You probably wrote off guest blogging around 2014. Matt Cutts declared it dead, the SEO crowd moved on, and the tactic became something spammy outreach agencies still emailed you about. Fair enough at the time.
But something has shifted. The rise of ChatGPT, Perplexity, Google's AI Overviews, and Claude has quietly rewired how people find businesses. And guest posts, of all things, are turning out to be one of the more effective ways to show up when an AI answers a question about your industry.
How AI Search Actually Chooses What to Mention
When someone asks ChatGPT for "a good boutique hotel in the Cotswolds for a weekend away", the model does not run a Google search and pick the top result. It draws on training data, real-time retrieval, and a weighted sense of which sources it trusts on a given topic.
That trust is built through repetition and context. If your hotel appears on ten different reputable websites, each time being discussed in a substantive way, the model starts to associate you with that category. Not because someone told it to. Because the pattern is there in the data.
This is very different to old-school SEO, where a backlink was a vote and volume was the game. AI systems care less about the link itself and more about the surrounding text. What are you being described as? What ideas are you associated with? Who else appears alongside you?
Why Guest Posts Fit This Model So Well
A guest post is, by definition, a substantial piece of writing on someone else's site, with your name attached, on a topic you have a view on. That is almost the ideal input for an AI system trying to work out who is an authority on what.
Compare it to the alternatives. A directory listing is a name and a URL, with no context. A press release is thin and formulaic. A backlink from a footer sponsorship tells a model nothing about what you actually do. A restaurant owner writing a proper piece on seasonal British produce for a food publication, however, tells the model: this person cooks this kind of food, is respected enough to be published here, and has a specific view on their craft.
The other thing worth noticing is that AI systems disproportionately pull from a smaller pool of trusted publications and industry sites than Google's index. Getting cited on ten well-chosen sites in your sector will do more than a hundred generic backlinks ever did.
What Actually Works Now
The mechanics of good guest posting have not really changed. The difference is what you are optimising for.
Write for sites your customers actually read. Not just sites with decent domain authority. If you run a family-friendly hotel in the Lake District, a post on a well-known travel or parenting publication is worth more than one on a generic hospitality blog, even if the traffic numbers say otherwise. AI models weight relevance heavily.
Say something. The days of a bland 800-word overview of "five tips for better customer service" are gone. If your post could have been written by anyone, it teaches the model nothing about you. Take a position. Disagree with the received wisdom. Bring a specific view that is recognisably yours. This is where taste actually pays off.
Get your name in the byline and the body. Not just a bio at the bottom. When AI models parse a page, they weight in-body mentions more heavily than boilerplate author boxes. If you are quoted, named, or referenced in the actual argument, you are part of the substance.
Cover the same ground repeatedly, from different angles. One post on one site is a data point. Fifteen posts across fifteen sites, all reinforcing that you think and write about, say, low-intervention winemaking or urban dentistry for anxious patients, becomes a pattern the model cannot ignore.
The Uncomfortable Bit
This is slow work. There is no dashboard for it. You will not see a spike in Perplexity citations next week because you published a guest post yesterday. The payoff builds over months as models retrain and retrieval systems index the new material.
But that is also why it works. Anything with a dashboard and a quick payoff is already crowded. The tactics that require patience, judgement, and actual writing ability are the ones with room left in them.
If you are serious about being the answer when an AI recommends someone in your category, start writing. On your own site, yes. But also, increasingly, on other people's. This is the same reason context matters more than tools in almost every AI decision now: the model is only as useful as what it has to draw on.
