If you've ever needed to pull real Google search results into your app — for a rank tracker, a research tool, a content analysis pipeline, or even a side project — you've probably hit the same wall: SERP APIs are expensive.
This post walks through how to fetch structured Google search data via a REST API in JavaScript, shows you the full response structure, and shares some honest thoughts on the cost gap between different providers I've tested.
What is a SERP API?
A SERP API (Search Engine Results Page API) returns the structured content of a search engine results page as JSON. Instead of loading google.com in a browser, you call an endpoint with a query and get back clean data: organic results, featured snippets, People Also Ask boxes, AI Overviews, ads, related searches — all parsed.
The use cases are wide: rank tracking, keyword research tools, content gap analysis, competitor monitoring, or feeding search context into an LLM pipeline.
The Basic Integration
I've been using Serpent API for this. Here's the simplest working implementation:
`javascript
// Basic SERP fetch — returns Google web results
const fetchSERP = async (query) => {
const response = await fetch(
`https://apiserpent.com/api/search?q=${encodeURIComponent(query)}`,
{ headers: { 'X-API-Key': 'sk_live_your_key' } }
);
const data = await response.json();
return data;
};
// Usage
const results = await fetchSERP('best javascript frameworks 2025');
console.log(results.results.organic); // array of organic results
`
That's the entire integration. One endpoint, one header, JSON response. No SDK required.
The Response Structure
Here's what a typical response object looks like (abbreviated):
json
**json — response structure
{
"success": true,
"query": "best javascript frameworks 2025",
"results": {
"organic": [
{
"position": 1,
"title": "Top JS Frameworks in 2025",
"url": "https://example.com/...",
"snippet": "React continues to dominate..."
}
// up to 100 results per query
],
"ai_overview": { /* Google AIO block if present */ },
"featured_snippet": { /* snippet box if present */ },
"people_also_ask": [ /* PAA questions + answers */ ],
"related_searches": [ /* related queries */ ],
"ads": [ /* top and bottom ads */ ]
},
"meta": {
"engine": "google",
"total_results": 100,
"country": "us"
}
}**
Switching Search Engines
The SERP API supports Google (default), Bing, Yahoo, and DuckDuckGo. You switch with a simple query parameter:
`
javascript
// Use Bing instead of Google
const bingResults = await fetch(
'https://apiserpent.com/api/search?q=seo+tools&engine=bing',
{ headers: { 'X-API-Key': 'sk_live_your_key' } }
);
// DuckDuckGo — same pattern
const ddgResults = await fetch(
'https://apiserpent.com/api/search?q=seo+tools&engine=duckduckgo',
{ headers: { 'X-API-Key': 'sk_live_your_key' } }
);
`
All four engines return the same normalized JSON structure. No extra parsing logic.
Bonus: Tracking AI Citations
This is the feature I didn't expect to need but now use constantly. The AI Ranking endpoint queries ChatGPT, Claude, Gemini, and Perplexity on your behalf and returns whether your brand/domain is cited in their responses.
`plaintext
`
javascript — ai ranking
// Check if your brand appears in AI responses
const aiVisibility = await fetch(
'https://apiserpent.com/api/ai-rank?q=best+serp+api&brand=apiserpent.com',
{ headers: { 'X-API-Key': 'sk_live_your_key' } }
);
`
`
// Returns: visibility_score, cited_by[], sources[]
This is increasingly relevant as users get answers directly from AI tools rather than clicking through to websites. It's an early signal for what's being called GEO — Generative Engine Optimization.
Full pricing breakdown: apiserpent.com/pricing
Getting Started
- Sign up at apiserpent.com (Google OAuth, 30 seconds)
- Get your API key from the dashboard
- You start with 10 free searches — no credit card
- Read the docs: apiserpent.com/docs
- The documentation is genuinely clean. I had a working prototype calling real Google data in under 30 minutes.
If you're building anything that touches search data — rank trackers, SEO tools, content pipelines, LLM context — this is worth 10 minutes of your time to test.
Happy to answer questions about integration patterns, response edge cases (PAA structure is a bit nested), or the AI ranking feature in the comments.