Docs
Endpoint reference and pricing
One URL, many actions. This is the whole surface.
SEOCheck research team ·

Actions and costs
| Action | Credits | What it returns |
|---|---|---|
| domain | free | Basic record and DR for one domain |
| overview | free | Summary of a domain's link profile |
| stats | free | Index-wide statistics |
| suggest | free | Domain suggestions as you type |
| bulk_dr | 1 | DR for up to 2,000 domains |
| compare | 1 | Side-by-side for up to 20 domains |
| top | 1 | Top domains |
| dr_distribution | 1 | Spread of DR among linking domains |
| tld_breakdown | 1 | Links by top-level domain |
| refdomains | 2 | Referring domains |
| backlinks | 2 | Individual backlinks |
| anchors | 2 | Anchor texts |
| toppages | 2 | Most linked pages |
| anchor_cloud | 2 | Anchor weights |
| intersect | 2 | Shared linkers for 2 to 10 domains |
| search | 3 | Full-text over anchors, titles and URLs |
| keyword | 3 | Who is linked with an exact anchor |
Auth and shape
Base URL: https://seocheck.dev/api.php?action=<action>&q=<domain>&key=sk_.... Prefer headers? Send Authorization: Bearer sk_... instead.
This backlink analysis API answers in JSON and allows cross-origin calls, so a browser can hit it directly. Every response has an ok field. Authenticated ones also carry _credits, which holds the cost of that call and your remaining balance.
Errors are blunt. HTTP 401 means the key is missing or wrong. HTTP 402 means the balance ran out. Both are cheap to handle, and neither should be retried in a loop.
Before the first paid call
- Run
overviewon a domain you know. It is free. - Check that your key works with one 1-credit call such as
compare. - Read
_creditsin the response and confirm the cost matches the table above. - Set a script alarm for HTTP 402 so a dry balance does not fail silently.
Paging and sorting
Rows per call go up to 1,000. Ask for fewer or more with per, from 10 to 5000, walk with page, and order with sort and order.
curl -s "https://seocheck.dev/api.php?action=backlinks&q=example.com&per=500&page=2&sort=dr&order=desc&key=sk_live_..." | jq '._credits'
Parameters that change the answer
Every action takes the same two basics, action and q, and then a short set of options. Read them first. Most mistakes come from options that look optional but change the cost or the shape of the rows. Read the options before you copy a call from a forum or a blog.
The examples below assume a key in the key parameter. Replace the placeholder with your own. Keep the key in an environment variable, not in a file you commit. Every example in this backlink analysis api reference uses the same base URL.
Paging with page and per
The per option sets how many rows come back, from 10 up to 5000. The page option moves you through the set. A call that asks for 500 rows and then page 2 returns the next block, not the next pull of the same rows.
Deep pagination stops at 50,000 rows. If a job needs more than that, narrow the query with sort or split the work by domain, so every call stays inside the cap.
Sorting with sort and order
Sort on the server, not in your script. Asking for sort=dr&order=desc returns the strongest linking domains first, so the rows you care about arrive on the first page. Filtering after a full download wastes both time and credits.
Use order=asc when you want the weakest rows, for example to find links that are thin or to spot a sudden drop in a profile.
Lists: bulk_dr, compare and intersect
The list actions take several domains in one call. bulk_dr accepts up to 2,000 domains for one credit, with a GET query string or a POST body. compare takes up to 20 domains, and intersect takes two to ten.
Since the price does not depend on the row count, a long list in one call is cheaper than many short calls. Say you check 1,500 prospect sites on Monday. That is one call, not 1,500.
Search and keyword calls cost more
The search action runs full-text over anchors, titles and URLs. The keyword action asks who is linked with an exact anchor. Both cost three credits per call, so use them to answer a specific question rather than to browse.
Choosing per and page for a first pull
For a first look at a new domain, start with a small per, such as 50. You will see the shape of the rows, the field names and the cost in one short response. Raise per only when you know the rows are what you need. The same habit works for every action in the table above.
A common mistake is to request 5000 rows on the first try and then discard most of them. The response is larger, the wait is longer and the saving you hoped for does not appear.
Reading the credit meter
Every authenticated response ends with _credits, which gives the cost of that call and your balance after it. Log it. Add it up at month end and you will know your real bill to the cent.
A rough budget helps. A thousand refdomains calls cost 2,000 credits, which is $0.20. A thousand search calls cost 3,000 credits, or $0.30. A daily bulk_dr run on a 2,000-domain list costs 1 credit, so about $0.003 per month.
Compared with plans that start at $129 per month for 200,000 API units, as Ahrefs lists them, the shape of the bill is different. See the comparison for how.

Reading errors and the _credits field
A response can fail in two different ways, and your code should treat them differently. Those are different problems. A bad key is a configuration problem. An empty balance is a billing problem. Handling both as one generic error hides which fix you need.
Read the ok field first. Then read the _credits block. Only after that should you look at the rows, because a missing or short result is easier to explain when you already know what the call cost.
Handle 401 and 402 separately
HTTP 401 means the key is missing or wrong. Stop the job and fix the key. HTTP 402 means the balance is spent. Top up from the account page, then carry on. Retrying a 401 never helps, and retrying a 402 without topping up never helps either.
Log the balance before and after a run
Write down the _credits balance at the start of a run and at the end. The difference should match what you expected from the calls you made. A larger jump means a loop ran more often than you meant it to.
For a quick check, multiply the number of calls by their credit cost. If the answer is 40 and the balance dropped by 4,000, something is wrong with the script, not with the price list.
Check the rows you asked for
A successful response with fewer rows than expected is not automatically an error. A small domain may simply have few links. Compare the row count with the per value you sent, and treat a gap as a question to answer, not a fault to hide.
Test the cheap calls first
Before you wire a call into a job, run it once by hand with the smallest sensible options. Check the response, the field names and the _credits value. A job that breaks on the second row is much harder to debug than a call that failed on its first try.
A first session with the API
Four calls, in order, so that every later call is one you already understand. Only the first one is free of cost, and that is intentional.
Run overview on a domain you know
It is free and needs no key. Compare the summary with what you see in the search console and the site itself, so you know what a normal answer looks like.
Test the key with one 1-credit call
Send compare with your domain and one rival. Confirm the ok field is true and that _credits shows a cost of one credit.
Pull a 2-credit listing with a small per
Ask for refdomains with per=50 and sort=dr. Check that the row count and the credit cost match the table above.
Write the cost into your log
Record the balance before and after. From now on, every script you write starts with that habit, and a surprise bill becomes a line in a log instead of a mystery.
Pricing
Prepaid credit. One credit is $0.0001. There is no subscription and no monthly minimum.
Pay as you go
$0.10per 1,000 one-credit calls
- $100 free credit on sign-up
- No card needed
- Top up with USDT (TRC-20)
- Billed once per call, not per row
Limits you should know
The index covers 181,132,097 domains and 2,774,625,913 domain links. It refreshes roughly monthly. That is slower than a daily crawl.
Deep pagination stops at 50,000 rows. For bigger pulls, narrow the query with sort or split by domain.
Sign-in is Google only, and top-ups are in USDT. Start with the DR checker, try a rival audit with the competitor guide, or read why some teams drop the Ahrefs API for pure link work.
FAQ
Do I pay per row?
No. A call is billed once, whether it returns 10 rows or 1,000.
What happens when credit runs out?
The API returns HTTP 402. Top up with USDT (TRC-20) from the account page and carry on.
Where do I find my key?
In the dashboard after Google sign-in.