How to Use a Rank Tracker API for Keyword Snapshots
Get a domain's keyword ranking snapshot into an agent or script, store its scope, and compare reports over time.
Get keyword ranking data into your reporting workflow. Paste this into your agent:
Set up https://monid.ai/SKILL.md and use Monid to retrieve a small keyword ranking snapshot for my domain, preserving the report date, country, scope, positions and ranking URLs.
One prompt starts the workflow. 1 min setup.
SEO analysts using a rank tracker API need more than a position on a screen: they need a dated record they can join to the next report. Even a two-row export has a boundary: it cannot describe every keyword a domain ranks for.
Manual exports become awkward when the target, country, date or URL scope changes between files. A position without those settings is difficult to compare, and a report of stored observations is not the same thing as a fresh search results page.
This guide uses Monid, the OpenRouter for agent tools, to retrieve a small keyword report with one key and one balance. You will get a real sample, a command to repeat, and a storage plan that keeps each comparison honest.
Our November 20 test returned two rows for wikipedia.org in the US. The first returned row is shown here; complete: false tells us this limited response is not a complete domain report.
| Keyword | Best position | Ranking URL | Search volume | Estimated traffic |
|---|---|---|---|---|
| wikipedia | 1 | https://www.wikipedia.org/ | 3,320,000 | 2,147,662 |
The traffic figure is a modeled estimate, not measured visits. The report date selects a report; it does not identify when the underlying search result was checked.
How to use a rank tracker API with Monid
Use a small, scoped keyword report first, then save its settings alongside the returned rows. An endpoint is a callable operation: the Ahrefs organic keywords endpoint returns a domain or URL's ranked keywords and positions through Monid.
Step 1. Set up access
For an agent, get a key at app.monid.ai and give it the following instruction. The agent quickstart explains the setup.
set up https://monid.ai/SKILL.md
For a terminal, install the CLI and add your key. The CLI quickstart covers the same setup for a human operator.
npm install -g @monid-ai/cli
monid keys add -k <your-api-key> -l main
Step 2. Inspect the report
Discover finds tools and inspect describes their inputs and price; neither operation runs the paid report. Search for ranked keywords, then inspect the exact path before you choose a limit.
monid discover -q "google keyword rank position serp"
monid inspect -p ahrefs -e /site-explorer/organic-keywords
The report takes target and date as required query parameters. Set country to the market you want and mode to the scope you intend to compare; subdomains covers the domain and its subdomains, while exact targets the URL as written. Start with a small limit because billing follows returned rows.
Step 3. Request a small snapshot
Run this example to request at most two rows for a public domain. The query is a JSON object passed with --query, not a request body; replace the domain, date and country for your own report.
monid run -p ahrefs -e /site-explorer/organic-keywords \
--query '{"target":"wikipedia.org","date":"2026-11-20","country":"us","mode":"subdomains","limit":2}' -w -j
Our call completed with two rows and cost $0.03096, Verified 2026-11-20. Its timestamps were 1.909 seconds apart; that observation is not a service guarantee. The unit price is $0.01548 per returned row, Verified 2026-11-20, on the tool page. The reported charge matched two billed units at that rate.
Step 4. Keep the scope with the rows
Save the response and its scope together, including the target, report date, country, mode, protocol and limit. The response also includes rowCount and complete; preserve both so a later reader can distinguish a bounded sample from a complete response.
For a repeatable tracker, write each report to a dated file or database table and append new observations rather than overwriting the old ones.
Why scope matters before comparing rankings
A rank is useful only when you know what was measured and which report produced it. Keep the same target, country, scope and source across the observations you compare, and store the settings even when they look obvious today.
Choose a target deliberately
A domain-wide report answers a different question from a report for a single page. A domain view can show which pages contribute to visibility; an exact URL view isolates one page, but it does not represent the rest of the site. Decide which decision the report supports before you save the first row.
The exact and prefix modes match the target as written. In those modes, a www URL and its bare-domain counterpart are different targets in the report, even when a browser redirects one to the other. Store the exact input and avoid treating an empty result for one spelling as proof that the other spelling has no rankings.
Keep geography and time attached
The example selects the US and a report date of November 20, 2026. A report for another country is a separate observation, so compare it as a separate series instead of averaging its positions into the US report.
A report date is not the timestamp of every underlying SERP observation. The inspected endpoint describes stored SERP checks and returns no per-row check date. If the decision needs a fresh view of a search page, use the Google SERP checker guide to examine that different job.
What are the ways to get ranking data?
Choose a route based on whether the end product is a report, a historical dataset, or an on-demand view of a search page. The input and the freshness requirement matter more than the name of the interface.
Compare the routes
| Route | What you get | Best fit | Boundary |
|---|---|---|---|
| Monid keyword report | Structured rows for a target and report date | An agent or script that stores snapshots | Stored observations, not a live SERP check |
| Reporting dashboard | A visual workspace | An analyst who reviews reports manually | An export and separate storage are needed for a custom pipeline |
| Search results check | A view of a search page | Investigating the page layout for a specific query | A result page is not a domain-wide keyword history |
| Your own storage | A history built from saved reports | Comparing identical report scopes over time | You must choose a schedule and handle missing or partial observations |
The guide to storing keyword positions over time explains how to build a historical series from repeated observations. These routes can work together. Use a report to identify a keyword worth investigating, then use a results-page workflow to examine the page itself. Do not join a best-position field from one report to a different report's position field and assume the two are interchangeable; the endpoint documentation warns that their values can differ even for the same keyword and date.
A separate Bing position tracking guide covers a different search engine. Keep that series separate from a Google-oriented report so the label on a chart always identifies the source.
How to read and store the snapshot
The returned row is a report observation, not an instruction to change a page. Read the keyword and URL first, then interpret the position, volume and traffic estimate in the context of the report scope.
Interpret each field
keyword is the query associated with the row; best_position is the best position reported for that keyword; and best_position_url identifies the ranking URL. In our sample, the keyword wikipedia returned position 1 and the public Wikipedia home page as its ranking URL. That is one reported observation, not evidence that the same position appears for every searcher.
volume is a search-volume field and sum_traffic is estimated traffic. The sample's 3,320,000 volume and 2,147,662 estimated traffic are different measures, so do not add them or label the traffic figure as actual visits. Use modeled traffic for relative sizing within the same provider and report, and use measured analytics when the question is actual site visits.
The second returned row also had position 1. Both positions describe the returned two-row sample only. The response explicitly marked the sample incomplete, so it cannot establish the number of all keywords or the full distribution of positions for the domain.
Store an observation rather than a conclusion
Use a key such as source, target, country, mode, report date and keyword to identify an observation. Keep the ranking URL, position, volume and estimated traffic as values, and retain the raw response or a reference to it for later review.
For a movement table, join observations only after matching those scope fields. A move from position 8 to position 3 in the same series is a five-position improvement; a missing row is not automatically a fall to zero. It can reflect a response limit or a keyword absent from that particular report. Label missing values separately from measured positions.
A row limit is a cost and completeness decision. A report with complete: false should be labeled partial in downstream dashboards; comparing two partial lists can otherwise make a keyword appear to disappear simply because it was outside one returned set. Increase coverage only after choosing a budget and a way to handle the returned set.
How to use the data in your own code
Treat each run as a batch with its own metadata, and let your script transform the response only after it checks the run status. Monid's CLI can wait for the completed response with -w, as in the tested command above.
Build a durable comparison
Start with a raw response store, a normalized observations table and a report metadata table. The raw record preserves the original values; the normalized table supports joins; and the metadata table records the input, run identifier, result count and completeness flag. This separation makes it possible to correct a transformation without paying for another report.
Use a stable identifier for each report and record the provider and endpoint as part of the series. Keep numbers as numbers, retain the full URL, and preserve the report date in a date column rather than embedding it only in a filename. A chart can then show the position for a keyword and link each point back to its source report.
Set a budget before scheduling
The inspected endpoint defaults to 100 rows and accepts a limit up to 100. Set an explicit limit in automated jobs: the limit bounds returned rows, while the charge depends on how many are returned. Discover and inspect are free; the data request is the paid operation. The endpoint's inspect notes say an empty result is free; this test returned two rows and did not test an empty response or a failed request.
Your scheduler, retries, storage and alert rules are your own application work. Record a failed or missing report as a missing observation and investigate it before producing a movement alert. Do not reuse an old value under a new date, because that disguises a gap in the series.
When to use another route
Use a dedicated search results check when you need the page as it appears for a query, or use an existing SEO dashboard when manual review is the whole job. This is the Monid blog, and Monid provides access to the Ahrefs report used here; the report's stored observations do not provide an on-demand search page.
Match the data to the decision
A fixed list of keywords that must all be checked at a chosen time needs a workflow designed around those queries. This domain report returns ranked keywords for a target and can be bounded by a row limit; it does not itself schedule checks or maintain your historical database. If you need those capabilities, build them around the report or select a tool that already supplies them.
Search environments also differ. Product search and web search measure different environments. Keep their results in separate series from the domain report in this article.
Which endpoint fits this job?
The tested report is the right starting point for a small domain keyword snapshot. Its price depends on returned rows, so an explicit limit is part of the request design.
| Use case | Best endpoint | Why? |
|---|---|---|
| Store a dated keyword snapshot for a domain | Ahrefs organic keywords | Keyword, position and URL in the tested response; $0.01548 per result, Verified 2026-11-20. |
Conclusion
You now have a tested keyword snapshot, a command to request another one, and a way to preserve its scope for comparison. Repeat the report with the same settings, label partial results, and investigate missing observations before treating them as movement. A consistent series gives you a clearer basis for deciding which pages and keywords need attention.
FAQ
Is the rank tracker API free?
Discovery and inspection are free; the tested report charges per returned row. The current tool listing shows the report's price and billing shape.
Do I need to install anything?
You can give the setup prompt and your key to an agent, or install the Monid CLI for terminal use. The CLI commands above show the latter route.
Do I need to write code?
No code is required to ask an agent for a small report. A script or database is useful when you want to save and compare repeated snapshots automatically.
How many keywords can one report return?
The inspected report accepts a limit of up to 100 rows per request. Set the limit explicitly and read complete before treating a response as a full report.
Does an empty or failed request cost anything?
The endpoint's inspect notes say an empty result is free. Our test returned two rows, so it does not establish how a failed request is billed.
Can I schedule ranking reports?
Yes, you can schedule calls in your own application and save each response. The endpoint itself does not create your schedule or historical database.
Are these positions live and exact for every searcher?
No. The report uses stored SERP observations, and its report date is not the date of each underlying check. Compare like-for-like reports and use a search-page check when the current page is the decision you need to investigate.
Published November 2026.


