Yandex Search Results as Structured JSON
Yandex Search remains the dominant search engine for Russian-speaking audiences, and its result pages carry information that Google-only rank tracking cannot see: Yandex-specific SERP features, ad placements, related searches and product carousel entries with price and store.
This page explains what a hosted Yandex Search API does, which fields it returns, how request options affect credit consumption, and when building your own headless-browser and proxy stack is the better choice.
What the API does with a Yandex Search URL
You send a Yandex Search results URL together with your API key. The service loads the page in a real browser, rotates the outbound IP, and returns the response your code can act on. Instead of managing proxy pools, CAPTCHA handling and headless Chrome yourself, you make one HTTP request per SERP.
The parsed output covers the fields most SEO and market-research workflows need: result position, destination URL, page title, snippet text, SERP features, ad placements, related searches, and — where Yandex shows a product carousel — product title, price, store and image.
A typical setup sequence looks like this:
- Create an account and obtain an API key.
- Install an SDK for your stack (Python, Node.js, Go, PHP, Ruby or Java) or call the endpoint with cURL.
- Send the Yandex Search URL with your key.
- Add options only when the page needs them: JavaScript rendering, stealth proxies, geo-targeting or extraction rules.
Field map: what comes back
Use this map to decide which fields your pipeline should persist. It reflects the parsed SERP output described for this integration; exact JSON key names should be confirmed against the current API documentation before you write parsers.
Where teams use it
Three workflows come up most often:
- Russian-market rank tracking. Pull Yandex's top organic results for each monitored keyword. A Google-only rank tracker will not reflect this audience, so Yandex positions need their own collection.
- Competitor share of voice. Because positions, URLs and SERP features arrive together, you can compute how often a domain appears and in which result types.
- Product discovery research. Yandex product carousel entries include title, price, store and image, which supports price-comparison and assortment analysis for the Russian market.
Cross-engine comparisons are also possible: the same provider offers a Google Search API, so you can collect both engines' SERPs and compare visibility side by side.
Request options and what they cost
Options are additive: you only pay for what a request enables. The credit rules below are the captured contract values and should be treated as the authoritative reference for cost planning.
| Option | Effect | Credit impact |
|---|---|---|
| JavaScript rendering | Executes page scripts in a real browser | 5 credits per successful request |
| Premium proxies | Routes through premium proxy pool | 10 credits (JS off) or 25 credits (JS on) per successful request |
| Stealth proxy | Higher-tolerance proxy for CAPTCHA-heavy targets | 75 credits per request |
| Geo-targeting | Routes through 200+ countries and territories (40+ on classic proxies) | Included in plan features |
| Screenshots & extraction rules | Full-page screenshots; CSS or XPath structured extraction | Included in plan features |
A worked example: a monthly rank-tracking job of 2,000 keywords with JavaScript rendering enabled consumes 2,000 × 5 = 10,000 credits. If you also need stealth proxies on every request, the same job consumes 2,000 × 75 = 150,000 credits — a 15× difference that should drive your option choices.
Concurrency limits vs. target throttling
These are two different constraints and are easy to confuse:
- Concurrency limits are set per plan. They cap how many requests you can have in flight at once. The plan table below lists the exact number for each tier.
- Target throttling happens when Yandex itself slows or blocks aggressive request patterns. The API does not impose a per-target rate limit, so if you see throttling, space out requests or spread them across sessions.
If your job is failing with timeouts rather than blocks, check concurrency first. If it is failing with challenge pages, check whether stealth proxies are needed for that target.
Getting structured JSON
Two extraction paths produce parsed JSON:
- Selector-based: pass
extract_ruleswith CSS selectors mapped to field names. This is deterministic and cheap, and works well once the page structure is stable. - Natural-language: pass
ai_extract_ruleswith a plain-language description of the fields you want. No selectors required; useful when the page layout shifts or when you are exploring a new result type.
Both return parsed JSON, so downstream code does not need an HTML parser. For LLM pipelines, the same provider also offers a Markdown output mode that converts pages into text suitable for retrieval-augmented generation.
When to build your own stack instead
A hosted API is not the only option. Consider building in-house if:
- Your volume is very low and occasional manual collection is acceptable.
- You need a rendering or proxy behavior the API does not expose.
- You already operate a browser fleet and proxy pool at scale.
Conversely, the hosted route tends to win when you need Yandex SERPs on a schedule, across many keywords, without dedicating engineering time to proxy rotation, CAPTCHA handling and browser maintenance. The decision usually comes down to engineering hours versus per-request cost.
Legal and compliance notes
Public-facing pages can usually be collected, but you should review the target site's terms of service before deploying and avoid post-login content. The provider states that it does not collect or store personal data from scraped sites unless the user explicitly requests it, and describes itself as GDPR and CCPA compliant. Those statements are the provider's own; verify them against your own legal requirements before relying on them.
| Field | What it contains | Typical use |
|---|---|---|
| Position | Rank of the result in the organic list | Rank tracking, share of voice |
| URL | Destination link of the result | Domain-level aggregation |
| Title | Result headline | Snippet quality checks |
| Snippet | Descriptive text under the title | Content-gap analysis |
| SERP features | Special result blocks shown on the page | Feature visibility tracking |
| Ad placements | Paid result positions | Paid vs. organic comparison |
| Related searches | Query suggestions shown on the SERP | Keyword discovery |
| Product carousel: title | Product name in the carousel | Assortment research |
| Product carousel: price | Listed price | Price comparison |
| Product carousel: store | Merchant name | Retailer coverage |
| Product carousel: image | Product image reference | Catalog enrichment |
Field availability depends on the SERP layout Yandex serves for a given query and locale. Confirm exact JSON key names against current API documentation.
Frequently asked questions
Do I need a Yandex account to use the API?
No. You need an account with the API provider to obtain an API key. You then send the Yandex Search URL and the service handles the rest. The provider's signup offer and free-credit terms are set by the owner and should be confirmed on the current pricing page.
Is scraping Yandex Search legal?
Public-facing pages can usually be collected, but you should review Yandex's terms of service before deploying and avoid post-login content. Legal requirements vary by jurisdiction; this is not legal advice.
What rate limits apply?
The API enforces concurrency limits per plan, not per-target rate limits. If Yandex throttles aggressive request patterns, space out requests or use multiple sessions. The plan table lists the exact concurrency figure for each tier.
How do I get structured JSON from Yandex Search?
Pass extract_rules with CSS selectors mapped to field names, or use ai_extract_rules with a natural-language specification. The response comes back as parsed JSON.
Which programming languages are supported?
SDKs are published for Python, Node.js, Go, PHP, Ruby and Java. Plain cURL works too, so you can use whatever your stack already relies on.
How much does a request cost?
Credit cost per request depends on the options you enable: 5 credits for JavaScript rendering, 10 credits for premium proxies with JavaScript rendering disabled, 25 credits for premium proxies with JavaScript rendering enabled, and 75 for stealth proxies. Plan prices and any free-credit allowance are commercial terms set by the provider; confirm them on the current pricing page.