How to Clean a Backlink Export Into Referring Domains

Convert a page-level backlink export into a reviewable referring-domain report while preserving counts and the original evidence.

FlashDomains Editorial·August 21, 2026·7 min read

A backlink export is usually organized around links or target pages, not the organizations behind them. The same publisher can appear hundreds of times through article URLs, pagination, author archives and syndicated copies. Counting those rows as independent websites exaggerates the size of the referring-domain set.

A clean referring-domain report groups source URLs by registrable root domain while retaining frequency and provenance. Start with the Backlink audit preset when your export is ready.

What a useful referring-domain dataset contains

Field	Keep it?	Why
Original source URL	Yes	Lets you inspect the actual linking page
Root domain	Yes	Provides the grouping key
Hostname	Often	Reveals hosted platforms and meaningful subdomains
Occurrence count	Yes	Shows concentration in the export
Target URL and anchor	Yes, in the source file	Needed for link-context analysis

Backlink-export cleanup workflow

Export the page-level records. Retain source URL, target URL, anchor text and any provider metrics. Crawlers such as Screaming Frog can bulk-export link data; the domain extractor does not replace collection.
Extract root domains from the source-URL column. Root mode turns page URLs into one registrable website key.
Deduplicate and count. Merge repeated roots while keeping their occurrence count and source rows.
Filter obvious noise. Review your own domain, staging services, link shorteners and hosted-platform boundaries deliberately rather than deleting them automatically.
Return to the source evidence. Use the mapped line or original export whenever a domain needs manual evaluation.

Practical reference: examples, comparisons and workflowExpand for a worked example, a step-by-step checklist and related guides.

Repeatable process

Clean a backlink export without losing evidence

Produce an organization-level view while keeping a route back to every source link.

1
Keep the source URL column
Do not replace the original export; copy the linking-page URLs into a working dataset.
2
Extract registrable domains
Collapse paths and subdomains when the goal is a unique referring-site count.
3
Count occurrences
Retain how many exported rows map to each domain instead of deleting duplicates blindly.
4
Join metrics after cleanup
Attach authority, traffic, status or review fields to the normalized domain table.

Side-by-side

Link rows and referring domains serve different analyses

Keep both views available rather than forcing one table to do every job.

Swipe the table sideways to compare all columns.

Link rows and referring domains serve different analyses
Decision	Backlink-row view	Referring-domain view
Unit	One discovered link or page	One registrable domain
Best for	Placement and page-level review	Diversity, outreach and concentration
Duplicates	Often meaningful	Grouped into frequency counts

Where it helps

Analyses unlocked by clean referring domains

Use the cleaned output as a focused input to the next analysis—not as a substitute for that analysis.

Link concentration

See whether a large backlink total depends on only a few sites.

Prospecting

Remove organizations that already link before building a new outreach list.

Competitor overlap

Compare unique linking sites across several exports.

Manual review queues

Prioritize suspicious or strategically important domains for deeper inspection.

Worked example

Collapse linking pages into referring domains

Input

https://publisher.com/post-a
https://publisher.com/post-b
https://news.example.org/story

Result

publisher.com — 2 links
example.org — 1 link

Keep the original rows if you later need anchor text, target URL or placement details.

Root domain or hostname for backlink analysis?

Root domains are the normal starting point because they estimate unique registered sites. Hostname mode becomes useful when one platform hosts unrelated customers on private suffixes, or when separate subdomains represent materially different properties. Run both views when uncertainty matters.

The detailed URL-to-domain workflow explains suffix parsing and mixed input. Do not use a spreadsheet rule that simply keeps the final two labels; it fails on domains such as publisher.co.uk.

What frequency does—and does not—tell you

A high count means a domain appears many times in this export. It does not automatically mean the domain is authoritative, harmful or more valuable. Sitewide navigation, pagination and duplicated crawl paths can inflate counts. Treat frequency as a queueing signal, then inspect link placement, relevance and the original pages.

Common backlink-cleaning mistakes

Deleting the original URL column after grouping.
Counting links and referring domains as the same metric.
Collapsing hosted customer sites to the platform domain without private-suffix rules.
Removing subdomains before deciding whether they represent independent properties.
Treating ownership details as proof of link quality. A WHOIS or RDAP lookup adds registration context, not an SEO verdict.

Frequently asked questions

Does deduplication remove useful information?

It can if you export only the final TXT list. Use CSV or JSON when you need counts, source lines and original values alongside the grouped domain.

Can this calculate domain authority?

No. It cleans and structures identifiers. Provider metrics, relevance, traffic and editorial context require separate analysis.

Should I disavow every suspicious domain?

No. Domain extraction is not a disavow recommendation. Review Google’s current guidance and the underlying links before taking action.