How AI is restructuring the path from “I need something” to “I’m buying this,” and where humans still matter.
When I was recently planning a family trip to Japan, I noticed I’d split the entire process in two without thinking about it. I talked to friends who’d taken their kids. I had my own ideas about what we wanted the trip to feel like. That was all human: the vision, the emotional blueprint. Then AI did the logistical work, suggesting neighborhoods to stay in, that made sense given what we wanted to see, checking which days museums were closed, and building an itinerary that worked for my husband’s work schedule.
I didn’t want AI to tell me where to go, but I appreciated its help in getting me there once I knew what mattered.
It got me wondering whether the same split was happening across every consumer journey. Turns out, the data says it is.
64% of consumers now use AI tools to research purchases. A year ago, that number was 45%. (BusinessWire)
That’s a big jump in a short time, faster than most brands have clocked, and it goes deeper than “people are Googling less.” Among consumers who use AI while shopping, it’s now the second most influential source in their purchase journey, ahead of retailer websites, social media, and recommendations from friends and family.
What’s actually happening is the same split I made planning Japan, just at scale. People are handing AI the parts of buying they want handled and keeping the parts they don’t.
The split looks different depending on where you are in the journey.
Discovery: How people find your brand has changed
Discovery used to be relatively visible to brands. Someone searched for something, clicked on your site, landed on your content. You could see the trail.
That’s less and less true. There are now two AI-informed paths to discovery, and brands have limited visibility into either of them.
On one path, people are going to AI deliberately with purchase intent. Typing “best recovery drink for after a long run” or “best developmental toys for a 12 month old” into ChatGPT. BCG found that people aren’t just using AI to answer questions they already have. They’re using it to figure out what they want in the first place. Someone asks how to make a healthy taco and ends up with a grocery list and store recommendations. Someone describes a vague skin concern and gets a full routine with product suggestions. The journey doesn’t always start with a product in mind. And the brand that gets recommended in that conversation has no way of knowing it happened.
On the other path, people don’t even think about the fact that AI is involved. Google’s AI Overviews appear at the top of a search page before any organic link. Pew Research tracked browsing behavior and found that when an AI summary appeared in search results, people clicked on a link only 8% of the time, down from 15% without one. The content brands have invested in (SEO, blogs, influencer partnerships, product pages) still matters, because AI pulls from those sources to generate its answers. But the consumer reads the AI’s synthesis instead of your original content. They get the recommendation without ever visiting your site, seeing your packaging, or experiencing your brand the way you designed it.
There’s a tree-falls-in-the-forest quality to all of this. Whether a customer found you through ChatGPT or through a Google AI Overview, if they never visit your website, never see your ad, never click your link, do you know it happened? Can you learn from it?
Research: People outsource the analysis. They keep the experience.
Over 80% of AI shoppers say it’s most effective for researching and comparing products. AI is genuinely excellent at the analytical side of evaluating options: processing specs, weighing trade-offs, matching criteria. In auto, 25% of new car buyers used AI during their research and reported higher satisfaction than non-AI buyers. In CPG, people are using AI to compare ingredients, find products that meet dietary needs, and build meal plans, sometimes buying directly through integrations like Instacart’s ChatGPT app. When the task is analytical, AI does it well and people are happy to let it.
But not all research is analytical. A parent browsing a toy store with their toddler isn’t trying to be efficient. A home cook wandering a farmers market isn’t optimizing for the best ingredient, they’re looking for inspiration. Those are emotional experiences that happen to look like research. They’re the parts people want to keep for themselves. Contentsquare’s data reflects this: AI guidance is most preferred in electronics (55%) and travel logistics (38%), and drops significantly in categories where the process is more personal, sensory, and tied to identity.
What’s interesting is that both can exist in the same purchase. A runner might happily let AI narrow the field. “Best running shoe for flat feet under $150” is exactly the kind of concrete, criteria-driven question AI handles well. That same runner still wants to go to the shop, try on three pairs, and feel how they land on a short jog around the block. The initial filtering was cognitive. The next level of filtering was emotional. Same purchase, two very different needs.
Validation: AI made research faster. What comes next takes longer.
This is the part that surprised me. You’d expect incorporating AI into the journey to make it shorter. Fewer steps, faster decisions. The opposite is happening.
Before AI, the average shopper took 1.6 online steps before making a purchase. Now, shoppers who use AI take 3.8 steps before buying. (IAB/Talk Shoppe, 2025)
People have always read reviews and asked friends before buying. What’s changed is the sequence. AI can significantly compress the research phase, handing you an initial shortlist in seconds. Then a distinct validation phase opens up. 95% of consumers take at least one additional action after an AI shopping session before they feel confident enough to commit.
Just like research, validation has an analytical side and an emotional side.
AI plays a role here too. BCG found that the most frequently cited benefit of using AI for shopping was confidence: feeling sure you’ve considered the right options and haven’t missed anything. Envision Horizons found that 51% of shoppers abandoned a purchase after AI flagged a concern about a brand, whether that was questionable claims, thin content, or a lack of credible information. AI can confirm your shortlist or dismantle it. It validates the analysis.
Then there’s a different kind of validation people still need. Someone shopping for a protein powder might let AI narrow the field based on macros, price, and dietary restrictions, then still want to read Reddit threads about how it actually tastes, check the ingredient label themselves, or ask a gym buddy what they use. The decision to put something in your body requires trust that data alone can’t provide.
A parent might let AI compare car seat safety ratings, then need to hear from another parent that it actually fits in a Honda Civic backseat and is easy to install at 6am. The data checked out. The feeling of “this is the right choice for my family” comes from talking to someone who’s been there.
Validation is growing precisely because AI makes research so fast. When you skip the slow process of figuring it out yourself, you need to build confidence another way. The conviction that used to come from doing the work now has to come from somewhere else.
Purchase: The moment people want to own
You might expect that once AI has done the research, narrowed the options, and even helped build confidence, the logical next step would be to let it finish the job. Place the order. Complete the transaction.
Almost nobody wants that.
Only 8% of US shoppers are fully comfortable with AI completing a purchase on their behalf. Nearly three-quarters want some form of restriction on what AI can do at checkout. Consumers want AI to help them find information, compare prices, and narrow choices, while keeping final decision-making control for themselves.
The easy explanation is security. People don’t want to hand over their credit card to a chatbot. That’s part of it. But the research suggests something deeper. Even in complex scenarios where AI could genuinely reduce effort and improve outcomes, consumers’ desire to retain autonomy outweighs the convenience. The researchers concluded that we don’t want AI to take over purchasing because our autonomy matters. It allows us to express our values and shape our identity.
The act of choosing isn’t just a step in the funnel. It’s an act of agency, how people express choice, identity, preference. The parent who picks the car seat is saying something about what kind of parent they are. The runner who chooses the shoe is expressing something about the kind of runner they want to be. The person who fills their grocery cart is making dozens of small identity statements, whether they realize it or not.
The research can be outsourced. The decision can’t.
The bigger question
Planning that Japan trip, I didn’t have language for what I was doing. I just knew which parts I wanted help with and which parts I wanted to do myself. Looking at the data, people are making the same split everywhere: handing AI the analytical work and keeping the parts that feel personal. The reasons differ depending on where they are in the journey, but the instinct is consistent. What stays constant is that at every phase, there are moments people want to keep for themselves.
Behavioral research supports what the data keeps showing: people welcome AI for functional decisions, comparing features, finding the best price, narrowing options, and resist it for experiential ones, where taste, emotion, and personal judgment matter more. The more efficiently AI handles the analytical work, the more weight the human moments carry.
Where those human moments sit is different for every category and every audience. A first-time parent researching strollers has a different balance than a runner choosing shoes or a household buying protein powder. As the running shoe example shows, the split doesn’t just happen between categories. It can happen inside a single purchase. The same person who lets AI do the filtering still wants to try on the shoes themselves.
If your last journey map was built more than a couple of years ago, it was built before AI reshaped how people discover, research, and build the confidence to commit. The journey has changed. As we explored in Building a Customer Journey That You’ll Use, the journey map only matters if it reflects what’s actually happening.
We suspect the brands that win won’t be the ones with the best dashboards, but the ones that recognize where the ‘human moment’ truly lives…
Want to explore how AI has reshaped the journey for your customers? Let’s talk.
Written By:
Clea Stone Monsurate
Clea has always been fascinated by human behavior, especially our relationship with technology. For her college thesis, she produced a documentary video installation about Death on Facebook. At The Sound, she leverages her background in marketing, social psychology, and film to bring consumers to life and develop actionable insights for her clients in diverse industries all over the globe.
More by this author
