What a link gap analysis is
A link gap analysis — also called a link intersect — compares your backlink profile against those of several competitors and returns the referring domains that link to some or all of them and not to you. It is a set operation on link data, nothing more mysterious than that.
The reasoning behind it is sound. A site that has linked to three of your competitors has demonstrated three things at once: it links out at all, it is willing to link to companies in your category, and it publishes the kind of page where such a link fits. That is a far better qualified prospect than a domain pulled from a search operator, because the willingness has already been proven.
It is prospecting, not strategy. The output is a list of opportunities that already exist, which means every one of them was created by somebody else's activity. That has a strategic consequence people rarely state: a program built only on gap analysis will always be following. It is the correct place to start and the wrong place to stop.
How it works, step by step
- Choose competitors by search results, not by revenue. Your business rivals are not necessarily your ranking rivals. Take the sites that outrank you for the queries you care about, even if they are publishers or aggregators rather than direct competitors.
- Use three to five of them. Two produces noise; ten produces a list of generic domains that link to everybody.
- Run the intersect at domain level. Referring domains, not links. Link-level comparison is dominated by sitewide placements — on my own profile the top 100 domains account for 84.3% of all links, and one domain alone sent 219,159 of them from a single sitewide placement.
- Filter for overlap. Domains linking to three or more competitors and none to you are the core list. Domains linking to only one are usually that competitor's own relationships.
- Qualify by relevance first, metrics second. A moderate site that covers your subject beats a high-authority general one.
- Classify how each link was obtained. Open the actual linking page. Was it a guest post, a directory, a news mention, a resource list, a review, a paid placement? The classification is the whole value of the exercise.
- Route each prospect to the tactic that fits. The gap analysis does not win links. It tells you which tactic to run.
What it costs in time and effort
The analysis itself is cheap: a link index subscription and an hour. Every major tool has this feature — Majestic, Ahrefs, Semrush, Moz and others all offer a comparison or intersect view, and they differ mainly in how big and how fresh their index is. Running the same comparison in two tools is worth doing once, because index coverage varies more than the vendors like to admit.
The expensive part is classification. Opening several hundred linking pages and recording how each link came to exist is genuinely tedious and cannot be automated, because the thing you are judging — was this earned, negotiated, or bought — is a judgment. Budget a day for a serious list.
After that, the cost is whatever the routed tactics cost. Gap analysis has no acquisition cost of its own, which is why it looks so efficient in tool marketing and why teams are surprised when the list does not convert itself into links.
When it works and when it does not
It works best in mature, crowded markets where several established competitors have each been building links for years. The overlap in those markets is rich and the patterns are legible.
It works well as a diagnostic after losing ground: if a competitor has overtaken you, the gap list often shows exactly what they did, and how recently.
It does not work well when your competitors got their links in ways you cannot repeat — a national news cycle, an acquisition, a founder with a media profile, or a decade of relationships. It does not work when they bought them, which you will discover during classification and should treat as information rather than instruction.
And it frequently returns nothing useful, which is itself a finding. A null result — no meaningful overlap, or an overlap composed entirely of directories and syndication — usually means one of three things: your competitors are not doing link building either, the market has no natural linking community, or the ranking difference you are trying to explain is not about links at all. Every one of those conclusions is worth having. A null result is not a failed analysis; it is an answer that redirects your effort, and I would rather spend a day learning that links are not the constraint than six months acquiring links that were never the problem.
Common mistakes
- Comparing at link level instead of domain level. One sitewide placement will make a competitor look untouchable when it is a single editorial decision by one person.
- Picking the wrong competitors. Comparing a local firm against a national marketplace produces a list of unwinnable prospects and a demoralized team.
- Skipping classification. A list of domains with no explanation of how each link happened is not actionable. It is a spreadsheet.
- Copying links that should not be copied. Discovering a competitor's paid links and buying the same ones is how a gap analysis becomes a liability.
- Assuming the link caused the ranking. Correlation with the top of the results is not a mechanism, and the strongest pages in most markets rank for reasons that include links without being reduced to them.
- Treating the list as permanent. Rerun it every six months. Links decay on both sides, and the gap you measured last year has changed shape.
A worked example
A specialist B2B software company compares itself against four rivals that consistently outrank it. The intersect returns 340 referring domains that link to at least two competitors and none to the company.
Filtering to three-or-more overlap cuts it to 78. Classification of those 78, one page at a time, produces a breakdown that is more useful than the list: 26 are software directories and comparison sites; 19 are guest posts, most of them written by the same two competitor employees; 14 are industry association and event pages; 9 are genuine editorial mentions in trade press; 6 are podcast episode pages; and 4 are obviously paid placements on sites that sell them openly.
The routing follows directly. The 26 directories are a checklist for an afternoon. The 14 association and event pages are memberships and speaking slots, which fit the company's existing budget. The 9 editorial mentions become a digital PR target list. The 19 guest posts show that competitors have staff who write, which is a resourcing decision rather than a link tactic. The 4 paid placements are noted and ignored.
The analysis produced no links. It produced a plan, and a costed one, in a day.
How to measure it
Measure the analysis on the quality of the plan it produces, not on links, because it does not produce links. Useful outputs: prospects identified, prospects classified as winnable, tactics chosen, and — six months later — links won from that list.
Track gap closure over time. Rerun the same comparison quarterly and watch the count of high-relevance domains linking to competitors and not to you. That number falling is real progress; total referring domains rising is not the same thing, because you can add fifty irrelevant domains and close no gap at all.
Record null results deliberately. If a comparison returns nothing winnable, write down what that implies and revisit it in six months. The most expensive mistake in this area is running the analysis, finding nothing, and quietly running it again next month in the hope of a different answer.
The verdict
Every link building program should start with a gap analysis, and no link building program should be built on one. As prospecting it is unmatched: the prospects are pre-qualified by their own behavior, the classification teaches you how your market actually acquires links, and it costs a day.
As strategy it has a hard ceiling. It can only ever find opportunities that somebody else created first, so a company that works exclusively from gap lists is permanently second. The links that separate the leader from everyone else are the ones nobody could have found by comparison, because they did not exist until that company earned them.
Run it, classify it properly, route the prospects to the tactics that fit, and rerun it twice a year. Then spend the rest of your time on the things a competitor's profile cannot tell you about.
Questions
How many competitors should I compare against?
Three to five, chosen by who outranks you rather than by who you compete with commercially. Fewer than three and the overlap is coincidence; more than five and you mostly surface general-purpose domains that link to everyone. If your ranking rivals are publishers rather than businesses, include them anyway — they are the sites you are actually competing with.
Which tool is best for a link gap analysis?
All the major link indexes offer it — Majestic, Ahrefs, Semrush and Moz among them — and the meaningful difference is index size and freshness rather than the feature itself. Coverage varies more than vendors admit, so if the decision matters, run the same comparison in two tools and compare the domain lists rather than the totals.
What does it mean if the analysis finds nothing?
Usually one of three things: your competitors are not building links either, your market has no community of sites that link out, or the ranking gap you are investigating is not caused by links. All three are useful conclusions. A null result redirects your budget toward whatever the real constraint is, which is worth more than a list of prospects you cannot win.
Should I try to get every link my competitors have?
No. Some were paid, some came from relationships you cannot replicate, and some are worthless. Parity is also not the goal — matching a competitor's profile exactly leaves you with no advantage. Use the list to understand how links are acquired in your market, take the winnable ones, and put the remaining effort into links they cannot copy.