What a backlink gap actually measures
A backlink gap is a set difference: a referring domain appears in a competitor’s supplied records but not in the baseline records for your site. The workbench above compares one baseline and up to two competitors in your browser. It does not discover backlinks or log in to a provider. Semrush also compares referring-domain sets; its documented service is a separate product, not the data source of this exercise.
Let O be your observed source domains and A and B the competitors’ sets. The broad gap is (A ∪ B) − O. A stricter shared-competitor gap is (A ∩ B) − O. Neither expression says that a publisher would accept your contribution, that the link is still present, or that acquiring it would improve rankings.
Make the exports comparable
Choose competitors that serve a comparable audience or content need. Export the same kind of record, use equivalent target scope and link filters, and record each actual export date. A whole-domain export and one filtered to a single landing page answer different questions. If dates differ, the workbench displays a warning rather than pretending the snapshots are simultaneous.
Each file needs Source URL and Target URL columns; Anchor, Rel and Nofollow preserve optional context. One file must resolve to one target domain, and the targets must differ. Invalid URLs, ambiguous columns and mixed targets stop the comparison so a broken baseline cannot quietly create false gaps. The limit is 2,000 records and 1 MB per file. Keep the original exports: unrecognized columns are not included in the evidence download.
Normalize domains without losing evidence
The comparison groups source URLs by registrable domain, including private suffixes so separately hosted tenants such as different blogspot.com blogs stay separate. Ordinary subdomains of the same domain are combined. Trailing dots and internationalized host spelling are normalized; full source/target URLs, fragments, anchor and recorded Rel/Nofollow remain in retained evidence. Links within the detected target domain are excluded as internal records.
Exact repeated observations collapse with a Copies count; different paths, fragments or attribute evidence can remain distinct. Domain membership is binary even when there are many observations. A competitor match counts datasets, not the number of links. This is a declared grouping rule, not proof that separate domains have independent owners. Use the analysis guide to inspect raw distributions before comparison.
Reproduce the five-domain matrix
Load the fictional example. Your file has four records: two external observations, one repeated record and one internal record. Competitor A has five records that reduce to four external observations; B has three. The combined matrix has five source domains, three export gaps, one domain shared with your export and one found only in your export.
Alpha is present in all three datasets and is not a gap. Beta has two observations for A but only one competitor match. Gamma appears for A and B, so it is the sole result when you require both competitors and keep the missing-from-baseline filter. Delta appears only for B. Owned-only remains visible when you show all domains and allow zero competitor matches. No real publisher or index was contacted for this exercise.
| Source domains | Your export | Competitor A | Competitor B | Competitor matches |
|---|---|---|---|---|
| alpha.example | 1 | 1 | 1 | 2 |
| beta.example | 0 | 2 | 0 | 1 |
| gamma.example | 0 | 1 | 1 | 2 |
| delta.example | 0 | 0 | 1 | 1 |
| owned-only.example | 1 | 0 | 0 | 0 |
Qualify a gap before choosing an action
Open a domain row to inspect the supplied source URLs and record a decision. Review the source page’s topic, audience, language, relationship to the competitor, link placement and destination. Check whether your own resource would add a useful contribution. A research citation, a paid placement and an automatically generated directory entry may produce similar set membership but deserve different decisions.
The example worksheet records three hypothetical judgments. Gamma is worth considering only if the cited research need fits your resource; Beta stays on hold until the placement context is understood; Delta is excluded for a mismatch. These are invented teaching notes, not findings from those domains. Choose Consider, Hold or Exclude with a reason and a concrete next check; there is no automatic quality score.
| Source domains | Review decision | Evidence and next action |
|---|---|---|
| gamma.example | Consider | Consider only after checking that the research citation fits our resource. |
| beta.example | Hold for evidence | Hold until the placement context and commercial relationship are verified. |
| delta.example | Exclude | Exclude in this fictional exercise because the audience and language do not fit. |
Download the example and your review
The four downloads below let you reproduce the example and read a completed decision worksheet. In the workbench, the complete matrix includes every domain; the filtered download follows the current filters. Pagination changes only the visible ten-domain page. Full evidence includes every retained external observation, its original supported attributes, source file, date and duplicate count. The example source URL in each matrix column is only one example, not the entire evidence set.
Nothing is automatically saved to your account or browser storage. Download your decisions before changing input, because new files, dates or scope clear the comparison and review notes. CSV formula protection may prefix risky text. Keep the original files privately alongside the matrix and evidence export; a downloaded review worksheet is an output, not an importable saved session.
Download complete matrix (CSV)
Repeat the comparison without inventing growth
A domain absent from one export may still link to you. Search Console’s Links report is sampled. Different collection windows, filters and grouping can also change observed membership. A zero here means no accepted observation in this supplied dataset, not verified absence on the web. Preserve these limitations in the source-and-filters field.
For a repeat review, use comparable dates and scope, preserve the previous exports, and distinguish newly observed domains from newly published links. The workbench is a snapshot comparison, not a historical acquisition tracker. Compare measured business outcomes separately and never turn a change in source counts into an unsupported claim about ranking or traffic growth.
Choose the next workflow
Use the link monitor when you have a concrete source/target pair to verify over time. An unresolved fetch needs investigation rather than an automatic “lost link” conclusion. If a candidate passes your manual research, the outreach workspace can organize prospects and drafts; this comparison does not send messages or add a prospect automatically.
This guide owns reproducible domain-set comparison and qualification. A full competitor dossier also examines content, linked pages and acquisition context. An audit owns verified defects and remediation. Use the report template to communicate your dated evidence and decisions without presenting export gaps as guaranteed opportunities.