The skyscraper technique: does it still work?

The skyscraper technique is a link building method in three parts: find a page in your subject that many sites already link to, publish a clearly better page on the same subject, and ask the sites that link to the original to link to yours. Brian Dean of Backlinko named it and published it in 2013, with a case study that reported 17 links from 160 emails. It became one of the most copied methods in SEO, and that popularity is part of why it is harder now. The evidence on how well it works today is thin. What exists is one founder’s case study, an opinion poll, one industry survey of usage and general data on outreach replies. Read together they suggest a method that still has a use in a narrow set of cases and disappoints as a default.

Where it came from and what was claimed

Dean’s article on Backlinko carries a publication date of 18 April 2013 and has been updated since, most recently in 2024 on the version read for this lesson. It sets out three steps in these words: find link-worthy content, make something even better, and reach out to the right people. The name comes from his comparison with buildings. Nobody admires a tall building and then wonders how big the eighth tallest one is, so the aim is to build the tallest page on a subject. The worked example was his list of Google’s 200 ranking factors, which took an existing type of list and made it longer and more thoroughly documented than the versions already ranking.

The numbers in that case study are the ones repeated everywhere since. Dean reports that he sent 160 outreach emails and gained 17 links, which he gives as an 11 percent success rate, and that organic search traffic to his whole site rose 110 percent in 14 days. Those figures deserve to be read for what they are. They come from a single campaign, reported by the person who ran it and who was at the same moment promoting the method. The subject was SEO, and the people being emailed were SEO bloggers, an audience unusually willing to link and to edit old posts. The year was 2013, before most site owners had received hundreds of emails built on the same template. None of that makes the figures untrue. It means they describe a best case in favourable conditions, and they cannot be used as an expected rate.

The logic underneath is sound enough to explain the method’s appeal. Prospecting is usually the slow part of link building, and here the list comes ready-made: every site that links to the original page has already shown that it links to content on this subject. The method also removes the guesswork about demand, since a page with many links is proof that the topic attracts them. Both ideas survive in ordinary practice, and a link gap analysis uses the same reasoning when it looks for sites that link to competitors and not to you.

How it is done, step by step

The published method and later guides, including the one Ahrefs published in 2023, agree on the sequence. The steps below follow that sequence, with the qualification work that later guides added to Dean’s original three.

1. Find a page on your subject that has links from many referring domains, using the “best by links” report in a backlink tool.
2. Read why it was linked. Open a sample of the linking pages and note what each one cites: a statistic, a definition, a tool, the whole guide.
3. Decide what a better page would add. Newer data, a correction, a missing section, original research or a working tool all count. Extra length alone does not.
4. Build the page, and make the improvement visible near the top.
5. Export the pages linking to the original and remove dead pages, forums, scrapers and sites you would not want a link from.
6. Group the remaining prospects by what they cited, so each group gets a pitch about that specific thing.
7. Find the right person at each site and send a short, specific email, with one follow-up.
8. Record replies and links, and note which reasons for linking produced them.

The second and sixth steps are where most campaigns are won or lost. A site that linked to the original for one statistic has no interest in hearing that your guide is more complete. It may care that the statistic is five years old and that you hold the current figure. Sorting prospects by their reason for linking is covered in link prospecting, the wording of the pitch in link outreach emails, and getting the message to a person who can edit the page in finding contacts and verifying emails. The page itself is a linkable asset, and the tactic page on linkable assets covers what to build and what it costs.

What the published evidence says

A search for independent data on skyscraper success rates turns up very little. No study was found that takes a sample of skyscraper campaigns and reports how many prospects linked. What follows is everything of substance that could be opened and read, and its limits are stated with it.

Source	What it measured	Figure	Limits
Backlinko case study, 2013	One campaign by the method’s author	17 links from 160 emails, 11%	Self-reported, one niche, before the method spread
Backlinko and Pitchbox, 2019	12 million outreach emails of all kinds	8.5% received a response	Replies, not links, and not specific to skyscraper
Ahrefs poll, 2023	1,242 votes on Twitter and LinkedIn	61% said it still works	Opinion, self-selected, no results data
Aira survey, cited by Ahrefs in 2023	Which tactics SEOs use	18% use it, 20th on the list	Usage, not success, and read second-hand

The outreach study is the most useful of the four, even though it is not about this method. Backlinko and Pitchbox analysed 12 million outreach emails and published the results in April 2019. Only 8.5 percent received a response of any kind. A response includes refusals, requests for payment and automatic replies, so the share that becomes a link is smaller again. Set that beside the 11 percent link rate in the 2013 case study and the gap is plain: the original campaign won links at a higher rate than the average outreach email, six years later, won replies. The same study found that one follow-up raised replies by 65.8 percent and that personalised message bodies did 32.7 percent better, which supports the segmenting described above.

The Ahrefs article, by Si Quan Ong, is a practitioner’s account and says so. It lists the reasons campaigns fail: the same template sent to everyone, prospects not sorted by why they linked, too few prospects, the belief that bigger means better, and the fact that a known brand wins swaps an unknown one does not. The poll in that article shows that most respondents believe the method works, and the survey figure it cites from Aira’s State of Link Building report shows that fewer than one in five actually use it. That report was not opened for this lesson, so the figure is given as Ahrefs reported it. Belief running well ahead of use is a reasonable summary of where the method stands.

Practitioner guides commonly put the conversion at 1 to 5 percent of prospects. That range has no dated study behind it and should be treated as low confidence. For comparison, our linkable assets page cites Hunter’s outreach guide, which puts prospect-to-link conversion for standard link outreach at about 1.3 percent. If a skyscraper campaign behaves like ordinary outreach, a list of 200 qualified prospects yields a handful of links, and a plan that needs the 2013 rate to pay for the content will lose money. The link building statistics page collects the wider outreach figures.

Why it is harder than it was

Three things have changed since 2013, and each follows from the method’s own success. The pitch became a template. Dean published his email, thousands of people copied it, and site owners learned to recognise the opening line. An email that says the sender came across the recipient’s page and has created a similar article is now filed with the rest. The second change is on the content side. When several teams apply the method to the same topic, each one outbuilds the last, and the result is a row of very long guides that say the same things. Being longer than the page above you stopped being a reason to link once every page was long.

The third change is the weakness that was always in the request. The site owner has a finished article with a working link. Changing it takes effort, and the benefit goes to a stranger. Compare broken link building, where the existing link is dead and the email reports a fault the owner would want fixed. Skyscraper outreach offers no fault, only an opinion that another page is better, and the person being asked has no means and little reason to check. As the Ahrefs article puts it, a recognised brand makes that opinion easier to accept, which means the method favours sites that need it least. Many replies in commercial subjects also come back with a price, and the lesson on what to do when a site asks for payment covers that decision.

When it is still worth trying

The method earns its place where the linker has a real reason to reopen the page. The clearest case is an original that has gone stale: a statistics page citing figures from years ago, a guide to a regulation that has since changed, a tool that no longer loads. There the email is closer to a correction than a pitch, and it borrows the strength of broken link building. The second case is where your page holds something the original cannot match, such as first-hand data, a calculator or a current dataset. A page of that kind also earns links without outreach, which the lesson on data-led PR campaigns explains, so the skyscraper emails become one channel for an asset that has others.

It is a poor choice where the original is good, recent and published by a brand people know, where the linking pages are old posts on sites nobody maintains, or where your improvement is more words and a nicer layout. It is also a poor choice as the centre of a programme. The honest way to budget it is to price the page as an asset you wanted anyway, treat the skyscraper list as one of several prospect sources, and send the first fifty emails as a test before committing to the rest. If those fifty produce nothing, the list or the reason is wrong, and more volume will not fix either.