Content Marketing is changing. Not just recently, though; it’s been building for years, says Andrew Wheeler, CEO of Skyword, an Enterprise Content Marketing agency. But something is pushing change faster today – ahem, AI! – and because of it, the website is no longer the starting point for buyers.
AI answers are becoming today’s new first impression, and brands risk losing visibility before a potential customer ever reaches the website. What to do? They turn to Generative Engine Optimization (GEO). Some Marketers use the term Answer Engine Optimization (AEO) instead, although GEO generally refers to the broader challenge of building authority across generative AI systems.
The problem is that many are treating AI search like Search Engine Optimization (SEO) when it first came out. But the smart ones are focused on building real authority and credibility, making them the answer customers seek. But it’s not an easy road.
Is GEO so different from SEO?
When brands started working with SEO, we saw a lot of keyword stuffing in content and metadata, invisible text on the page for Google to read, and real backlink farms. Today, we’re seeing the same behavior, just different technology.
With GEO, we get fake citations, AI-generated recommendation spam, and fake expert ecosystems. And that’s just to get surfaced in the Large Language Models (LLMs). The website still matters, and what’s happening, says Wheeler, is that smart brands are trying to feed both masters.
Google is trying to fight this by building out spam and privacy policies, and it’s driving differentiation for them relative to other frontier models (like OpenAI, Anthropic, and Perplexity), though likely not for long, he argues:
Right now, I think a lot of brands are treating, are still treating AI optimization like a shortcut, you know. Again, how do we get mentioned, how do we influence prompts, and how do we appear in summaries? And the brands that do it right, the long-term winners, are not going to be the brands that are just trying for quick hits or manipulating outputs. They’re going to be the brands that become sources of information across the broader digital ecosystem.
The difference between influencing an answer and becoming a trusted source
Wheeler believes that AI systems will increasingly evaluate patterns of credibility. They are going to learn which brands are repeatedly cited, such as how trusted third-party publications reference them, which experts consistently appear in authoritative contexts, and whether they are real. What brands demonstrate expertise across different channels and conversations? Then the AI and its prediction engines will rely more on the sources that are repeatedly cited to know who’s most authoritative.
Skyword research on AI buyers shows that when AI-generated answers conflict with a brand’s website, buyers look for third-party signals of accuracy and authority. So this makes a lot of sense.
But Wheeler contends there’s an important piece to consider: the difference between influencing an answer and becoming a trusted source. AI systems will not develop trust from a single mention. He said they will develop that trust through repetition and statistical reinforcement across authoritative contexts:
The only way that I believe a brand can get consistently cited [is] if their content offers clear expertise, a distinctive point of view, third-party validation, consistent reinforcement across channels. You can’t just write an article and stick it on your site.
One key is to find out where your audience is spending their time and make sure you’re reinforcing your message in all of these channels. But that is getting harder all the time, especially getting third-party media sites to cover your brand, getting your leadership out there to share important perspectives, and creating valuable social media content that isn’t ‘AI slop‘.
Wheeler’s company is banking on people wanting to read content created by real humans:
We believe that folks like yourself are masters of the craft, and we can’t outsource content creation to the machines. We can outsource content manipulation, content derivative, and content optimization after the original piece is created. And so, yes, it’s getting harder, but if we have that core asset that was created by Andrew, now I can constrain prompts around creating a version of this for this audience, a version of this for this channel. Do not introduce anything new that wasn’t in the original piece. No room for hallucination, no room for fake citation; go off the original.
It’s this approach that Wheeler believes will allow brands to stay authentic.
What about the website?
Again, the website is important, but often neglected in this effort to build authoritative content for LLMs. Think about it this way: a brand builds a great blog with lots of content that’s picked up by third-party publications, widely shared on social media, and where the leaders are invited to podcasts and webinars, etc.
The brand now appears in AI answers with the website cited as the source, and some people click through. But the website doesn’t match the content surfaced in the LLMs. Sure, the blog is there, but the brand and product messaging don’t match the content cited in the blog or podcast.
There is an entire Content Marketing strategy a brand needs to think about; one that includes how the entire website aligns to a story that meets the needs of its intended audience. Wheeler says:
That’s where it’s really, really important. It has to be a very purposeful, thematic sort of view for the website, making sure that it’s not just one and done. […] But we’ve been preaching this for years, and I don’t mean just we, Skyword, but we more broadly; it can’t be just a, here are the 25 keywords I want to rank on. It has to be much more thematic and thoughtful in the way that the audience is going to be looking for information.
Focus on the category, not the prompts
Skyword has a history of helping brands win in search, so how is it adapting to supporting generative engine optimization? Wheeler explained that the key is not to identify a list of prompts people use and create content for them. The right approach is to understand what category you earn the right to win in and create a strategic content plan around it, says Wheeler:
How do we actually establish a system around it that the LLMs will see as credible and want to cite because they see folks referencing you, they see credible experts talking about the brand, quoted in material?
The idea is to create content with unique expert insights so it continually improves how the brand shows up in AI search. Wheeler said the outcome from this work is great source attribution and AI answers, increasing brand searches, and ultimately, improving business outcomes.
Brands are stuck on discoverability in an SEO sense, AKA ‘just get found’. But Wheeler wants CMOs to think differently:
We’re advising CMOs to prioritize a bit differently than they had been. Again, not to keep using the word, but to think about building category authority early and really strengthening expert visibility, whether those experts are outside or within their organization.
Think about the logo swap test, he suggests – would someone be able to tell what content comes from your brand versus your competitors?:
You have to have that differentiated insight. Spend time auditing how the brand is appearing across different AI systems. The folks that are doing this well are compounding the results, and it’s going to be an advantage. And you don’t want to fall so far behind that it’s harder to catch up. So, we recommend that CMOs prioritize experimentation, but balance. You want to experiment tactically, not strategically. If you start changing everything up all the time, you don’t know what’s working.
The other important thing to remember is that this work takes time. AI can help you be more efficient, but it cannot drive credibility. Brands should not take shortcuts, concludes Wheeler:
The competitive advantage in AI search is not content volume. It’s really about whether the AI system recognizes your brand as a credible authority…That should be the full point, not that if I just create more content for what people are prompting and searching for, I’ll be okay. It’s really important to think about it more from an authority ecosystem.
My take
Here’s an interesting stat from Skyword’s AI buyer research – 47% of consumers have taken at least one significant action based on AI-generated information, such as avoiding a purchase, switching brands, warning others, or contacting a company.
What does this tell you? Your brand needs to show up in AI search in a positive way. And by positive, I mean a consistent perspective that speaks to who the brand is, what it stands for, and how it wants to help people improve, grow, and do something. Writing content for single prompts won’t do that. Listicles that put your brand in a more positive light than your competitors won’t do that.
In the rush to get more content out so the LLMs see the brand, marketers are missing the mark and wondering what they are doing wrong. The hard work of a real content strategy is more important now than ever.
