
Make PR your AI superpower

AI shapes how customers find your brand, and reputation determines which brands they choose. Drawing on the findings of our Brand Discovery Index, Beth Nunnington, VP of PR, gives practical advice on how brands should adapt their PR strategy for an AI-driven discovery landscape.
Research is happening before buyers even start talking to your brand. To put your brand in the best position for discovery, this means you should start by mapping the sources of influence that shape buying in your category. For a SaaS brand, that might be G2, niche trade titles, comparison reviews on the likes of NerdWallet, thought leadership in Forbes and Business Insider, relevant subreddits and industry podcasts. For a consumer brand it might be Wirecutter, CNN Underscored and a group of TikTok creators.
Next, show up consistently across them: named spokespeople giving expert commentary, original data that journalists and AI platforms both cite, inclusion in comparison and best-of content, and a respectful, community-led approach to communities like Reddit (don’t even attempt a hard sell here - it will backfire). Princeton’s research found that adding statistics, quotations and citations to content can lift visibility in AI answers by up to 40%, which matches what we see from data-led and expert-led PR. Recency matters too. AI weights fresh sources, so a steady drumbeat of activity will do more than a one-off big launch a year.
Start with relevance: is the coverage talking about the category the brand wants to be known for? From there, four types carry real weight. Listicles and best-of roundups matter because they’re exactly what AI platforms pull from when someone asks which brand to choose. Product placement in product roundups, reviews and comparison pieces puts the brand in front of buyers at the point they’re deciding. Unique data studies give a brand something new to own, which helps it stand out from the flood of AI slop repeating the same generic points. And thought leadership from named spokespeople is even more crucial than ever, because humans buy from humans, and AI platforms look for real expertise attached to real names.
Then there are the wins that look good in a coverage report but do little for discovery: a quirky survey story that holds no relevance to the brand, a viral stunt that never mentions what you sell, a high-authority link in an article about something unrelated, or a wire release picked up by automated sites nobody reads. A niche podcast your buyers listen to can be worth more than an unrelated brand mention on a major national news site.
The Index shows us that 75% of US marketers audit Share of Model at least monthly, yet 64% still make performance decisions without adequate competitor visibility. Use it to shape the strategy and decide which publications to prioritise first. Share of Model tells you how often AI recommends you compared with your competitors, and you can break it down by prompt and by platform.
The platform view is really useful, because ChatGPT, Claude, Gemini and Perplexity all favor different sources. A brand can do well on one and hardly show up on another. So we look at which publications each platform is citing when it recommends a competitor, and then layer that over the platforms and publications the brand’s audience is actually engaging with.
The overlap gives us a clear target list. It also shows us how AI is describing a brand next to its competitors. If a competitor keeps getting called the best-value option, we know exactly what story we need more third-party proof for. From there we track it against our activity to see which placements made a difference, and put more behind what’s working.
To ensure AI is working in your brand’s favor, you need to do three things: strategically target the right sources, measure what matters and bring teams together around a consistent story.
AI will keep repeating an old or negative story for as long as the strongest sources are saying it. It has no way of knowing a problem was fixed unless something credible tells it. How long it takes to change depends on how entrenched the story is and how much new evidence you can put out there.
With consistent activity you’d usually start to see AI answers change within a few months, but anything with years of old press behind it will take longer. The most important thing is to deal with the issue directly. If there are real problems with your product, fix them, and make sure negative reviews are responded to. Getting involved on Reddit can help too, as long as you’re offering useful advice with no sales pitch attached.
Once you’re sure everything has been fixed, invite your key press to review the product and get it in front of creators. Creator content sits somewhere between earned, owned and paid, but we find it’s a great way to help rebuild trust alongside PR and paid activity. It’s also worth correcting errors at the source, whether that’s an outdated line on Wikipedia or a comparison article still listing a product you’ve discontinued.
Pushing out more owned content isn’t going to fix the issue by itself. AI gives far less weight to what a brand says about itself, and a big promotional campaign on your own channels could even make the gap between what you say and what everyone else says even wider.
When we run the Brand Discovery Index again next year, I’d expect the 99% who are confident they’re visible where customers discover them to fall the most. As more teams audit how AI describes them, more will find gaps they didn’t know were there, and it’s hard to stay that sure of yourself once you’ve seen a competitor recommended for a prompt you thought you owned.
I’d actually be glad to see that number come down, because it would mean brands are looking more closely. I’d also expect the 38% who say getting AI to recommend them is their biggest challenge to go up, as AI answers become a bigger part of how people research and the commercial impact gets easier to see. ChatGPT-referred ecommerce visits already convert at more than double the rate of organic search, so it’s going to be much harder for brands to ignore.
AI recommends the brands it can find credible proof for. Producing that proof, independent coverage, expert voices, third party sources, is what PR does. That’s what makes it the secret superpower for discoverability in the AI search era.
Win the research journey before buyers reach your website
Research is happening before buyers even start talking to your brand. To put your brand in the best position for discovery, this means you should start by mapping the sources of influence that shape buying in your category. For a SaaS brand, that might be G2, niche trade titles, comparison reviews on the likes of NerdWallet, thought leadership in Forbes and Business Insider, relevant subreddits and industry podcasts. For a consumer brand it might be Wirecutter, CNN Underscored and a group of TikTok creators.
Next, show up consistently across them: named spokespeople giving expert commentary, original data that journalists and AI platforms both cite, inclusion in comparison and best-of content, and a respectful, community-led approach to communities like Reddit (don’t even attempt a hard sell here - it will backfire). Princeton’s research found that adding statistics, quotations and citations to content can lift visibility in AI answers by up to 40%, which matches what we see from data-led and expert-led PR. Recency matters too. AI weights fresh sources, so a steady drumbeat of activity will do more than a one-off big launch a year.
The coverage that influences AI discovery
Start with relevance: is the coverage talking about the category the brand wants to be known for? From there, four types carry real weight. Listicles and best-of roundups matter because they’re exactly what AI platforms pull from when someone asks which brand to choose. Product placement in product roundups, reviews and comparison pieces puts the brand in front of buyers at the point they’re deciding. Unique data studies give a brand something new to own, which helps it stand out from the flood of AI slop repeating the same generic points. And thought leadership from named spokespeople is even more crucial than ever, because humans buy from humans, and AI platforms look for real expertise attached to real names.
Then there are the wins that look good in a coverage report but do little for discovery: a quirky survey story that holds no relevance to the brand, a viral stunt that never mentions what you sell, a high-authority link in an article about something unrelated, or a wire release picked up by automated sites nobody reads. A niche podcast your buyers listen to can be worth more than an unrelated brand mention on a major national news site.
Share of Model becomes a strategic PR tool
The Index shows us that 75% of US marketers audit Share of Model at least monthly, yet 64% still make performance decisions without adequate competitor visibility. Use it to shape the strategy and decide which publications to prioritise first. Share of Model tells you how often AI recommends you compared with your competitors, and you can break it down by prompt and by platform.
The platform view is really useful, because ChatGPT, Claude, Gemini and Perplexity all favor different sources. A brand can do well on one and hardly show up on another. So we look at which publications each platform is citing when it recommends a competitor, and then layer that over the platforms and publications the brand’s audience is actually engaging with.
The overlap gives us a clear target list. It also shows us how AI is describing a brand next to its competitors. If a competitor keeps getting called the best-value option, we know exactly what story we need more third-party proof for. From there we track it against our activity to see which placements made a difference, and put more behind what’s working.
The 3-step PR strategy for algorithms
To ensure AI is working in your brand’s favor, you need to do three things: strategically target the right sources, measure what matters and bring teams together around a consistent story.
- The strategy: Build it around the sources of influence that matter to both the machines and your target audience. That means working closely with the SEO and GEO experts to build a performance-driven media list, and optimizing press materials and content for both the keywords and the prompts people use when they search. Doing that consistently keeps our messaging tight and helps AI place us in the right category.
- The measures: If PR is still judged on coverage volume and domain authority, it will keep chasing wins that look good in a report. I’d want to see citation share in the sources that matter for our category, how often we show up in priority AI prompts, branded search, and how closely AI’s description of us matches the way we describe ourselves.
- The team: I’d want PR working much more closely with the influencer, content and social teams. When every channel tells the same story, people and AI platforms build the same picture of who the brand is, and that consistency is what gives a brand a clear identity online.
When AI gets it wrong
AI will keep repeating an old or negative story for as long as the strongest sources are saying it. It has no way of knowing a problem was fixed unless something credible tells it. How long it takes to change depends on how entrenched the story is and how much new evidence you can put out there.
With consistent activity you’d usually start to see AI answers change within a few months, but anything with years of old press behind it will take longer. The most important thing is to deal with the issue directly. If there are real problems with your product, fix them, and make sure negative reviews are responded to. Getting involved on Reddit can help too, as long as you’re offering useful advice with no sales pitch attached.
Once you’re sure everything has been fixed, invite your key press to review the product and get it in front of creators. Creator content sits somewhere between earned, owned and paid, but we find it’s a great way to help rebuild trust alongside PR and paid activity. It’s also worth correcting errors at the source, whether that’s an outdated line on Wikipedia or a comparison article still listing a product you’ve discontinued.
Pushing out more owned content isn’t going to fix the issue by itself. AI gives far less weight to what a brand says about itself, and a big promotional campaign on your own channels could even make the gap between what you say and what everyone else says even wider.
The future of brand discovery
When we run the Brand Discovery Index again next year, I’d expect the 99% who are confident they’re visible where customers discover them to fall the most. As more teams audit how AI describes them, more will find gaps they didn’t know were there, and it’s hard to stay that sure of yourself once you’ve seen a competitor recommended for a prompt you thought you owned.
I’d actually be glad to see that number come down, because it would mean brands are looking more closely. I’d also expect the 38% who say getting AI to recommend them is their biggest challenge to go up, as AI answers become a bigger part of how people research and the commercial impact gets easier to see. ChatGPT-referred ecommerce visits already convert at more than double the rate of organic search, so it’s going to be much harder for brands to ignore.
AI recommends the brands it can find credible proof for. Producing that proof, independent coverage, expert voices, third party sources, is what PR does. That’s what makes it the secret superpower for discoverability in the AI search era.








