usdigitalmarketingindex.com SEO
How should unique referring domains be measured after a digital PR campaign?
Updated November 15, 2026
Count each linking site once, however many links it sent, and count it only if it was not already linking before the campaign and its link can be traced to the campaign. That requires three things fixed in advance: a dated baseline of the domains that already link, a rule for what counts as one domain, and a rule for what counts as attributable. With those in place, the campaign’s figure is the number of new, normalized, attributable domains observed in a stated window, reported with the source and date of each observation. Without them, the same campaign can be reported with a small number or a large one, and neither means anything.
Campaign baseline
Take the baseline before launch.
- Export referring domains from each source you will use: the Links report in Search Console and, if used, a third-party backlink index.
- Record the export date for each.
- Keep the files unaltered.
Use the same sources at the end. Each tool sees a different portion of the web, so a count from one tool at the start cannot be compared with a count from another at the end. Where two sources are used, keep them in separate columns and do not add them together.
The baseline answers one question later: was this domain already linking?
Unique domain normalization
Decide what one domain is, and apply it to every row.
| Rule | Example | Counted as |
|---|---|---|
| Ignore protocol | http and https versions | One |
| Ignore www | www.example.com and example.com | One |
| Subdomains of one publisher | news.example.com and blog.example.com | One, by default |
| Hosted platforms | alice.platform.example and bob.platform.example | Separate; different publishers on a shared host |
| Country domains of one publisher | example.com and example.co.uk | Separate domains; note common ownership |
| Syndicated copies | The same article on 30 sites of one newspaper group | 30 domains; flag as one piece of coverage |
| Multiple links from one site | Five articles on one site | One domain |
| Sitewide links | A footer link on 4,000 pages | One domain |
The hosted-platform and syndication rows are where counts inflate or deflate. Write the rule down and note in the report how many domains fall under each special case.
Attribution and timing
A new domain is not automatically a campaign result. Links arrive for other reasons during the same weeks.
Attribute a domain to the campaign only when at least one of these holds:
- the linking page discusses the campaign’s story, data, or asset;
- the link points to the campaign’s page;
- the outlet was contacted and the coverage followed;
- the article cites another outlet’s coverage of the campaign.
Everything else that is new goes in a separate line, “new, not attributed”.
Timing rules to set in advance:
| Parameter | Decision to record |
|---|---|
| Window start | Launch date |
| Window end | A fixed date, chosen before launch |
| First seen date | The date the tool first reported the link, which trails publication |
| Publication date | The date on the article, where shown |
| Late coverage | Reported in a follow-up, not added silently to the original figure |
| Lost links | Domains whose link disappeared within the window, reported separately |
Tools discover links with a delay that differs by tool and by site. A count taken on the last day of the window understates what was published in it. Take a second reading later, and report both with their dates.
Qualifying earned citations
A campaign report schema:
| Field | Content |
|---|---|
| Normalized domain | After the rules above |
| Linking URL | The page |
| Target URL | Where it points |
| Publication date | From the article |
| First seen | Date and source tool |
| In baseline? | Yes or no |
| Attribution basis | Which of the attribution rules applies |
| Coverage group | Identifier shared by syndicated copies of one article |
| Link attributes | <!–INLINECODE0–> values as published |
| Classification | Earned editorial, paid or sponsored, directory, scraper, other |
| Qualifies | Yes or no |
| Verified by | Person and date the page was opened |
Qualifying rules for the headline figure:
- not in the baseline;
- attributable to the campaign;
- the linking page was opened and the link seen by a person;
- classified as earned editorial coverage;
- counted once per normalized domain.
Excluded from the headline, and reported on their own lines: scraper copies of coverage, paid or sponsored placements, pre-existing domains that linked again, and new domains with no connection to the campaign.
A reporting layout, with made-up figures:
| Line | Count |
|---|---|
| New domains seen in window (source A, read on stated date) | 61 |
| Less: not attributable to the campaign | 14 |
| Less: scraper and aggregator copies | 19 |
| Less: paid or sponsored placements | 3 |
| Qualifying unique referring domains | 25 |
| Of which syndicated copies of one article | 8 (from 1 piece of coverage) |
| Distinct pieces of coverage | 18 |
| Existing domains that linked again | 6 |
| Mentions without a link | 11 |
The subtraction checks out: 61 less 14, 19, and 3 leaves 25. Of those 25, 8 are syndicated copies of a single article, so the distinct pieces of coverage number 17 others plus that one, 18.
State what the figure is and is not. It is a count of sites that linked because of the campaign. It is not a measure of ranking effect, and it does not weigh one domain against another. A second column describing each qualifying domain in words, its subject and audience, tells the reader more than the total.
The classification column depends on telling editorial coverage from purchased placement, and the criteria for that are set out in the guide to what distinguishes earned digital PR links from links purchased as placements.
The method in order: baseline, normalize, attribute, qualify, then count, and publish the rules with the number.
Add your comment to this answer