Judge a competitor's link profile, not their link total

Open a competitor's link report and the biggest number is at the top. It is also the least informative thing on the page. A total counts every link, including the four hundred sitting in a sitewide footer that appears on every page of one partner site, and each of those counts as a separate vote from a source that had already voted.

Divide the total by the number of distinct domains linking and the picture usually changes shape. A high result means repetition; the real endorsement count is far below the headline, and so is the gap you would need to close.

What the total hides

Sitewide elements are the main source of inflation. A footer link, a template sidebar, a partner badge, a franchise listing repeated across a network: each one produces a link for every page it appears on. None of them adds a new site willing to vouch for the domain.

Repetition is not worthless, but it does not scale the way the raw number implies. Reading the total as a measure of how much link building stands between you and a rival will make a beatable competitor look untouchable.

There is a second thing the total cannot do: tell you whether any of it is relevant. A profile assembled from directories and scraper sites produces a large count and very little ranking help. Only the domain-level view shows that.

Authority scores are a tier, not a target

Most link reports also carry a 0-100 authority score, reported as the rank field in a backlink overview, and that is the figure people most often turn into a goal. It makes a poor one. The score is a vendor's own invention, computed from the link graph. Search engines do not read it. It tracks rankings because both of them track a strong link profile. Push the score up without adding links and the rankings stay exactly where they were.

Read it as a weight class. It answers one useful question — is this competitor in a different league from me, or roughly mine? — and then its work is done. Campaigns aimed at the score itself buy links that lift the number and leave the rankings where they were.

Relevance is the stronger filter and it is the one the score cannot express. A link from a modest site inside your own topic is worth more than a high-scoring link from a site with nothing to do with you, which is why the score belongs after the relevance judgment, never before it.

Three readings that beat the total

Distinct referring domains. The count of separate sites linking. This is the closest thing to a real measure of how much independent endorsement a profile carries, and it is the number to compare against your own rather than the total.

Spam in the neighborhood. A spam signal on the profile says what company the domain keeps. An established brand carrying a moderate score is normal: two decades of directory listings and scraper mentions accumulate. The point of reading it is not to judge the competitor, it is to clean your target list. Filter the low-quality domains out before you start treating what remains as outreach candidates.

Dead links. Links pointing at URLs on the competitor's domain that no longer resolve are the most reachable opportunity in the whole report. Each one is a publisher that judged the subject worth linking to, and whose link now arrives at nothing. Find the dead page, look up who is still linking to it, and decide whether you have something that would do the same job. The approach works because you are reporting a problem and offering a fix, not asking for a favor.

Run that last check against your own domain before you run it against a rival's. Sites that have been rebuilt even once routinely find a slice of their earned links landing on URLs that 404. That is equity already paid for, recoverable with redirects rather than outreach.

Do it with an agent

Pull both profiles through the backlink checker and let the Doverity MCP do the comparison, so you set the quality filters instead of inheriting defaults.

Using the Doverity MCP, set [mydomain.com] and [competitor.com] next
to each other and read what their links say.

1. Call get_backlinks_overview for each domain. Put backlinks,
   referring domains, and rank for the two of them into one table,
   then divide backlinks by referring domains for each and say what
   that ratio implies about the method behind their links.
2. For the competitor, call get_backlinks_profile in one_per_domain
   mode with hideSpam enabled. Read that result as a shortlist rather
   than the complete profile, and tell me how many distinct domains it
   holds.
3. Compare that shortlist against my own referring domains. List every
   domain that links to them but has never linked to me, and for each
   one name the page of theirs that earned the link.
4. Call get_backlinks_profile again, keeping only rows in a broken
   state. For each one report the site linking in and the URL on their
   site which resolves to nothing.
5. Repeat step 4 against [mydomain.com] and add up the links I have
   already earned that now land on a URL which resolves to nothing.

On the step 3 list, weigh topical relevance to [what I sell] ahead of
any authority figure, and tell me which sources you would skip and why.

Step 5 outranks every outreach step on this page. Recovering links you already earned costs a redirect; earning a new one costs a campaign.

Competitor backlink analysis FAQ

How do I run a competitor backlink analysis?

Start with the overview for the headline figures. Then pull link rows one per domain, so the table shows separate sites instead of the same site counted over and over. Keep the relevant, low-spam ones, set them beside your own referring domains, and whatever they hold that you lack is the gap worth working on.

What counts as a good spam score?

No threshold suits every case, and the signal comes from a third party rather than from anything search engines publish. Read it side by side with similar domains. An established brand can sit in the low twenties without it meaning much, because two decades of directory listings add up; a site a year old at the same level has likely collected links it never earned. Treat it as a way to trim a prospect list.

Should I try to match every link my competitor has?

No, and the headline count is mostly repetition anyway. Directory entries, scraper copies, and syndicated listings make up a large part of any profile, none of it counts for much, and most of it would not link to you at any price. A handful of domains also link over and over from footers, sidebars, page templates, or a franchise network's location pages, which is how a site reaches hundreds of thousands of links without anything like that many endorsements. Divide the total by the referring-domain count: a result well above one says the headline is measuring how many pages those sites have, not how many of them vouch for the domain. What you could realistically reach is a slice of the total, and the slice that matters — relevant sites with a habit of linking to good material — is smaller still.

Are a competitor's broken links worth pursuing?

Few things in the report are as promising, since the site linking in has already shown it will cover this subject and its link is presently arriving at nothing. It lands when the replacement you offer is genuinely better than the page that died, and it falls flat when all you want is a swap.