The best way to convert a customer is to understand them

I've sat through hundreds of user interviews in my career. They can be incredibly enlightening but frustrating experiences for a product or business leader. Enlightening for obvious reasons. You get out of your bubble and get a much deeper understanding of how customers tick and what they care about. Frustrating because with each interview, as you understand the customer, you desperately want to jump in and explain to them how the product could benefit them based on your newfound understanding. But you don't, because 1) that wouldn't make for an unbiased research study, and 2) it wouldn't really make a difference when you have hundreds of thousands or millions of customers.

I remember one interview in particular where a customer spent ten minutes explaining a workaround they had built for a problem our product already solved. They just hadn't found the feature. Every part of me wanted to stop the interview and show them. That moment stuck with me, because it captured the whole problem. We understood exactly what that customer needed, and likely how many others were out there needing similar guidance in individual ways.

The best sellers of luxury or B2B products and services take a similar approach to those user interviews, with one important difference: they get to act on what they learn. They look to deeply understand the prospect or customer before discussing anything about what they're offering or what they should buy. This isn't just intuition. When Neil Rackham and his team studied more than 35,000 sales calls for what became SPIN Selling, they found that the most successful salespeople in larger sales spent their time asking questions about the customer's situation, problems, and needs, rather than pitching features and pushing for a close.¹ The pitch comes later, and when it does, it's built around what the customer actually said.

However, most businesses take the opposite approach out of necessity. They guess at what a customer needs, what is holding them back, or what questions they have. They do this because, due to their scale, unit economics, or both, they can't afford to understand each individual customer. When you have millions of customers, you can't put an expert seller in front of every one of them.

This leads to behavior on the business side like broad discount offers which don't distinguish between someone ready to buy and someone hesitating due to price. It's worth remembering that people hesitate for very different reasons. When the Baymard Institute looked at why shoppers abandon their carts, which happens about 70% of the time, the answers ranged from unexpected extra costs, to having to create an account, to slow delivery, to not trusting the site with their credit card information.² A discount only answers one of those. For everyone else, it's money given away without solving the actual problem.

Discounting broadly also has a cost that is easy to miss. Research by Eric Anderson and Duncan Simester, based on three large field experiments, found that deeper discounts increased future purchases from first-time customers but actually reduced future purchases from established customers. In other words, the same discount that wins over one customer can quietly teach another to wait.³

It also leads to "best guesses" on messaging, like promoting privacy and security, which may ease some users' concerns while creating concerns for others who hadn't considered security being a problem. Every broad message is a bet that most people share the same hesitation, and every time that bet is wrong, it either misses the point or creates a new worry that wasn't there before.

For consumers, this has trained some unfortunate habits. In a 2026 survey, 51% of Americans said they intentionally abandon their online shopping carts in hopes that the retailer will send them a discount code later.⁴ At the same time, the average person has become blind to marketing and onboarding emails, or even automatically archives them. In one survey, more than half of consumers said they delete at least half of the marketing emails they receive without ever opening them.⁵ Businesses send more messages to make up for not understanding the customer, and customers tune out more of them in response.

For years, the best alternative to deeply understanding the individual customer has been various forms of behavioral tracking. If we know the customer clicked on this ad or visited this page, we know a little bit about that customer and can serve them in a more personalized way. But for the consumer, this feels like spying. According to Pew Research Center, 81% of Americans are concerned about how companies use the data they collect about them, and 73% feel they have little or no control over it.⁶

For the business, tracking isn't an amazing approximation anyway. You can only glean so much from small snapshots of what a person does. A customer who visits the pricing page three times might be close to buying, or might be confused, or might be comparing you to a competitor. The data tells you what they did, but not why. And yet customers do want to be understood. McKinsey found that 71% of consumers expect companies to deliver personalized interactions, and 76% get frustrated when that doesn't happen.⁷ People aren't against being understood. They're against being watched.

The opportunity now exists to truly understand each customer individually in a much more transparent and direct way. Instead of tracking where people go and what they do, AI enables you to have substantive dialogues with customers at scale. Similar to an expert seller, it can dig into what they care about, what their concerns are, and what their goals are, all toward the end of giving them a better experience that's tailored to them and their needs. And because the customer is the one telling you, it doesn't feel like spying. It feels like being listened to.

However, AI will also enable companies to take this a step further, in that, unlike an expert seller, an AI can be an expert in much more than selling. When the time calls for it, and there is nuance worth understanding, the AI can be an expert researcher. When the customer is lost and needs guidance, the AI can be an expert educator. When the customer is looking for a deal, the AI can become an expert negotiator. Very few people are great at all three, and even fewer can switch between them in the middle of a conversation.

Picture three people who sign up for the same budgeting app on the same afternoon, and all three stop at the screen asking them to connect a bank account. On a dashboard, they look identical. Same step, same drop-off, same segment. Most companies would send all three the same follow-up, and there's a good chance it would lean on "bank-level security," since that's the most common worry at that step.

Ask each of them why, though, and you get three different people. The first, a recent college graduate, says it honestly felt sketchy to hand a new app her bank login. The second, who manages the household budget with his wife, never thought twice about security. He just couldn't tell whether the app would work for two people sharing money. The third, a freelancer, likes the app but isn't convinced it's worth $8 a month when a spreadsheet is free.

Each of those answers calls for a different expert. The graduate needs an educator, someone who can explain plainly that the connection is read-only and the app never sees her password. The husband needs a researcher first, someone curious enough to ask how he and his wife split their spending today, before pointing him to shared accounts. And for him, the security email wouldn't just have missed the mark. It might have introduced a worry he didn't have. The freelancer needs something closer to a negotiator, someone who can weigh whether a longer trial or an annual plan makes sense for him. One conversation can play all three roles, and it can tell which one the moment calls for, because the customer just said so.

There's a fourth thing that comes out of those conversations, and it doesn't show up for any of the three customers. The fact that couples who share money are getting stuck at that screen is a product insight, not a conversion problem. No amount of tracking would have surfaced it.

While dedicated research conversations and dedicated sales calls will always exist, we believe AI will make the hybrid of the two more common. Instead of a research call where the person with the answer is sitting on the sidelines, itching to tell the interviewee the answer, an AI agent will be able to seamlessly switch between "research mode," where it's trying to maximize understanding, and "selling mode," when it sees an opportunity to correct a misconception or provide some education.

This only works if it's done honestly. Researchers have long warned against "selling under the guise of research," and for good reason. If customers feel like a question is just a setup for a pitch, they'll stop answering. The goal isn't to disguise selling as research. It's to stop forcing a false choice between the two. The agent should be open about who it represents, genuinely interested in the answer, and only move into helping when the customer would actually benefit from it. A customer who just wants to share feedback should be able to do exactly that, without a pitch at the end.

When it's done well, this gives the best of both worlds. In the worst case, you learn from every customer. You find out why they hesitated, what they expected, and what would have changed their mind, and those reasons become the themes your team works on next. In the best case, that learning drives a conversion, because the customer got a real answer to their real question at the moment it mattered.