Google AI Overviews now appear for a significant share of informational queries, and their presence directly impacts organic click-through rates. SEO teams need to know which of their target keywords trigger AI Overviews, what the overview says, and which sources are cited. This tutorial builds a batch monitoring system that checks hundreds of keywords for AI Overview presence, extracts the summary text and cited sources, and outputs a report showing coverage trends over time.
Prerequisites
- Python 3.10 or higher
- requests library installed
- A Scavio API key
- A keyword list to monitor (CSV or Python list)
Walkthrough
Step 1: Load keywords to monitor
Read keywords from a CSV file or define them inline. Each keyword will be checked for AI Overview presence.
Python
import csv
def load_keywords(path: str) -> list[str]:
with open(path) as f:
reader = csv.reader(f)
return [row[0] for row in reader]
# Or define inline
KEYWORDS = ["what is rag", "how to fine tune llm", "python vs rust 2026"]Step 2: Check AI Overview for each keyword
Query each keyword with ai_overview enabled and extract the overview text and sources if present.
Python
def check_ai_overview(keyword: str) -> dict:
r = requests.post(
"https://api.scavio.dev/api/v1/search",
headers={"x-api-key": API_KEY},
json={"query": keyword, "country_code": "us", "ai_overview": True}
)
r.raise_for_status()
data = r.json()
ai = data.get("ai_overview")
return {
"keyword": keyword,
"has_overview": ai is not None,
"text": ai.get("text", "") if ai else "",
"sources": ai.get("sources", []) if ai else [],
}Step 3: Batch process all keywords
Run the check across all keywords with rate limiting. Track progress and handle errors gracefully.
Python
import time
def batch_check(keywords: list[str]) -> list[dict]:
results = []
for i, kw in enumerate(keywords):
try:
result = check_ai_overview(kw)
results.append(result)
except Exception as e:
results.append({"keyword": kw, "has_overview": None, "error": str(e)})
if i % 10 == 9:
time.sleep(1)
return resultsStep 4: Generate the coverage report
Summarize what percentage of keywords trigger AI Overviews and which domains are most frequently cited.
Python
from collections import Counter
def report(results: list[dict]) -> None:
total = len(results)
with_overview = sum(1 for r in results if r.get("has_overview"))
print(f"AI Overview coverage: {with_overview}/{total} ({with_overview/total*100:.0f}%)")
all_sources = []
for r in results:
for s in r.get("sources", []):
domain = s.get("link", "").split("/")[2] if s.get("link") else "unknown"
all_sources.append(domain)
print("\nTop cited domains:")
for domain, count in Counter(all_sources).most_common(10):
print(f" {domain}: {count} citations")Python Example
Python
import os
import time
import json
import requests
from collections import Counter
API_KEY = os.environ.get("SCAVIO_API_KEY", "your_scavio_api_key")
ENDPOINT = "https://api.scavio.dev/api/v1/search"
KEYWORDS = ["what is rag", "how to fine tune llm", "python vs rust 2026", "best vector database"]
def check(kw: str) -> dict:
r = requests.post(ENDPOINT, headers={"x-api-key": API_KEY},
json={"query": kw, "country_code": "us", "ai_overview": True})
r.raise_for_status()
ai = r.json().get("ai_overview")
return {"keyword": kw, "has_overview": ai is not None,
"text": (ai or {}).get("text", "")[:200],
"sources": (ai or {}).get("sources", [])}
if __name__ == "__main__":
results = []
for kw in KEYWORDS:
results.append(check(kw))
time.sleep(0.5)
with_ai = sum(1 for r in results if r["has_overview"])
print(f"Coverage: {with_ai}/{len(results)}")
for r in results:
status = "YES" if r["has_overview"] else "NO"
print(f" [{status}] {r['keyword']}")JavaScript Example
JavaScript
const API_KEY = process.env.SCAVIO_API_KEY || "your_scavio_api_key";
const ENDPOINT = "https://api.scavio.dev/api/v1/search";
async function check(kw) {
const res = await fetch(ENDPOINT, {
method: "POST",
headers: { "x-api-key": API_KEY, "Content-Type": "application/json" },
body: JSON.stringify({ query: kw, country_code: "us", ai_overview: true })
});
const data = await res.json();
return { keyword: kw, hasOverview: !!data.ai_overview, text: data.ai_overview?.text?.slice(0, 200) || "" };
}
async function main() {
const keywords = ["what is rag", "how to fine tune llm", "python vs rust 2026"];
const results = [];
for (const kw of keywords) {
results.push(await check(kw));
}
const count = results.filter(r => r.hasOverview).length;
console.log(`AI Overview coverage: ${count}/${results.length}`);
results.forEach(r => console.log(` [${r.hasOverview ? "YES" : "NO"}] ${r.keyword}`));
}
main().catch(console.error);Expected Output
JSON
AI Overview coverage: 3/4 (75%)
[YES] what is rag
[YES] how to fine tune llm
[YES] python vs rust 2026
[NO] best vector database
Top cited domains:
aws.amazon.com: 3 citations
docs.python.org: 2 citations
huggingface.co: 2 citations