Serpstat API Guide 2026: Plans, Credits, Limits and a First Request

What the Serpstat API is, in one paragraph

The Serpstat API gives programmatic access to the same data as the web interface: keywords, domains, backlinks, audits and rank tracking. It runs as a JSON-RPC 2.0 interface, so you send a POST request with a method name and parameters and get JSON back ^[1]. Scripts, dashboards, spreadsheets and AI agents all use the same API, and so does the MCP server. It is the best-value part of Serpstat: 200,000 credits a month come with Team at $129, or $100 billed annually. The limits are real, though. Below Agency you get one request per second, and there is no sandbox. For dashboards and AI agents that is fine. For a data product, you need a separate license.

Which plans include it, and how many credits you get

API access starts on Team, and Individual has none ^[2]. Every Team-level plan comes with a monthly pool of API credits, and the rate limit only rises on Agency. The Team API x2 variant is the cheapest way to double the pool: it costs $40 more a month than Team, or $135 instead of $100 billed annually, and gets the same 400,000 credits as Team x2.

API credits renew every month, and unused credits do not carry over. An older support page said they do, but the help article is newer and specific to the API, so we treat it as current. Spend or lose. When the pool runs dry, calls stop until the next cycle or a top-up. You can buy extra credits for one module, the API included, without upgrading the plan, and top-ups do not expire while the subscription runs. The credits dashboard shows one row per module, and the API pool is separate from Site Analysis or Keyword Research, so a heavy month in the interface does not touch it.

How credits are spent

The basic rule is simple. In most methods, each row in the response costs one credit, so a request that returns 1,000 keywords costs 1,000 credits. The help article adds a catch: "Every request costs at least one credit, even if the response is empty." ^[1] A few methods cost more. The Summary report costs 5 credits a call and a single-page audit costs 10, and in site audits run through MCP each page costs one credit, or ten with JavaScript rendering. The exact cost of every method sits at the top of its page in the API documentation. Ask for fewer rows when you can. Filters save money.

The free reads, and the expensive call

Rank Tracker read methods do not use credits. Reading positions you already track is free, while the tracking itself spends Rank Tracker checks, so a daily dashboard of tracked rankings costs no API credits at all. That makes Rank Tracker the cheapest data to automate. The opposite end is the full pull. Each method returns up to 60,000 rows, and bigger pulls mean paging or contacting Serpstat. One maximal request uses 30% of Team's monthly 200,000 credits, while the same budget spread over a month buys roughly 6,600 rows a day. One careless call can do that. Paging through smaller requests is almost always the cheaper and safer design.

The limits that shape your design

Three limits decide how you build. The first is speed. Team and Team x2 allow one request per second, and Agency allows ten. The MCP help article calls one a second sufficient for most workflows, and for dashboards and agents it is. For bulk jobs it is slow. A script can make at most 3,600 calls an hour and about 86,400 a day, which sounds like plenty until a batch of keyword lookups for 50 clients needs several calls each and queues behind a scheduled audit export on the same token.

The second limit works in your favor. The web interface has a daily search allowance of 100 to 5,000 by plan, but that limit applies only to the web interface. The API is bounded only by monthly credits and the rate limit, which makes it the better route for heavy research on Team. The third limit is the missing sandbox. All requests run against live data and use credits, so every test call spends real money. Plan a development budget before you write the first line.

What the data is, and where it comes from

The API returns data from Serpstat's databases, not live search results. Keyword databases update on a rolling basis, and frequency varies by region. Each request sets the search engine database in the se parameter, for example g_nl for Google Netherlands, and the statistics page lists every code from g_us to bing_us. Data depth varies widely by country. Check your market's keyword count before you build around it. For live results at the moment of the call, Serpstat points to its SERP API, available on request, and to the newer real-time method covered below.

Getting a token and sending a first request

In your Serpstat profile settings, generate a personal API token. The same token unlocks the browser extension, the Google Sheets add-on, the Looker Studio connector and the MCP server. Keep it secret, and rotate it if it leaks. Each user account gets one token. Team members get their own and draw on credits the owner allocates, which lets an agency give each developer a capped budget. It also means one person's script cannot drain everyone's credits, provided allocations are set.

Requests go to the v4 endpoint at api.serpstat.com with your token in the URL, and the body names the method, an id and the parameters. Methods have names such as SerpstatDomainProcedure.getDomainKeywords, and the method and the se parameter in the example below come from Serpstat's changelog and help center. Asking for 10 rows, as this call does, costs about 10 credits. Start small. A minimal call has four parts:

Testing without burning the budget

Without a sandbox, testing is budgeting. The help center suggests adding a team member as a test account, giving it a small credit allocation and using its token for development, then switching to the main token only for production; that allocation comes out of the owner's pool. Ask for 10 rows, not 1,000. Read Rank Tracker data, which is free. Use Serpstat's Colab notebook to check response shapes before writing parsers. A careful first week rarely needs more than a few thousand credits. The Team trial also includes the API, and to evaluate the data before buying you can ask in chat for test API access or trial credits.

The methods, grouped by the job they do

Research data: domains, keywords, links and pages

Domain methods return a domain's keywords, competitors, top pages, visibility history and traffic. Since March 2026 a language-tagged version, getDomainKeywordsByLanguage, adds a language code to each row at one credit per result, and domain history now includes AI Overview fields. Keyword methods cover volume, difficulty, CPC, related keywords and top-100 results. The getKeywordFullTop method returns Google's top 100 for a keyword and sorts by 13 parameters; it replaced the deprecated getKeywordTop in May 2024 ^[3], so update old scripts. Backlink methods return backlinks, referring domains and anchors, and they return many rows. Filter by date or type before you pull a large profile. URL methods show what a page ranks for and the keywords it misses, for content-gap work.

Rank Tracker, audits and account automation

Rank Tracker methods return keyword and URL position history, free to read. Since April 23, 2026 the API covers all Rank Tracker entities: you can create projects, add keywords, set tags, regions and schedules, and trigger checks on demand. August 2026 added getProjectSummary, diffPositions, detectUrlCannibalization and getTagMetrics. Site Audit methods start, stop and configure audits and return issues by category, while audits spend Site Audit credits as they crawl. Project methods create, list and delete projects, and Team Management methods add, remove, activate and deactivate members. Together they let an agency automate a new client from account to first report.

AI Overview, trends, markets and live SERPs

getAIOSummary returns counts of keywords where your brand appears in AI Overview with a link, without a link, or as a domain only, and getAIOBrandOpportunities returns the full list; they arrived on April 27 and May 11, 2026. Few SEO APIs at this price expose AI Overview data this way. getKeywordTrends returns demand over up to 48 months, with year-over-year change and peak and trough months, which suits seasonality checks. getDomainsRatingData takes up to 200 domains and returns traffic, visibility, backlinks and more for each. Good for prospect lists. Finally, getKeywordSerp retrieves live Google results at the time of the call. Serpstat says this first release targets the MCP server and AI agents. Expect changes.

Connecting without much code

Most teams never write an HTTP client. Serpstat's Python guide uses a Google Colab notebook with examples for each method, needing only Python 3 and the requests library: paste your token, run a cell, inspect the JSON, then copy the working function into your own script. Analysts in R get an integration for analysis and dashboards built with Shiny. The Batch Analysis add-on pulls domain, URL and keyword reports into Google Sheets, where each domain, URL or keyword counts as one API request and each run writes a new tab. A sheet that checks 500 keywords every Monday spends about 2,165 credits a month, by our arithmetic.

The other routes are narrower. The Looker Studio connector offers two reports, Domain history and Competitors, which can be shared with clients who have no Serpstat account. Since July 3, 2026 Serpstat is a community node in n8n, with most standard methods available, which is the fastest no-code route: a node, a token, a schedule. An Airbyte connector moves data into warehouses, and many reports export straight to BigQuery. Make and any platform that sends a POST with JSON can call the API too; keep the token in the platform's secrets. The MCP server is a conversational layer on the same API and spends the same credits, so budget both from one pool. Our Serpstat MCP guide covers it in full.

What you may do with the data

You may use API data internally and in reports for your clients, and scripts and AI agents may call it directly. That covers most agency and in-house work. Products are a different matter: "Embedding, reselling, or showing raw Serpstat data inside a customer-facing product or dashboard requires a separate data licensing agreement." ^[1] One Trustpilot reviewer complains the API cannot be used commercially, but the rule is narrower. An agency can send clients a PDF or a Looker Studio report built on Serpstat data. A SaaS company that shows Serpstat keyword volumes inside its own app needs the license, even if the numbers are cached for weeks. Ask before you build.

API-only packages, SERP crawling and datasets

API-only packages exist on request. An older support page describes an API plan with faster requests, priced from $180 down to $100 per million rows depending on volume and term, so contact sales with your expected volume. SERP crawling sells live top-100 results apart from the plans, at $0.0013 per query under a million queries a month and $0.001 above, with a $300 minimum order ^[4]. It is a one-time service per order. For a product, the route is data.serpstat.com, which sells full keyword, domain, backlink, SERP and AI Overview datasets by API, bulk export or custom pipeline, and offers licensing for platforms that embed Serpstat data.

Budgeting credits: two worked examples

On Team, 200,000 credits a month allow about 6,600 rows a day. A weekly pull of 1,000 keywords for 10 competitors costs about 43,000 credits a month, and a daily 500-row keyword check on top brings it to about 58,000, a load Team covers more than three times over. A narrower competitor pull of 1,000 ranking keywords for each of five rivals costs about 5,000 credits, or roughly 21,650 a month if repeated weekly, about a tenth of the plan. All of these figures are our arithmetic.

An agency with 15 clients that pulls 2,000 domain keywords per client each month spends 30,000 credits. A weekly Summary report for each adds about 325, and daily Rank Tracker reads add nothing, so the total is about 30,325 credits of Team's 200,000, with a large margin left for MCP. Check usage weekly at first, in Profile, Credits and Subscriptions or through the API's credit stats. The common mistakes repeat:

• Pulling full reports when ten rows would do.
• Forgetting that empty responses still cost one credit.
• Running JavaScript audits through MCP at ten credits a page.
• Sharing one token across scripts.
• Building a product without a license.
• Copying old blog snippets with deprecated methods; check them against api-docs.serpstat.com.

Serpstat's API next to Semrush, Ahrefs and SE Ranking

Semrush lists API data integration on Advanced at $549 a month, and sells API units in packages of 2 to 20 million that expire if unused. Serpstat starts the API at $129. Semrush suits enterprise data teams. Ahrefs includes API and MCP access on Lite at $129, but with up to 100 rows per request and 200,000 integration units a month. Serpstat allows up to 60,000 rows per request, so pulling 10,000 keyword rows takes one request on Serpstat and 100 on Ahrefs Lite. For large pulls, Serpstat is roomier.

SE Ranking Core includes 25,000 API credits a month and MCP access, and sells an API add-on with 3 million credits for $45 a month on annual billing, plus a standalone API plan from $179 a month. On raw credit counts that add-on is larger than Serpstat Team's 200,000, so compare what a credit buys on each side before you decide. Use Serpstat's API if you are an agency building client dashboards, an in-house team feeding BI tools, or anyone running AI agents on SEO data. Skip it if you need live SERP data for every call, more than ten requests a second, or the right to show raw data in your own product. Skip it on Individual, too. There is no API.

The bottom line

The Serpstat API is the reason to buy Team. It brings 200,000 credits a month, free Rank Tracker reads and, since 2026, real write access for projects and tracking. Its weak points are speed and licensing. Take the Team trial, generate a token, run the Colab notebook and pull ten rows. If the data fits, build on it. If you need to sell the data, talk to Serpstat about a license before writing a line of product code. For how Team compares with the other plans, read our Serpstat pricing guide.