Growth marketing used to mean owning Google rankings and paid social feeds. That definition is now incomplete. A meaningful share of high-intent product research happens inside ChatGPT, Perplexity, and Google AI Overviews before a brand website ever loads, and most growth budgets still are not built for it.
What changed in how people research before they buy?
Search behavior split into two channels that used to be one. People still type queries into Google, but a growing share of the same research now happens as a conversation with ChatGPT, Perplexity, or Gemini, where the AI system reads dozens of sources, synthesizes an answer, and never shows the user a list of blue links. 5W's own research, published in The GEO Reckoning, tracks how often brands get cited inside those AI answers instead of just ranked on a results page. A brand that ranks first on Google can still be absent from the answer ChatGPT gives, because the two systems weigh different signals.
I have written before about how this gap shows up in category after category. Our wine brand AI visibility research found a brand named in 36.9 percent of AI answers about its category, yet cited from its own website in exactly zero percent of those answers. The brand exists in the AI's training data. It just is not the source the AI trusts enough to link back to. That distinction, appearance versus citation, is the one most growth teams have not learned to measure yet.
Why does this break the old growth marketing playbook?
Classic growth marketing runs on a feedback loop: spend on a channel, measure the click, retarget the visitor, repeat. That loop assumes the user lands on your site. When Perplexity or an AI Overview answers the question directly, there is no click to measure and no visitor to retarget. A growth team that only reports on sessions and CAC from paid and organic search is now blind to a share of the demand it used to see, and cannot optimize for a channel it cannot measure.
We saw the same pattern in our esports AI visibility gap research: teams were named by AI chatbots often, but rarely cited as the source behind the answer. Being mentioned is not the same as being credited, and only the second one sends any signal back to a growth dashboard.
What does an AI-era growth marketing program actually add?
The fix is not replacing performance marketing. It is running Generative Engine Optimization (GEO) alongside paid media, SEO, and lifecycle marketing as one system, so the same brand shows up whether a customer searches on Google or asks an AI engine directly. That is the model 5W's growth marketing practice runs for clients: structured content and entity signals for GEO, paid and organic channels for demand capture, and retention programs so the customers acquired through either channel stick around.
Our peptide brands AI visibility index showed how fast this can move in a single category. Brands that had never invested in structured, citable content were absent from AI answers entirely, while smaller competitors with clearer sourcing showed up consistently. Growth marketing decided by which brand kept its facts easiest to verify, not by which one spent more on media.
How should a marketing leader measure this?
Blended CAC and LTV still matter, but they now need a second layer: how often does the brand actually get cited when someone asks an AI engine a category question, and does that citation lead anywhere a customer can act on it. A brand that cannot answer both questions is optimizing half of its funnel and reporting on the other half as if it does not exist.
That second layer needs its own reporting cadence, not a footnote in a quarterly deck. AI citation share moves in weeks, not quarters, as engines re-crawl and re-weight sources, so a growth team checking it once a year will always be reacting to a shift that already happened.
The agencies and in-house teams that treat GEO as a separate, secondary project will keep measuring a shrinking share of how their customers actually find them. The ones that fold it into growth marketing from the start will be the ones whose acquisition numbers still make sense in two years.