Every brand, from a scrappy startup to a Fortune 500, faces the same fundamental problem: How do you get discovered in a sea of consumer choices?
For decades, the answers shifted: first it was getting shelf space at brick and mortar retailers, then Amazon search as online sales took off around 2010. Then Google SEO. In every era, the challenge was the same — being visible in a world where gatekeepers decide who gets surfaced and who gets ignored. Now, the gatekeepers are AI agents.
Consumers are asking the large LLM’s ( ChatGPT, Claude, or Gemini) for answers instead of browsing endless search results. And instead of 10 blue links as was the norm in the period of search, they’re getting 3-4 curated recommendations. If your brand isn’t in that shortlist, you’re invisible. While SEO is part of this story (and yes, its power is waning), the bigger truth is that discoverability itself is being rewritten in the age of LLM’s.
The problem of being found isn’t new.
- Retail era: Your product lived or died by whether a buyer gave you shelf space at Walmart or Whole Foods.
- Amazon era: The battle shifted to keyword bidding and product reviews. If you didn’t optimize your brand’s PDP on Amazon or focus on your ratings and reviews, your ability to scale was limited.
- Google era: SEO/Affiliate ruled. Ranking on page one was the difference between growth and irrelevance. At Casper, for instance, our single largest lead gen source was from affiliates.
Today, consumers are beginning their searches on the large LLM’s in increasing numbers. “In a survey of 12,000 consumers, 58% (vs. 25% in 2023) say they now turn to Generative AI tools for product or service recommendations,” according to HBR and “roughly 1 in 7 brands now report that customers mention LLMs as part of how they found them,” according to Fairing.
Why AI Agents Matter Now? Simply put, AI assistants are compressing the consumer journey.
- From 10 results in the legacy search world to 3 recommendations on the LLM’s.
The funnel collapses, leaving fewer winners. - Agents prefer authority. Data from highly cited domains (Wikipedia, NYT, PubMed, government databases) shows up disproportionately in both training and grounding, whereas low-authority sources (random blogs, small affiliate sites) are less influential or even filtered out altogether.
- They overweight credible sources, reviews, and structured data heavily.
- The consumer trusts the answer. Unlike search where consumers click multiple links and visit a brand’s website to comparison shop and learn more, AI feels “final.”
That means brands can’t just “show up” anymore. They need to be chosen based on a newly defined set of criteria.
How do AI agents decide which brands to surface? It comes down to the data they trust.
- Authoritative mentions
- Coverage in credible outlets (NYT, Bon Appétit, Men’s Health).
- Inclusion in curated “best of” lists from high-trust publishers.
- Structured product data
- Schema.org markup, JSON-LD, nutrition facts, ingredients, UPC registry, etc
- Product feeds in ecommerce platforms like Amazon, Instacart.
- Knowledge bases
- Wikipedia/Wikidata entries.
- Google/Bing Knowledge Graph profiles.
- User-generated data
- Verified reviews on Amazon, Google, Yelp.
- Conversations on Reddit, forums, and social – heavily weighted by quantity of likes and overall AI summary of comments.
- Licensed or integrated data – brands and content owners with large enough reach to command a licensing royalty from the large LLM’s.
- Reddit – OpenAI ($60M).
- Publishers – Anthropic/Google.
- Future: brands licensing their catalogs directly.
The AIO Playbook for Brands
Here’s how brands should approach discoverability in the AI era.
Highest Impact (P0)
- Secure earned media in authoritative publications. This is different from being featured in affiliate marketing from leading publications. It means getting your brand mentioned naturally in articles.
- Build review depth and credibility on Amazon, Google, and key marketplaces. Brands should consider R&R aggregators like Bazaarvoice.
- Ensure structured product data is machine-readable and consistent. Structured product data means putting all your product details (name, ingredients, nutrition facts, price, availability, etc.) into a standardized, machine-readable format (like schema.org or retailer feeds) so AI agents and search engines can parse them without guessing. If that data is inconsistent across sites (e.g. your website says 45 calories, Amazon says 40), agents lose trust and are less likely to recommend your brand.
Medium Impact (P1)
- Create and maintain a Wikipedia/Wikidata presence.
- Publish agent-friendly content (FAQ/Q&A style posts). This is important because the nature of LLM’s are primarily question and answer engines so text in this format resonates.
Long-Term Bets (P2)
- Explore data licensing with LLM providers. As mentioned earlier proprietary data could be very valuable to the large models (brand whitepapers, research studies, product efficacy data (e.g., skincare trials, cleaning product lab tests).
- Instruction manuals, troubleshooting guides.
- Safety certifications and compliance docs.
- Integrate products into ecommerce APIs (Shopify, RangeMe, Mirakl, etc).
- Continuously test AI outputs to monitor visibility. LLM’s will often ask for feedback on a response. Brand Managers should play around to drive the AI in a certain direction using this feedback loop.
Prioritization: What Actually Drives Discoverability
Not all tactics matter equally though. From my consulting experience, the top three drivers are:
- Credible media mentions
- Verified customer reviews
- Structured product data
Everything else (Wikipedia, licensing, testing) builds on that foundation.
Using the lifestyle apparel brand, Alo, as an example, If a consumer asks ChatGPT: “What are the top three premium activewear brands?” Here’s what determines whether Alo shows up in that answer:
- Press Coverage (Authority)
- If Vogue, Women’s Health, or Forbes have profiled Alo, the AI has credible, authoritative mentions to ground its recommendation.
- Without this, Alo may not register as a trusted, “top-tier” option.
- Reviews (Consumer Sentiment)
- Verified reviews on sites like Nordstrom, Alo’s own site, or Google give the agent real consumer data to summarize. Without this, the AI has no signal that customers like the fit, comfort, or quality. Reddit also is an increasingly large source of data for the LLM’s given the recent licensing deals they’ve done. Brands should regularly monitor discussions around your products to spot trends or help drive the narrative.
- Structured Data (Machine Readability)
- Product info (materials, sizes, prices, care instructions) must be consistently tagged across Alo’s site, Amazon, and retail partners. As we’ve seen, even small variances are flags to the models.
- Without structured data, the AI can’t correctly parse that Alo sells “premium yoga leggings” vs “basic apparel.”
SEO isn’t dead, but it’s no longer the main battlefield. The real battle has always been discoverability and AI agents just rewrote the rules.
For me, this shift ties directly back to BD. My work has always centered on making brands discoverable through the right channels — whether that meant scaling the Lime x Uber partnership to put e-scooters & bikes in front of millions of riders, or structuring the Verizon x Blue Apron collaboration that drove customer acquisition, or pioneering the Alexa x Blue Apron skill that made recipes voice-accessible. AI agents are simply the next gatekeepers of distribution. Optimizing for them isn’t just about marketing tactics, it’s a business development challenge as well: securing partnerships, structuring data pipelines, and ensuring your brand shows up when the machine decides what options to surface. In other words, AI agent optimization is the new distribution strategy.