Turning customer conversations from a cost center into an invaluable resource

Before commerce moved online and grew to massive scale, businesses engaged with customers in personal ways almost by default. The shopkeeper knew your name, noticed when you picked something up and put it back, and asked you about it. Because businesses were small, one-to-one connection was still possible, and that connection did two things at once. It helped win the sale, and it taught the business something about what its customers actually needed.

As businesses like Amazon, Google, and Facebook grew to serve millions and then billions of people online, that mindset shifted out of necessity. When offering those services, online so cost effectively for customers, at such a massive scale, one-to-one communication becomes a non-starter. So businesses did anything in their power to avoid engaging directly with the customer in any personal way.

On the positive side of that, the user experience improved. You could do more yourself without talking to someone, and fewer people ended up confused and lost. You could change your plan, reset your password, or track your order in the middle of the night without waiting on hold for anyone.

On the negative side of that, if you did need or want help, you were sent through an endless maze of help center articles, or simply told there was no way for you to contact the company you relied on so heavily. While this wasn't great for the customer, it made great business sense. In 2019, according to Gartner, a customer contacting you through a live channel cost $8.01 per contact, whereas pushing them through self-service channels cost $0.10 per contact.¹ With a difference that large, every conversation avoided was money saved. This was the era of no-reply email addresses, endless phone trees, purposefully hidden "contact us" pages, and limited investment in customer research.

This approach was also reflected in how the success of customer-facing teams is measured: tickets deflected, cost per ticket, escalation rate, and a simple satisfaction question at the end that few customers answer. Fewer customers answer every year. Qualtrics found in late 2025 that fewer than one in three consumers give feedback at all, which is an all-time low.²

The problem is that this data never actually speaks to the "why," and it is often misleading. A deflected ticket could mean the customer's problem was solved, or it could mean the customer gave up. A three out of five on a satisfaction survey could mean almost anything. As Jeff Bezos has said, "When the data and the anecdotes disagree, the anecdotes are usually right."³ For a long time, businesses have had plenty of data and very few anecdotes.

A hidden consequence of this approach is that businesses lose touch with their customers, and any customer on the fence with questions or hesitation ends up going in a different direction. If you think about who actually needs a conversation, it usually isn't the customer who already knows what they want. It's the one who is unsure: the trial user who can't tell if the product is right for them, the shopper with one unanswered question, or the subscriber who is thinking about cancelling. These are the moments where a single good answer can change the outcome, and they are exactly the moments self-service handles worst. According to Gartner, only 14% of customer service issues are fully resolved through self-service options.⁴

Most of these customers will never tell you what went wrong. Research going back to John Goodman's TARP studies found that around half of customers with a serious problem never complain.⁵ They just leave. The base assumption from customers is now that it will be hard or impossible to get a straight answer online, so many of them don't even try. The business never finds out what happened, and the same problem quietly costs them the next customer too.

However, with AI's emergence, there is an opportunity to return to those earlier days of business, even at a massive scale. For the first time, a business can have a real one-to-one conversation with every customer, at the moment it matters, and in the customer's own words. It no longer takes a giant call center or a research team that can only get to a handful of interviews a month.

It would be easy to say this is all because AI is cheap, but that isn't the whole story. Gartner actually expects the cost of AI per resolution to rise past $3 by 2030, and it argues that leading companies will use AI "to drive customer engagement rather than to cut costs."⁶ We think that's exactly right. When a deep, personalized conversation costs a small fraction of what it used to, the question stops being "how do we avoid this conversation?" and becomes "what is this conversation worth?"

The answer is that it's worth a lot. Every conversation either wins back a customer who might have left, or teaches you something about why they almost did, and often it does both. Because of this, we believe the paradigm will flip. Companies will go from desperately trying to avoid conversations to cut costs, to desperately trying to have them to better understand customers on the path to winning them over.

It is easy to find a good analogy with B2B businesses. In B2B, almost nothing is bought without talking to someone. Experts in sales and customer success strive to engage the customer and deeply understand them and their problems on the path to converting them into a customer, keeping them as a customer, or upselling them to new products and services. There are whole departments focused on trying to get customers to engage at different points in the journey.

Why is that? Because even a quick conversation can yield incredible value in understanding the customer, which can make or break the relationship over the long term. It's also because every competitor is desperately trying to have these same conversations, and so the bar is raised. A B2B company that refused to talk to its buyers wouldn't last very long. B2C businesses never really had this option, because the economics didn't work at consumer scale. Now they do.

Here’s a quick example to make this more concrete. A customer signs up for a meal-kit trial, browses the menu, and then stops before choosing a plan. In the old approach, they get a reminder email, probably with a discount, which is the same email everyone else who stopped at that step gets.

With a conversation, things look different. An agent reaches out and asks what held them back. The customer replies that it seemed a bit pricey, and that their family has a gluten sensitivity, so they weren't sure the meals would work for them. Those are two very different hesitations, and a discount doesn't really answer either one. The agent explains how the cost compares to what the family spends on takeout, and points them to the gluten-free menu. The customer chooses a plan.

Just as important, the business learns something it never would have learned from a dashboard: some of the people who stop during the trial are families with dietary needs who can't find the right menu. That becomes something the team can fix, so the next family doesn't get stuck at all.

At Woven, we believe that B2C businesses, even with their massive scale, will shift to operate in similar ways to B2B, enabled by AI. They will become proactive in starting conversations with customers instead of hiding behind phone trees and help centers. They won't let a customer leave without trying to understand why. They will seek to understand each customer as an individual and provide them with the individualized service that best suits them and their needs.

The first businesses to successfully make this transformation will have a huge opportunity to remake their brands as being truly customer first. And then, over time, the bar will be raised similar to B2B, and this type of customer engagement will become the rule, not the exception.

Paul Graham tells founders, "You should take extraordinary measures not just to acquire users, but also to make them happy… The feedback you get from engaging directly with your earliest users will be the best you ever get."⁷ For most of the internet's history, that advice only worked while a company was small. In this new era, businesses have the opportunity to take that feedback to heart even with millions or billions of customers.

If you're wondering where to start, you don't need to change everything at once. Pick the one moment where customers most often get stuck or slip away, whether that's an abandoned trial, an unfinished onboarding, or a cancellation, and start a conversation there. Ask them why, and really listen to the answer.