Competitor backlink analysis, and what to do with the list

Backlink gap analysis produces a very satisfying spreadsheet and, for most teams, very little. The reason is that the export mixes links you could get with links you never will, and nobody does the sorting.

Here is the version that ends with a list you can act on.

Getting the data

You need an index. Ahrefs from $129 a month has the best one. Semrush from $139 is close. Both charge extra per user above the first, which matters for teams.

Run a backlink gap: domains linking to two or more competitors but not to you. The two competitor minimum matters, because a link both of them have is a link the category gives out, not a one off relationship.

Export it. Then do the part everyone skips.

Sort into four piles

Pile 1: repeatable. Directories, category listings, "best X" articles, resource pages, podcast appearances, tool roundups.

These exist because somebody put companies on a list. Lists take more entries. This pile is where almost all of your realistic wins are, and it is usually 10 to 20% of the export.

Pile 2: earned by something specific. A link to their original research, their free tool, their data study.

You cannot get these links. You can get equivalent ones by making an equivalent thing, which is a different project with a different budget.

Pile 3: relationship. Their investor, their customer, their founder's old colleague, a partner integration.

Not available. Move on without resentment.

Pile 4: noise. Scrapers, expired domains, foreign language spam, aggregators. In most exports this is the largest pile and it is worth nothing.

Only pile 1 is a work list. If pile 1 has forty entries, you have a good quarter's work. If it has four, link building is not your constraint and you should go and fix something else.

What to look at beyond the count

Which page earned the links. Sort their links by target URL. If one page has most of them, open it and understand why. Usually it is a tool, a data study, a template or a genuinely useful resource, and that is the shape of thing that earns links in your category.

Anchor text. What words people use when linking to them. This is how the web describes your competitor in its own words, and it is a free read on their positioning.

Velocity. Are they gaining links steadily, or was there one spike two years ago? A spike is a campaign. Steady growth is an operation.

Where the links are from. Industry publications, local press, partner sites, universities. The mix tells you which door is open in your category.

The pile 1 workflow

For each repeatable link, one of three actions.

Directory or listing. Submit. Most are free, some charge, a few are worthless. Prioritise by whether the listing page itself ranks for anything.

Listicle or roundup. Email the author. Short, no flattery, a specific reason your inclusion helps their reader, and an offer of any information they need. Success rate is low and the ones that work compound.

Resource page. Same, with a clearer case, because resource pages are maintained by someone who wants them complete.

Work through it in a batch. This is administrative work that benefits from being done in one sitting rather than spread over months.

The second reason this matters now

Backlinks have always been a ranking factor. They now have a second effect worth naming.

When an AI assistant answers a buying question, it reads a small number of pages, and those pages skew heavily towards lists, directories and comparison articles. Being listed on them puts you in AI answers as well as in search results.

That reframes pile 1 usefully. A directory listing is not only a link. It is a place where a model might read your name while assembling a shortlist. That makes the unglamorous submission work worth more than its link value alone suggests, and it is why we would put pile 1 ahead of almost any other link building activity right now.

To find which of those sources matter for your category specifically, ask your buyer questions in ChatGPT, Claude, Gemini and Perplexity and record every source cited. Cross reference that list with your pile 1. Anything on both should be done this month.

What to skip

Buying links. Not a recommendation and not a moral point, simply that the risk and the cost are both badly priced compared to pile 1 work.

Chasing their biggest link. The national publication that covered them is not going to cover you because you emailed. Spend the time on ten achievable ones.

Measuring success in link count. Count referring domains from pile 1 sources, and count whether your AI answer share moved. Total backlinks includes the noise pile and goes up on its own.