Broken link building, step by step: find, fix, and pitch

Broken link building is the one outreach pitch where you arrive with a fix instead of a favor. An editor has a dead link inside an article they still maintain, and you may have a page that can take its place. Most people waste that opening. They send a note that mentions the 404 in one line, then spend the rest of the email pitching their homepage. The editor did not ask for your homepage. They have a defect on a page that still gets traffic, and you can remove it.

The reason this works is mechanical. A link is worth something to an editor only because of what it does for the reader. When the target disappears, the sentence that cited it loses its support and the reader hits a dead end. Fixing that costs the editor one edit. The rest of this guide covers how to do that reliably: find dead links on pages that matter, confirm they are really dead, work out what each link was for, and only then build or pick a replacement to pitch.

Broken link building works at all only under three conditions

A dead link becomes an opening when three things are true at once.

1. The page still matters to its owner. It ranks, gets organic traffic, or is linked from elsewhere. An abandoned page nobody visits is not worth an email.
2. The dead link did a specific job in a specific sentence. It supported a number, defined a term, or pointed to a tool. If you cannot say what the sentence needed, you cannot fix it.
3. You have something that does the same job. Not something adjacent, but something that fills the same gap well enough that the editor does not have to rewrite the sentence.

If one of these fails, skip the page. Pitching a replacement for a link nobody needs is just a cold email with extra steps. This is also where the volume trap starts. Once you think in terms of how many dead links you can find, you stop checking the three conditions. The number of pitches you send matters less than how many of them pass all three.

Step one: find dead links on pages that still matter

Start with a list of source pages, not the whole web

Crawling the open web for 404s produces thousands of results, and most of them are worthless: forum threads, scraped pages, and sites that were abandoned years ago. Start narrower. Build a list of pages that are editorial, on your topic, still indexed, and still getting organic traffic.

The best candidates are resource pages (“useful tools,” “further reading,” “links”), industry roundups, and long guides that are several years old. They are old enough to have collected dead links and recent enough that someone still reads them. To check traffic, use what you already have. Search Console covers your own pages, and an SEO tool’s organic traffic estimate covers everyone else’s. Those estimates are rough, but you only need to know whether a page is alive, not its exact visit count.

Crawl the list and filter for broken outlinks

Once you have source URLs, run them through a crawler in list mode. Screaming Frog’s SEO Spider handles this well. Paste the URLs, crawl, and then review the outlinks from pages that return client errors (4xx) or server errors (5xx). The report tells you which page holds the link, what the anchor text says, and what status the target returned. That is everything you need for one row in your tracking sheet.

Site audit features in Ahrefs and Semrush can surface broken outbound links across a whole domain, which is faster when you already know which sites are in scope. For spot checks on a handful of pages, a browser extension such as Check My Links is enough.

Search the pattern, not just the tool

Crawlers miss resource pages that nothing else links to, and they cannot judge context. Search operators fill that gap. Queries such as intitle:"resources" "your topic" or "your topic" "useful links" surface pages that exist mainly to link out, and those are where dead links accumulate. Check the results by hand. A page that has been dead for five years and still ranks is a better target than a brand-new page with one broken link.

The inbound variant

There is a second version of this method, and it is harder to do well. You look for a competitor’s or a reference site’s dead page that still has other sites linking to it. An SEO tool’s broken backlinks report will show these. You then build a page that covers the same topic and ask the linking sites to point to yours instead.

This version works because the links already exist, so you are asking for an update rather than a new placement. The downside is that the linking sites may have already removed the link, and you are contacting people who never chose you as a resource in the first place. Treat it as a second channel, not the main one.

Step two: verify the link is really dead

Most false positives in a crawl come from a small set of patterns. Verify each one in a real browser before you spend a sentence on it. The table covers the common cases.

Soft 404s deserve particular attention. A page that returns 200 but shows “page not found” is a dead link to a reader, even though the status code says otherwise. Crawlers handle these inconsistently, so do not rely on a status code alone. Open the link yourself.

The redirect-to-homepage case needs care too. Technically the editor’s link still resolves, so some editors will say “it works.” But the reader lands somewhere unrelated, and the sentence has no support. Build the pitch around what the reader gets, not around the status code.

Step three: work out what the link was for

This is the step most people skip, and it decides whether your pitch succeeds. Editors replace citations like for like. A sentence that cites a statistic needs another source with that statistic. A sentence that defines a term needs a definition. A sentence that points to a tool needs a tool. An opinion post, however good, does not replace a data citation, and the editor will see the mismatch in seconds.

Sort each dead link into one of five jobs:

• Data source: the link supports a number or a finding.
• Explainer: the link defines a concept or explains a process.
• Tool or calculator: the reader was meant to use something.
• Example or case: the link shows a real instance of the claim.
• Template or download: the reader was meant to take something away.

Here is a hypothetical. An article written several years ago says most marketing teams track their link metrics in a spreadsheet, and it cites a survey report that now returns a 404. The job is “data source.” A 2026 blog post with your opinion on link tracking does not do that job. A current, sourced dataset that answers the same question does. Your replacement has to be the same kind of thing as the original, or the pitch fails before the editor reads past the first line.

The classification also tells you when to walk away. If the dead link was a data citation and you have no data, you have no pitch. Do not invent a statistic to fill the gap. The editor will check it, and you will have spent your credibility on the first email.

Step four: build or pick the replacement

You have three options. Only the first is usually fast.

1. Use an existing page. Search your own site for the topic. If a page already does the dead link’s job, confirm it is current and indexed. Then it is ready to pitch.
2. Refresh an old page. If you have an outdated version of the right content, update it, keep the URL, and add an accurate “last updated” date. Check the page’s existing backlinks first. An old URL may already be linked from elsewhere, and that history is worth keeping.
3. Make a new asset. Build something new only when the gap is real and you are willing to maintain it. A new asset that answers a question editors keep running into is worth the effort. One built only to earn a link is not.

Whichever you choose, run this test: if you were the editor, with this sentence in front of you, would you cite this page? If the answer is “maybe,” it is not ready. A replacement weaker than the original fails on the first reply, and it damages your standing with that editor for the next one. The logic matches the point in our piece on relevance over volume: a placement has to fit its context to do anything.

Two checks come before you pitch. First, confirm the replacement is indexed. A page that search engines have not crawled cannot pass value through a link, which is the same logic as our earlier point about placements that never get indexed. Second, confirm the page you are pitching to is still maintained. A dead link on a site that has not published in three years is a dead end for the pitch as well, because nobody is reading the inbox.

The outreach that gets replies

Find the right person

Start with the byline. The author may still work at the company, or the article may credit an editor. If neither helps, look for the site’s contact page or its editorial guidelines, which sometimes say how to suggest corrections. Some sites run a formal link-correction process. If one exists, use it, because it is the channel the editor expects.

Avoid generic addresses such as info@ unless nothing else exists. They have lower odds because they often reach someone who does not maintain the content.

Write the email

The email has one job: make the edit take thirty seconds. Here is a template you can adapt.

Subject: Broken link in your [topic] guide

Hi [first name],

The section on [specific subsection] links to [exact anchor text],
which now returns a 404.

I put together a [replacement type] covering the same [specific claim]
with current figures: [URL]

If it fits, the swap is replacing the old URL in that sentence. Happy
to suggest different wording if you would rather link something else.

[Your name]

Each line does work. The subject names the problem, so the email reads as a fix rather than a pitch. The first sentence names the exact section and the exact broken anchor. That proves you read the article and makes the problem concrete. The replacement line states what the new page covers, in terms of the claim the editor already made, which is the like-for-like test from step three. The last paragraph states the exact swap. That line is what gets replies, because it removes the decision the editor would otherwise have to make.

Notice what is missing. There is no compliment paragraph, no description of your company, and no request to “pick your brain.” Editors recognize a flattery opener, and it makes the email look like a template. If you have a real reason you read the page, one clause is enough.

Send one follow-up after about a week. Keep it short and restate the swap. Then stop. A third message signals a campaign, and an editor who ignored two specific emails is not going to change their mind on the third.

What to expect: the conversion math

Here is the honest version. Broken link building is a low-volume method that depends on judgment, and most of your prospects will never reply. Editors are busy. Many pages list contact addresses that do not reach anyone who maintains the content. Some editors reply but do not want to change a link. The people who say yes are a small share of the people you verified, and that is normal.

The figure below shows how the funnel narrows. The numbers are illustrative, not a benchmark, but the shape is typical. Each stage loses people, and each loss is a place to look for improvement.

The stages matter more than the final number, because each one points to a different fix. Low contact rates mean your address-finding is weak, so check bylines and contact pages before you blame the email. Low reply rates usually mean the subject line or first sentence is not specific enough to earn the open. Replies that never become a swap almost always mean the replacement is not as good as the original, or the editor does not want to change that link at all. Each leak tells you something different, so do not respond to all of them by sending more email.

Do not borrow someone else’s target rate either. The result depends on topic, on how stale your source pages are, and on how strong your replacements are. Run a small batch, measure each stage, and then scale the parts of the funnel that work.

Where broken link building goes wrong

The failures are predictable, and most of them happen before anyone replies.

• Mass-sending one template. Teams swap the URL and send the same email to two hundred pages. Editors recognize a template by the second sentence, and the placements that would have worked are lost in the noise. Personalize the first line for every single send, or send fewer emails.
• Weak replacements. A replacement that is shorter, older, or less specific than the dead original gets a polite no. Worse, it trains the editor to ignore the next email from you.
• Dead or abandoned sites. A page that is dead on a site nobody maintains is not a link you can win. Check the site’s recent activity before you invest in the pitch.
• Right page, wrong placement. The editor agrees, but the link lands in a footer, an author bio, or a sidebar widget. Those placements carry less weight than in-content editorial links. Keep the swap inside the body text, and once it is live, check its rel attribute. Our piece on dofollow and nofollow links covers why that attribute decides whether the link passes value.
• Unindexed targets. A replacement that search engines have not crawled gives the editor a link that passes nothing. Confirm indexing before you pitch, not after.

The short version

Broken link building is a sequence of small, checkable decisions. Find dead links on pages that still matter. Verify each one in a browser and in the Wayback Machine. Work out what the link was doing for the reader. Offer a replacement that does the same job, and only if it is actually better or equal to what came before. Then send one email that makes the swap easy, follow up once, and move on.

The part that is hard to do by hand across dozens of pages is the matching: finding placements that genuinely fit the content, making sure every one gets crawled and indexed, and checking that each link is dofollow and still live. That is the work OptiLink automates. It places relevant links from millions of vetted blogs, hand-places them where the link fits the content, checks every one for dofollow, and pushes each to get indexed. Relevance over volume, on autopilot. If you want to see how that works for your site, start with how OptiLink works.

Keep reading

Playbook

Anchor text optimization without tripping a filter

A hundred random links can do less than five relevant ones. Here's what makes a backlink pass real ranking value in 2026 — and what makes it worthless or worse.

Sofia Meyer September 24, 20262 min read

Playbook

Competitor backlink analysis that finds real opportunities

A hundred random links can do less than five relevant ones. Here's what makes a backlink pass real ranking value in 2026 — and what makes it worthless or worse.

Sofia Meyer September 26, 20262 min read

Playbook

Digital PR: the most durable link building there is

A hundred random links can do less than five relevant ones. Here's what makes a backlink pass real ranking value in 2026 — and what makes it worthless or worse.

Sofia Meyer October 13, 20262 min read

Explainer

Dofollow vs nofollow in 2026: what still passes value

The link attributes changed, Google's treatment changed, and a lot of advice is out of date. Here's what dofollow, nofollow, sponsored and ugc actually do now.

The OptiLink Team October 10, 20262 min read