$100 free credit on sign-up, then $0.10 per 1,000 calls

Import recipe

Determine Backlinks for Every Prospect in Your Import

A clean enrichment job is four requests and a date column. This page lists what to store, what each call costs and where imports quietly go wrong.

Isometric miniature bridges connecting many distinct islands to represent bulk prospect enrichment

Pre-flight list for the import script

  • Strip protocols, paths and www. so every row is a bare domain
  • De-duplicate before calling; a repeated domain still costs a credit slot
  • Chunk into batches of at most 2,000 for bulk_dr
  • Write the score and a fetch date into separate columns
  • Treat HTTP 401 and 402 as stop signals, not as empty results
  • Log the credit balance from _credits after every batch

Field reference: which action fills which column

Row limits and prices from the public price list. 1 credit = $0.0001.
ActionCreditsInputWhat you can store
domain / overview / stats0One domain in qHeadline counts and DR for a single lookup
suggest0Partial nameAutocomplete when a user types a publication
bulk_dr1Up to 2,000 domainsDR for each domain in the batch
refdomains2One domain, up to 1,000 rows per callThe list of domains linking the prospect
anchors2One domainAnchor texts pointing at it
toppages2One domainIts most linked pages

How to determine backlinks for 5,000 rows without overspending

Start with scoring. Stop there for most rows. A bulk_dr pass over 5,000 domains is three calls, so 3 credits in total, a tiny fraction of a cent per row.

Deeper pulls are for the slice you will actually email. If 300 prospects clear your DR cutoff, refdomains for each costs 600 credits, or six cents. Do the same for all 5,000 and the bill is $1, still small, but the output is mostly noise nobody reads. Keep it lean.

Order matters here. Score first, cut by threshold (see the DR cutoff table), then enrich the survivors. Afterwards you can compare the finalists in pairs.

Miniature figure inspecting a small plant on a rock to symbolize checking subdomains

A note on subdomains and redirects

Prospect lists carry messy hosts: m.example.com, blog.example.com, a vanity domain that redirects. Pick a rule and write it in the job comment.

Our recommendation is to score the registrable domain, keep the original host in a separate column, and flag any row where the two differ. A newsletter hosted on a platform subdomain is the usual trap, because the platform's authority says nothing about the writer. Check those by hand before the first send. Do not skip it.

Rows that were never domains

Most lookups that come back empty were never valid domains in the first place. Mixed input is the usual cause, and a column that blends web addresses, contact names and free text sends junk to the API. Clean it before spending a credit.

Picture a sheet of 4,000 contacts where the outlet column sometimes holds a name in front of the site. Keep the domain part only. Write each change into a note, so it can be undone later.

Strip the parts a browser needs and a domain does not

A pasted link carries a scheme, a path, a query string and sometimes a port. None of those belong in q. Remove them in one pass, then check that what remains contains a dot and no spaces.

Anything with a space in it is not a domain, so send it to a review list instead of the API. A quick count of those rows often shows the source of the mess, such as one importer that copied whole addresses from a browser bar.

  • Protocols such as https:// and http://
  • Everything after the first slash, including the path
  • Port numbers such as :8080

Score the outlet, not the byline

Many outlet lists store a person's name in the same column as the publication. A byline or a newsletter brand is a label for a contact, not a host. Score the publication domain, then keep the contact name in its own column.

Say three reporters share one newsroom. All three rows should receive the same DR, because the domain is the same. Dedupe the domains, then join the scores back to every contact, so one lookup serves the whole team.

Handle a domain that moved or redirects

Publications change hosts. A title can migrate to a new domain, or a long-running site can redirect to a parent brand. The old name in your list then describes a site the editor no longer runs.

Check the redirect once by hand. Record the final host and mark the row, so the old name is not scored again next month. Repeating that check on every refresh adds cost and tells you nothing new.

Batch call with a POST body

GET works for short lists. For a full chunk, send the domains in the body so the URL stays short.

bash
curl -X POST "https://linkfluence.app/api.php?action=bulk_dr&key=sk_live_..." \
  -d "domains=techcrunch.com,hubspot.com,theverge.com"

# then, for the shortlist only
curl "https://linkfluence.app/api.php?action=refdomains&q=hubspot.com&per=1000&key=sk_live_..."

Order the job so no credit is wasted

01

Normalise the column

Apply the cleaning rules, then count the rows that failed. A rejected row is a finding, so keep the list of them.

02

De-duplicate the whole file

One domain goes into the request once, however many contacts sit behind it.

03

Score in batches of 2,000

Send each chunk to bulk_dr, read ok first, then store the score and the fetch date for every row.

04

Pull depth only for the cut

Apply your threshold from the cutoff table, then call the 2-credit actions only for rows that passed it.

05

Log the balance

Write the _credits balance after each chunk, so a run that stops halfway shows exactly where it stopped.

Mistakes that corrupt a CRM, and the fix

MistakeWhat happensFix
Overwriting yesterday's score with a blank after a 402Prospects look unscored and drop out of the queueHalt the job on any non-200, keep the old value
Scoring subdomains such as blog.example.comCounts reflect the host, not the brand your contact works forDecide per row whether to score the root domain or the host, and record which
No date columnNobody can tell a March score from an October oneStore the fetch date beside every number
Re-pulling refdomains every weekPaying twice for data that refreshes monthlyCache per domain until the next index cycle
Using DR as an absolute truth in client talkA client compares it with Ahrefs and sees a mismatchLabel it as Linkfluence DR in every export
Brass compass on a wooden table pointing to a horizon, symbolizing clear rules

Questions about the enrichment run

The questions below come up on almost every import. Each answer is short, and each one points back to a rule set out earlier on this page.

Why does one row come back without a score?

Usually the input was not a bare domain, or the value was cleaned differently from the rest of the batch. Check the cleaned value first. A 401 or 402 means the whole job should stop, while one empty field is only a bad row, and the two need different handling.

Can I run the job once and never again?

You can, but a one-off score goes stale. The index refreshes roughly once a month, so re-score the active list after each refresh and keep the earlier value as history. A placement check covers the new links that appear between runs.

Is a number from 0 to 100 enough to decide?

It is enough to sort, not to decide alone. Use the score to order the list, then add a fit check and a timing check. No single number can tell you whether an editor covers your topic this quarter.

What belongs in the CRM notes for each row?

The score, the fetch date, the cleaned host, the rule that changed it and the batch number. With those five fields anyone can rebuild the number later without asking you.

How do I estimate the credits before I run a job?

Count the calls, not the rows. A bulk_dr call takes up to 2,000 domains for 1 credit, so 10,000 domains is five calls. Add the 2-credit pulls for the survivors, then multiply by $0.0001 per credit.

What a finished enrichment file should contain

A finished file is one you can explain six months later without the person who wrote the script. Every number should carry the reason it exists, and every skipped row should carry the reason it was skipped. That is most of the discipline. It costs almost nothing once the job runs in batches.

Think of the file as a record of decisions. The score is only one column. The cleaned host, the threshold that selected the row and the date of the last deeper pull tell the next person what happened and whether it still holds.

The columns that should survive the job

Keep the raw value exactly as it arrived, then the cleaned domain beside it. When a score looks wrong, the raw value is what lets you trace the mistake back to the import rather than blaming the API.

Store the DR, the fetch date and the batch number as separate fields. Mixing them into one text cell makes filtering by date impossible, and filtering by date is the first thing a reviewer will ask for.

  • Raw value as imported
  • Cleaned registrable domain
  • DR and the fetch date
  • Batch number and credits charged

Rows you should never silently drop

Some rows fail for reasons that matter later. A redirect, a parked domain or a site that returns nothing at all can all look like empty data. Mark those rows instead of deleting them, so the gap is visible in the file.

Say a prospect returns no referring domains on a deep pull. That may mean the site is new, the site is small or the row is bad. The file should say which, because each answer leads to a different action.

A file a client or teammate can audit

Anyone reading the sheet should be able to reproduce a single number from the notes alone. If they cannot, a field is missing from the notes, and the gap belongs to the notes rather than the reader.

Keep one export per run rather than overwriting the master sheet. Then a question about last month's shortlist has an answer, and the shortlist comparison can be rerun against the same inputs.