What broken link building is
Broken link building is the practice of finding a page that links out to a resource which no longer exists, then contacting whoever maintains that page and suggesting your own resource as the replacement. The link on their page returns a 404 (the HTTP status code a server sends when the requested page is not there). You are offering to fix a small defect in their site, and the fix happens to point at you.
The appeal is that the pitch has a reason to exist. You are not asking a stranger for a favor on the strength of nothing. You have found a real fault, you can describe it precisely, and the person on the other end has a mild professional interest in not publishing links that go nowhere. That is a genuinely better starting position than most outreach.
The tactic is safe. Nothing about it violates Google's link spam policies: no money changes hands, no reciprocal obligation is created, and the editor makes a free decision about whether your page is a fair substitute. The risk is not penalization. The risk is that you spend six weeks on it and get two links.
How it works, step by step
The workflow has five distinct stages, and each one loses you prospects.
- Pick a topic where dead resources cluster. Old tool directories, government or university reference pages, industry association lists, and anything last edited before a major platform shut down. Topics that were popular ten years ago and are still linked to are the richest ground.
- Find pages that link out heavily. Search operators are how you do this without a tool budget. Real patterns that work:
inurl:links intitle:resources "your topic""useful links" "your topic" -site:yourdomain.comintitle:"link*" "your topic" site:.edu"recommended reading" OR "further reading" "your topic"inurl:links.html "your topic"
- Crawl those pages for dead outbound links. A crawler in list mode, or a link checker extension for one-off checks. You want the status code of every external link on every candidate page.
- Qualify the dead target. What was on that URL? Use the Wayback Machine to see the original page. If you cannot see what it said, you cannot claim your page replaces it.
- Contact the person who maintains the page — not a generic inbox if you can avoid it. Report the broken link with its exact URL and anchor text, say what used to be there, and offer your page as one option.
What it costs in time and effort
The cost is almost entirely your attention, spread thin over a lot of low-yield checking. Building a prospect list of a few hundred candidate pages is an afternoon. Crawling them is machine time. Everything after that is manual: reading each page, deciding whether the dead link is one you can honestly replace, finding a human being to email, and writing something that does not read like a template.
Where the money goes, structurally, is threefold. First, a crawler or link checker, which is a modest fixed cost you already pay for other reasons. Second, an email finding and verification service, because sending to addresses that bounce damages your sending reputation faster than anything else you can do. Third — and this is the large one — the hours of a person who can judge relevance. That last cost is what makes agency broken link building expensive relative to its output, and why the tactic is quietly abandoned by most teams that try to run it at scale.
Expect the first link four to eight weeks after you start, and expect it to arrive from a prospect you had written off.
When it works and when it does not
It works when you already have the replacement resource. The single biggest determinant of success is whether the page you are offering is genuinely as good as the one that died. If the dead link pointed to a detailed technical reference and you are offering a blog post with a signup form on it, the editor can tell, and you have burned the contact.
It works in fields with long-lived reference pages: academic and library subject guides, professional associations, hobbyist communities, local government, and any sector where somebody built a links page in 2011 and still updates it occasionally.
It does not work in fast-moving commercial niches where nobody maintains link lists any more. It does not work when your replacement is thin. And it does not work if you cannot find a human — a contact form on a university department page is a coin flip at best.
Be honest with yourself about the response rate. Cold outreach of this kind converts in the low single digits by every credible account I have seen, and my own experience is consistent with that. On a list of 300 qualified prospects, a handful of links is a normal result, not a failure.
Common mistakes
- Pitching a replacement that is not a replacement. The most common failure and the one that does lasting damage to your name.
- Emailing about a link the editor does not control. Half the broken links on a page are in the comments, in a sidebar widget, or in a syndicated feed. Check where the link actually lives before you write.
- Leading with the ask. If the first line of your email is about your page rather than their broken link, the reason to reply has evaporated.
- Ignoring the page's own health. A page that has not been touched in eight years, has no traffic and sits four clicks from anything, will not be edited for you and would not be worth much if it were.
- Running it as a volume play. The moment you template the pitch and send a thousand, your reply rate collapses and your domain starts getting filtered. This tactic punishes automation more than most.
- Quitting at week three. The work front-loads and the results back-load. Most people stop precisely when the first replies would have started arriving.
A worked example
Say you publish a detailed guide to reading a commercial property lease. You start with intitle:resources "commercial lease" inurl:links and a few variants, and pull 240 URLs across small business support organizations, local chambers, and two university extension programs.
Crawling those 240 pages returns 1,900 outbound links, of which 63 are dead. You discard 31 immediately — they were tools, calculators, or PDFs you cannot replace with an article. Of the remaining 32, Wayback shows 19 were explanatory guides on lease terms. That is your real prospect list: nineteen.
You find named contacts for eleven of them. You write eleven individual emails, each naming the exact broken URL and its anchor text, describing what used to be at that address, and offering your guide as a possible replacement without insisting on it. Four reply. Two add the link within a month, one says they are rebuilding the page and will consider it, one says no.
Two links from 240 starting URLs. That ratio is normal. Whether it was worth it depends entirely on what those two pages are — and in this example, one of them is a chamber resource page that has been linked to by other people for a decade.
How to measure it
Measure the funnel, not the links. Track candidate pages crawled, dead links found, dead links qualified as genuinely replaceable, contacts identified, emails sent, replies received, and links placed. The ratios between those stages tell you where your process is failing; a single link count tells you nothing you can act on.
Two ratios matter most. Qualified-dead-links per hundred pages crawled tells you whether your prospecting query is any good — if it is under about five, change the operators rather than sending more email. Replies per hundred emails tells you whether your pitch and your contact data are any good; if it is near zero, the problem is almost never the tactic.
Then measure survival. Check the placed links again at 90 days and at a year. On my own profile, the median referring domain stops linking after 1,080 days, and a resource page that gets rebuilt takes your link with it. A link you won in February and never checked again is not evidence of anything.
The verdict
Broken link building deserves its reputation as the honest tactic, and it deserves rather less of its reputation as an efficient one. Every link is earned by a real editorial decision, it will not attract a manual action, and the pages it wins tend to be the durable kind — reference pages accumulate their own links over time, which is worth more than the raw metric on the day you get it.
But the arithmetic is brutal and nobody advertises it. You will examine hundreds of pages to send dozens of emails to win a small number of links, and the work does not get easier with volume because the qualifying step cannot be automated without destroying the thing that makes it work.
My recommendation: run it as a background activity, in batches, on a topic where you already have the strongest resource. Do not build a program around it and do not promise a client a monthly number from it. If you are choosing between this and link reclamation, start with reclamation — the prospects are warmer, the pages are still live, and the evidence for it is far stronger.
Questions
Does broken link building still work in 2026?
Yes, and nothing about the mechanism has changed. What has changed is the supply of maintained link pages, which has been shrinking for a decade, and the volume of automated outreach competing for the same editors' attention. The tactic still converts for people who qualify prospects properly and write individually. It has never converted for people sending templates in bulk.
What response rate should I expect?
Low single digits on reply rate is the honest expectation for cold outreach of this kind, and the link rate is lower still. If you are getting nothing at all from a hundred well-targeted emails, the problem is usually your contact data, your sending reputation, or a replacement page that is obviously not equivalent to what died.
Do I need a tool, or will search operators do?
Operators find the pages; you still need something to check status codes at volume. A desktop crawler in list mode is enough to start, and a link index helps you see which dead pages had links of their own, which is how you find opportunities other people missed. You do not need an outreach platform to run your first batch.
Is it safe to build broken links at scale?
The tactic is safe; the scaling is what causes problems. The links themselves are editorially given and carry no risk. Mass templated outreach damages your sending domain and your professional reputation, and it pushes people toward pitching replacements that do not replace anything, which is where the tactic stops being honest.