Part 2 in a series exploring how to navigate AI’s influence on your communication. View the first article here.
Paul Frick: Earlier this month, I posed a question that has been coming up more frequently in our strategy discussions: Is AI an audience in its own right? My opinion: AI is the latest intermediary between an organization and its audience, and one that we have to plan for to be truly effective.
But how? That’s the question Mimi Tan, Avoq’s lead UI/UX and content strategy expert, explores today. She’s worked with organizations like Special Olympics and NAACP to boost their search rankings through content strategy, and helped to develop the GEO (Generative Engine Optimization) offering at Avoq. Her answers confirm my suspicions, and will help you be more effective in getting your story to the right audiences.
Paul: Mimi — give us your perspective — is AI an audience for communicators?
AI is now the go-to brand intermediary, whether we like it or not. It’s your frenemy — it can talk you up but it can also cut you down. That’s why your AI brand narrative needs to be proactively managed. Lots of organizations are approaching GEO as a technical or content challenge, when really, they should approach it as reputation management.
We’re in the infancy of GEO right now. The LLMs (Large Language Models, the technology underlying AI) will continue to evolve, but we expect the models to converge with what human audiences prefer in the long run. As opposed to how things lean now, where LLMs favor what’s easy for them to parse and regurgitate.
We should probably pause here for clarification…how does GEO differ from SEO?
With GEO, organizations are vying for mentions and citations in AI-generated results. SEO, or Search Engine Optimization, is about competing for higher rankings in search results on Google and Bing. SEO uses a different model but many best practices overlap, such as well-structured content, descriptive language, and so on.
So when I ask AI a question about my company, what does it look at to provide the answer?”
AI is definitely looking for your brand story and any indicators of reputation. But it rewards multiple sources and it likes it when sources both within and outside of your control agree. AI wants to see consistency in messaging and corroboration of those points: consensus across the web.
AI does spread misinformation and disinformation based on where it’s sourcing answers. So it’s important to be aware of how you’re being portrayed in the various platforms based on prompts on your brand, executive leadership bios, key issues, etc., and to target your audiences based on the platforms where they are.
Why are answers to my AI queries all different?
That’s a big part of the challenge with managing AI and why you need a deliberate strategy. The answer you get will vary by prompt, platform, and timing. The good news is that many of the sources are ones that you can manage, like your website, social media, and blog. But it will also search sources that are harder to manage, like media outlets, Reddit, and Wikipedia. These LLM agents are looking for consistency and corroboration in sources, so they’re more likely to report answers that are validating what they’re seeing in other places.
The “trust factor” gives some indication of where to focus your management energy. But what’s considered a trusted source varies by the platform. They all have very different business interests at heart, and will prioritize their own favorite children. For instance, Gemini is owned by Google, so it favors sourcing from YouTube — its own platform — whereas Claude rarely cites it.
We love to boil things down to a top three. If you had 90 days to improve how AI presents an organization, what are the first three things you would do?
First, figure out your baseline and get the lay of the land.
Get your AI visibility score so you can measure progress over time.
To fully understand how the major AI platforms are talking about your organization, you need to use a GEO tool that can run prompts at scale, because the LLMs don’t always generate the same answers for the same prompt. Avoq’s GEOMetriq tool can do this and provide an overview report that helps you identify where there are gaps in your presence. The analysis can help you better understand which issues you need to proactively manage — and get ahead of.
Next, focus on the care and feeding of content on your owned channels, particularly your website.
- At least once a year, run a content audit, which will help you initiate stakeholder conversations about what content to keep, update, and sunset. If old content isn’t getting any eyeballs, consider archiving it to help clean up search results and ensure accurate AI-generated responses.
- Inconsistent, outdated, and inaccurate content will impede your message in AI-generated responses. A common example is staff bios for people who no longer work at your organization, outdated/inaccurate reports, or policies and resources that are no longer relevant. AI also favors content published in the past year, which is why you should review and update popular content and why you’ll often see “last updated” dates on web pages.
- Once a month, review website analytics to identify what pages are getting traffic and engagement, and what AI platforms are sending the most traffic to your site, to guide stakeholder conversations around your editorial planning. High-performing content can give you a sense of what’s resonating with your audiences — offering valuable clues on what related topics you may want to build out for driving traffic — which can offer a direct SEO boost and indirect GEO boost.
The third thing — and this is the stumbling block for most organizations — don’t only focus on the easy things. You have three primary levers you can pull to manage AI-generated responses: owned media and platforms, the earned media your organization generates, and to some extent, your paid media. Owned is going to be the first thing you attend to. But earned and paid are important, and they require patience because of the longer timelines involved to plan and develop action plans.
What can a strong content strategy can influence? Are there things that you shouldn’t expect to influence?
AI is looking for a few things.
- “Extractable” content that it can easily parse and remix, which is why clearly structured content like FAQs perform well.
- Citations and stats to support any claims.
- Multimedia, especially video.
- A human, conversational tone, which is easier for online audiences — both human and robot — to understand.
- Above all, corroboration from third-party sources to validate what you’re saying about yourself.
One study published by Muck Rack earlier this year estimates that nearly 40% of AI citations are coming from websites and corporate blogs. That doesn’t translate to direct control over 40% of what AI platforms source. It’s more a reminder that you should be paying close attention to the low-hanging fruit: your owned channels, with your website as the most important channel for AI influence that’s within your control.
Sources such as Wikipedia and earned media require a strategy, patience, and tempered expectations of what you can and can’t shape. It doesn’t mean these are impossible options, but you have to commit to a process and be willing to wait for results.
No matter your content strategy, remember to prioritize human-created, human-centric content. Do we really want to live in a world where we’re regularly relying on AI to regenerate other AI content, rinse and repeat? Ultimately, we’re the ones who decide whether the snake ends up eating its own tail.
