Harsh Jangid is Director at Coozmoo, an AI-powered digital marketing agency that helps SMBs accelerate growth and increase revenue.
A few months ago, one of my clients in the cosmetics space came to me confused. Paid search numbers looked fine. SEO rankings hadn’t moved. But conversions had dropped 21% over a matter of weeks, with no obvious cause on any dashboard we normally check.
It took us a while to find the real story, because it wasn’t hiding in Google Analytics; it was hiding in ChatGPT. When we typed in the kind of question their actual customers were asking, “What’s the best vitamin C serum for sensitive skin?” their brand simply didn’t come up. Competitors did. Not because those competitors made a better product but because their product data, reviews and content were structured in a way AI models could actually find, understand and cite.
That was the moment this stopped being an abstract trend and became a client-facing crisis. Once I started checking for it across other accounts, I saw the same pattern: Brands absent from AI answers were seeing conversions soften even while every traditional metric looked healthy. The erosion doesn’t show up as a traffic crash. It’s a slow leak; the customer never lands on your site to be “lost” because they never considered you at all.
The Customer Journey Has Grown A New, Invisible Step
For 20 years, digital marketing has run on one assumption: a person types a query, scans results or ads and clicks. SEO, paid search and conversion optimization are all built around influencing that human at that moment of choice.
That assumption is breaking. The “search” and the “shortlisting” now often happen inside a conversation with an AI agent before a human ever opens a browser tab. The shopper tells ChatGPT or Perplexity or Amazon’s Rufus their budget and use case, the model does the comparison shopping and by the time a link is clicked, the decision is largely made. The data backs up what I saw anecdotally: Retail sites are reporting AI-referred traffic growing well over 100% (paywall) year over year, and multiple analytics providers now report that shoppers arriving via ChatGPT or Gemini convert noticeably higher—and spend more per visit—than traditional organic traffic. The volume is still small next to Google’s. The trajectory isn’t.
AI isn’t just a new traffic source to add to the media plan. It’s becoming a new customer in the funnel, one that reads your site differently than a person does, has no loyalty to your brand name and will happily recommend whoever answered its question more clearly.
Most Businesses Are Simply Not Set Up For This
I’ll say plainly what I now tell every client: The vast majority of businesses I talk to, including sophisticated, well-funded ones, have no strategy here at all. They have an SEO plan and a paid social plan. They do not have a plan for being understood, trusted and recommended by a machine.
That’s a real gap—the same kind that once separated brands that took mobile seriously from those that treated it as an afterthought. The businesses that get this right early won’t just gain a temporary edge; they’ll set the baseline everyone else has to catch up to.
Part of the problem is structural. Marketing teams are organized around channels humans use: search, social, email and metrics tied to human sessions. Nobody owns “AI visibility” yet, so it falls into the gap between SEO, PR and IT. Part of it is a trust problem. Leadership doesn’t yet believe AI-referred volume justifies budget, which is the exact argument I heard about mobile search 15 years ago.
What ‘Marketing To Machines’ Actually Requires
Here’s where I think businesses need to focus first, based on what’s worked and badly failed for our clients:
• Make product information machine-readable, not just human-readable. AI models pull answers from structured data, clear specifications, comparison content and third-party validation—not from beautifully designed but semantically vague landing pages. If a model can’t parse what your product does and who it’s for, it recommends the competitor whose site makes that easy.
• Treat external reviews and mentions as a core visibility channel in machine-readable data. AI models weigh what’s said about you across review sites, forums and comparison articles more heavily than what you say about yourself. Thin third-party presence makes a brand nearly invisible to these systems.
• Answer the question, don’t just rank for the keyword. Traditional SEO optimizes for a search term. AI visibility rewards content that fully answers the underlying question a buyer would ask because that’s what models extract and cite, a genuine mindset shift for teams trained to think in keywords. One analysis our company did of 250-plus AI answers across ChatGPT, Perplexity, Gemini, Claude and AI Overviews found the same handful of content formats getting cited again and again.
• Start measuring it, even imperfectly. Most businesses can’t tell you what share of traffic or revenue comes from AI referrals because it isn’t tracked as its own channel. You can’t manage what you don’t measure, and almost nobody is measuring this well yet.
The Uncomfortable Truth
None of this is fully solved, not by Google, not by OpenAI, not by any agency, including mine. The rules of AI answer-engine optimization are still being written in real time, and anyone claiming to have it figured out is overselling. But that uncertainty is exactly why businesses willing to experiment now, while the playing field is still being drawn, stand to gain the most.
The question isn’t whether AI agents will influence more of your customers’ decisions—that’s already happening, quietly, inside conversations you can’t see. It’s whether, when your next customer asks a machine for a recommendation, your business is even in the running to be named.
Forbes Technology Council is an invitation-only community for world-class CIOs, CTOs and technology executives. Do I qualify?