Elevator Talk: Is Duplicate Content Actually an AI Visibility Play, or Just Wasted Crawl Budget?

My take is that duplicate content is probably net neutral. I think about it similarly to what happened with Penguin and backlinks, where spammy links increasingly just stopped counting rather than actively hurting you. Reposting the same content across multiple platforms probably won't hurt, but I'm also not convinced it gives you much.

What seems to matter more for LLM visibility is consistency and corroboration across the web. There's also a lot of context that goes into that. What topic are you trying to build authority around? What are you trying to reinforce or validate? Does it actually need corroboration? And are those signals showing up across credible, independent sources?

So I think repetition can have a role, but simply duplicating content everywhere is unlikely to create much additional value.

I think to fully understand the context of duplicative and repetitive content and how it affects LLMs, it helps to dive into how LLMs work a little more. When we prompt Chat or Gemini or whatever LLM of your choice, it uses RAG (Retrieval-Augmented Generation) to pull relevant data for a response. But LLMs have limited context windows and have to spend their compute wisely. So they don't want to read the same thing over and over — they need to get a consensus quickly and efficiently.

To combat this, they also use something called MMR, or Maximal Marginal Relevance. It's an algorithmic search/retrieval method that ensures results aren't just perfect matches saying the exact same thing, but instead picks the result that matches the query or intent while remaining different from what's already been chosen.

Essentially, MMR filters out redundancy. So if we're posting the exact same thing over and over, it's kinda like having a football team full of only running backs. Except it's worse, because AI would look at a team of running backs and go 'these are all the same, I don't need them all,' bench 10 of them, and I'd only have 1 player on the field. Instead we need a team of players that share the same goal but bring their own unique skills to the table. And honestly, AI is getting pretty good at finding out pay-to-play. Posting the same piece on the blog, then LinkedIn, then Substack or whatever is like a running back pretending to be a lineman and the quarterback, not a true team working together.

Real E-E-A-T is consensus. So different high-authority entities independently saying a version of the same thing, not one entity saying the same thing in three places. And that's not just an AI thing — people do the same thing. We look for trust signals before we take an action, like reading product reviews, checking out the socials, etc. Which, tbh, is more important since at the end of the day we're marketing to humans, not algorithms.

TLDR: Repetition builds our brand's association with a topic, but true visibility requires giving LLMs a good reason to pull every single thing we publish.

I land on repetition ≠ duplication ≠ syndication.

My hunch is that repetition can absolutely help with LLM visibility, but mostly when it creates consistent signals, not just more copies. If we keep articulating the same distinctive POV, terminology, expertise, and evidence across different pages and platforms, we're making it easier for models to associate us with that idea.

That's different from publishing the exact same article on our blog, LinkedIn Pulse, Substack, Medium, etc. At some point, those extra copies probably stop adding much new information or authority.

Distribution can amplify authority, but distribution itself isn't authority. If the footprint looks like 'this is everywhere because someone pushed it everywhere,' that's a weaker signal than 'this idea keeps showing up because other credible sources are engaging with, citing, or building on it.'

So I'd draw the line like this: Repetitive — good, when we're reinforcing a recognizable idea from different angles. Duplicative — probably diminishing returns, especially if we're just copy/pasting.

I suspect the AI-search play isn't 'publish the same thing everywhere.' It's 'be unmistakably associated with the same idea everywhere.'

A lot of my thoughts echo what Sam and Alyssa have already said, especially the part about having multiple locations say the same thing about your company or product and your expertise in slightly different phrasing in content that is different from each other. Having the exact same content across multiple platforms will likely come out net neutral like Sam's saying. It might get more users' eyes on it, but it's not going to help you much if at all for the robots side of things (SEO, AEO, etc.).

Brands need to have strong entity relationships to perform the way they want to in LLMs and organic, and that's not going to happen with duplicative content. It's going to happen with unique content across multiple high authority domains saying variations of the same thing about a brand's expertise, UVP, and product offerings.

One factor that could make syndication more interesting that I hadn't considered until now is the author entity. If the same author is consistently publishing around a topic across LinkedIn, Substack, industry publications, their own site, etc., that broader footprint may help establish a stronger connection between that person and their expertise. Things like consistent bios, credentials, topic focus, and publication history can all contribute to E-E-A-T and trust signals. So even if the duplicate content itself is neutral, there could still be value in what that distribution does for the author entity over time.

Regarding Sam's author entity thought, I think that makes total sense — while Google hasn't indicated there is a specific 'author authority score,' this would work similarly to our topical authority strategy. Google recommends clear authorship and bylines where readers would expect them, and its helpful-content guidance explicitly asks whether content demonstrates the creator's experience and expertise.

Connecting the same named author with the same subject matter across credible platforms could strengthen the web's understanding of 'Person X → expertise in Topic Y.'

I think including the Person/author structured data as part of the entity optimization would be super cool to test, specifically for clients currently doing content syndication.

If we are saying duplicating content (1:1) across multiple sources/domains is going to provide credibility to an idea or a take, I am cautious of trying that. Even if it does work, I doubt it's a very evergreen strategy and sounds more like a hack that would quickly be deemed 'spammy tactics' by Google. My reasoning being:

It's a very easy way to solidify an idea and will 1000% result in overuse and manipulation of search results. You can't give SEOs that low hanging of fruit and expect the industry to go overboard using it.

I mostly agree with Sam and Jasmin's takes of it being similar to links, in that it will either help you or do nothing, and if it is helping, the returns will diminish quickly. But I actually think that is best case scenario, and could see this being something Google does end up penalizing a domain for if they are not already. If we put out the same exact content on multiple pages, on different domains, with the intention and hope that Google finds and crawls it to establish credibility to an idea without giving it anything new or unique, Google is going to see that as a waste of compute power to crawl the same thing over and over again. Google HAS to crawl the web with efficiency in order to effectively find the right information for users. This is evident by many things, but particularly by the emergence and growth of the 'crawled not indexed' section in GSC.

All of that being said, I do believe that the goal of an LLM or any info retrieval system is to come to a consensus on a subject, and then deliver that consensus to the user. So repeating an idea across domains and channels is a good thing, but it should be tailored to the channel it is being distributed on.