Troubleshooting a donor scoring run: symptom, cause, fix
Most failures in a scoring script are boring. The key, the balance, the list formatting. This table covers them in the order they usually appear.
SourceRanker editorial team ·
Symptoms and fixes
Symptom
Likely cause
Fix
HTTP 401
Missing or wrong key
Check the key= value or Bearer header; copy it again from the account page
HTTP 402
Credit has run out
Top up in USDT (TRC-20) from the account page, then resume from the last saved batch
Response says ok is false
Bad parameter
Read the message; usual culprits are a missing action or an empty domains list
A domain scores 0 or comes back empty
Almost no known links, or a typo
Strip the protocol and any path; try the free domain action on it
Score unchanged for weeks
Index refreshes about monthly
Wait for the next cycle; do not re-run daily
A fresh link is not visible
Link is newer than the last crawl
Recheck next month
Batch seems truncated
More than 1,000 rows requested on a detail action
Page through with page and per; deep paging stops at 50,000
Totals differ from another tool
Different crawl and formula
Pick one metric per project
Reading the credits field
Each authenticated reply tells you what the call cost and what is left. Trust that field over your own arithmetic.
A common budgeting slip is treating row count as cost. Our calls are billed once per request, so scoring 2,000 domains in one bulk_dr call costs one credit, while 2,000 separate calls cost 2,000. The first is $0.0001. The second is $0.20, still small but needlessly slow.
Free actions return no _credits charge because they need no key. If you see a charge you did not expect, you probably called a paid action by name.
A third slip is retry loops. If your script retries a failed paid call five times without checking the status, you pay five times for the same rows. Check for 401 and 402 first and stop. Only retry on network faults, and cap it at two attempts.
Log the number after each batch.
Before you file a question about a bad result
Did you clean every row to a bare host name, with no protocol?
Is the list already de-duplicated?
Did the previous batch save to disk before this one started?
Have you compared against the free domain action for one sample row?
Is the other tool's number from a different crawl, and are you comparing like with like?
When the number is right and still looks wrong
A DR score reflects the count of referring domains and nothing about a site's niche. A well-linked recipe blog and a well-linked casino both score alike.
Open the site. Run anchors on it. A page whose inbound anchors are all commercial phrases is telling you something that DR cannot.
For the full list of actions, see the reference, and for a clean-up recipe the bulk guide.
Seasonal sites add another wrinkle. A tourist board or a tax-advice blog may collect links in bursts around events. The monthly index catches the burst late, so the DR you see can lag the real picture by up to a cycle. Note the date of your pull next to every score, as you would date a quote from Google Search Console.
Short rows hide long stories, so read before you cut.
Rule of thumb
Stop on 401 and 402, retry only on network faults
Keys and balances do not fix themselves between two attempts. Save the batch, fix the cause, resume from the saved file. Any other approach pays twice for the same rows.