Link building outreach that people actually answer

Somebody's software once read Jon Henshaw's privacy policy and decided it needed a link.

When I interviewed Jon in 2019, I brought up a MozCon talk he'd given years earlier on the psychology of relationships in link building. I'd been using it as training material at Kwasi, to remind the team that outreach isn't spreadsheet work and there are people behind the keyboard. Jon said that was several years ago, then told me about his inbox that week:

"There's nothing more that I hate to get than some stupid automated email. I mean, I got one like really bad the other day where their software had discovered my privacy policy page and thought that would be a good place to put their link... obviously no human actually wrote this email because nobody would be that dumb."

Later he did an impression of the genre: "hey, really, read your article... And I really feel like such and such would be a great fit. Just let me know." Then the aside that still makes me laugh: "they did not read my privacy policy." The story is in the first part of his written interview.

Jon started Raven Tools, and I used it for years to track backlinks and outreach. He knew exactly what was on the other end of that email, and it wasn't a person.

Why the bot fails before anyone reads it

Jon's bigger problem was that this kind of outreach "can ultimately or sometimes harm the brand they're representing." The email tells the reader one thing very clearly: nobody looked. Whatever brand is in the signature now owns that.

The way I define digital empathy is the ability to understand and share the feelings of the person on the other side of the screen. Automated outreach skips that step entirely, and Jon's test for it is almost embarrassingly simple. Put yourself in their seat: "if I were somebody who was running this blog... how would I feel if I received this type of thing?" His answer to what he'd respond well to started with "one, not that."

Oli Gardner got the same treatment from the other direction. Unbounce is in Vancouver, but people kept assuming it was in the Valley and emailing to catch up for drinks, from over a thousand miles away. His advice to them: "do your research. We're not there." Pretty basic understand the context fail, right there in the first line.

When I showed Jon the 16 principles and asked him to pick a favourite, nothing quite summed it up for him, so we worked out a seventeenth. He settled on "respectful, efficient communication", and he meant efficient for the reader: knowing enough about someone "because that way I'm not wasting their time."

What good outreach looked like to our guests

Jon's best example was a client in a corner of finance that, as he put it, is "not thought highly of". The sites worth being on were personal finance blogs. Pitch them a link to that client in one of "these stupid emails" and it "would be a 100% rejection rate."

So he flipped it. "Instead of going after the thing I wanted, I went after the thing they wanted." Most of those bloggers wanted to make some money from their site, so he offered to sponsor it properly: an ad or a newsletter sponsorship, committed for three months. The response rate, he said, was huge.

That did two jobs. First, the ads told him whether the site was worth anything: "does anybody even go there and will they even click on these links?" Second, months of paying invoices and setting up new placements built what he called rapport. Only then did he ask whether there might be an editorial opportunity, something he'd write for them for free. "They would say yes or no." It was slower, and it worked far better. He goes through the whole method in part two of the written interview.

A note if you try this: the ad is an ad, so mark it as one (rel="sponsored" on the link), and keep the editorial conversation separate. The point is the relationship, not a paid link in disguise.

Rand Fishkin answered his email because he remembered being ignored. Early in his career he'd write to influential people and hear nothing back. "I hate it... And so I want to make sure that I am not that person." If you send him "a well-formed, reasonable request that looks like a real conversation," he'll try to reply. It's in our Rand interview too, and it's why I think showing you care works both ways.
Paddy Moogan sent his outreach as himself, and the relationships outlived the links: "I've even had freelance work as a result of link building for clients, which is really, really weird."
Jon again, on why he bothered: "I may not end up getting a link or a mention, but... this is going to sound really corny, but I may end up with a really good friendship."

I said something similar on Jason Acidre's episode: outreach is "a bit like dating... You're not going to get married on the first interaction."

The bots got better at sounding like people

In 2017, Mike King walked me through a talk he'd given on using machine learning to scale prospecting. Train a model on the sites you've approved and rejected, then "whatever new list of prospects you have, you run it through this model, and then boom, you have your list of prospects instantly." He'd also shown how to train a chatbot to answer the standard objections, which is where you can't automate empathy comes from. Pretty clever for 2017. It's a standard feature now.

The 2026 version of Jon's email has read the privacy policy. It's probably complimented it. The personalisation token became a paragraph, and the paragraph sounds like a person until the fourth email that week sounds exactly the same.

Cision surveyed 1,899 journalists in early 2026 for its State of the Media report: 53% oppose AI generated pitches, and 72% say fewer than a quarter of the pitches they get are relevant. Muck Rack's State of Journalism 2026 found 88% immediately disregard pitches that miss their beat. Those are journalists, not bloggers, but a blogger has the same inbox and no editor to share it with.

Nicole Franco wrote in Search Engine Land last month: journalists "are drowning in emails that all sound like they were written by the same robot", and blasting more of them means "you're not scaling outreach, you're scaling deletion."

Jon saw that coming in 2019. The mass email can't work, he said, "especially when people... are getting the same message from different places as well." The tools have made that message cheaper to send, and a real one easier to spot.

Before you hit send

This is the list I'd run any outreach email past, whether a person or a tool wrote it.

Can you name the person? Not "Hi there", not the site name. If you can't find a name, you haven't looked hard enough to be emailing.
Did a human look at the site? A model can shortlist. Someone still needs to read three recent posts before anything goes out.
Run Jon's test. If you ran their site, how would you feel getting this email? If the honest answer is "not great", don't send it.
What's in it for them, in their terms? Money, a story, data, a better resource for their readers. "A great fit" is not a reason.
Does your page belong on theirs? If you had to squint to find a spot for the link, so will they.
Is it from you? Real name, real address you'll answer from, like Paddy.
Would Rand reply? Well formed, reasonable, and reads like the start of a real conversation.
Is it short? That's the "efficient" half of Jon's three words.
Is money clearly money? If you're paying, it's sponsorship and the link is marked as sponsored.
Are you fine with silence? One polite follow up, then leave it. A second is a nudge. A fifth is a sequence.
Would you be happy to see it quoted on a podcast? Jon's privacy policy email ended up on one, and not in a good way.

Somewhere there's an agency whose software pitched Jon Henshaw's privacy policy, and they have no idea it happened, because the bot never mentioned it. Jon, on the other hand, told a podcast.

If you want the human version, start with talk to the people behind the screen. Jon still writes about search and the web at Coywolf, and he still, as far as I know, doesn't want a link on his privacy policy.