "Heat pump running costs", "citationPillId": 1, "url": "https://www.example-hvac.com/blog/heat-pump-cost", "domain": "example-hvac.com", "position": 2 }
      ],
      "relatedLinks": [
        // Under chip 1's flyout, not in sources, so no position.
        { "label": "Compare heat pumps", "citationPillId": 1, "url": "https://www.example-compare.com/heat-pumps", "domain": "example-compare.com" }
      ],
      "videos": [],
      "ads": [
        { "position": 1, "title": "Heat Pump Installation Quotes", "url": "https://www.example-installers.com/", "type": "TEXT", "domain": "example-installers.com", "description": "Compare local installers..." }
      ]
    }
  }
}

Cost

Add include.aioverview as an object. It adds +$0.00071 when Google shows an overview; include.paaAioverview is free:

The add-on is charged only when the response holds an overview: a request with aioverview: null is not charged the +$0.00071. GET /v1/countries?model=aioverview lists supported markets.

Example: chip structure, co-citation and organic overlap

The AI Overview monitoring post covers presence rate, citation share and rank gap for one domain. This script looks at the structure instead: which domains appear in the most chips, which domains are cited together in one chip, how many cited URLs also rank in the organic top 10, and how often overviews carry ads or videos.

import os
import time
from collections import Counter
from datetime import date
from itertools import combinations

import requests

API = "https://api.answerline.dev"
HEADERS = {"Authorization": f"Bearer {os.environ['API_KEY']}"}
KEYWORDS = ["are heat pumps worth it", "heat pump vs furnace", "heat pump running cost", "heat pump in cold climate"]
DAY = date.today().isoformat()


def run_tasks(tasks):
    resp = requests.post(f"{API}/v1/async/task/batch", headers=HEADERS, json=tasks, timeout=60)
    resp.raise_for_status()
    pending = {item["task"]["id"]: tasks[item["index"]] for item in resp.json()["results"] if item["success"]}
    done = []
    while pending:
        time.sleep(10)
        for task_id in list(pending):
            body = requests.get(f"{API}/v1/async/task/{task_id}", headers=HEADERS, timeout=30).json()
            if body["task"]["status"] in ("COMPLETED", "FAILED"):
                task = pending.pop(task_id)
                if body["task"]["status"] == "COMPLETED":
                    done.append((task["payload"]["query"], body["response"]["result"]))
    return done


tasks = [
    {
        "taskType": "GOOGLE",
        "payload": {"query": kw, "country": "US", "include": {"aioverview": {}}},
        "idempotencyKey": f"aio-structure-{i}-{DAY}",
    }
    for i, kw in enumerate(KEYWORDS)
]

chips_per_domain, pairs = Counter(), Counter()
overviews = cited_urls = cited_in_top10 = with_ads = with_videos = 0
for keyword, result in run_tasks(tasks):
    overview = result.get("aioverview")
    if overview is None:
        continue
    overviews += 1
    with_ads += bool(overview.get("ads"))
    with_videos += bool(overview.get("videos"))

    chips = {}
    for pill in overview.get("citationPills", []):
        chips.setdefault(pill["citationPillId"], set()).add(pill["domain"].removeprefix("www."))
    for domains in chips.values():
        chips_per_domain.update(domains)
        pairs.update(combinations(sorted(domains), 2))

    top10 = {r["link"] for r in result["organicResults"] if r["position"] <= 10}
    for source in overview.get("sources", []):
        cited_urls += 1
        cited_in_top10 += source["url"] in top10

print(f"overviews: {overviews}/{len(KEYWORDS)}  with ads: {with_ads}  with videos: {with_videos}")
print(f"cited URLs also ranking top 10 organically: {cited_in_top10}/{cited_urls}")
print("Domains by number of chips:", chips_per_domain.most_common(10))
print("Domains cited together in one chip:", pairs.most_common(10))

Exact URL matching understates overlap when URL variants differ; normalize URLs first if the number drives decisions.

Pitfalls

Three states need separate handling: key absent (not requested), null (requested, no overview), and an object without citationPills or relatedLinks (an overview without chips or flyouts).

The same query can show an overview on one run and not the next. Measure presence and citations as rates over several runs per keyword, with country fixed, and budget for the +$0.00071 on every overview Google shows; the ones that come back null cost nothing extra.

markdown can hold more URLs than sources, because related links appear inline in it. Count rail citations from sources and citationPills and flyout links from relatedLinks; don’t extract URLs from markdown.

Count distinct citationPillId values for chip counts. IDs start at 1 here and at 0 on ChatGPT.

Related

• Monitoring Google AI Overviews and AI Overviews with async batches and webhooks.
• AI Overview API options and the Google AI Overview API post.
• People Also Ask, where include.paaAioverview fills in AI-answered questions.
• AI Mode sources, Google’s conversational surface.
• Google Search engine guide for targeting, batches and billing.

More Google Search data

All Google Search fields · API reference