AI customer service: 4 lessons on what your customers actually want

Is your AI customer service a bridge or a barrier? 

In our recent customer experience research, we found that 73% of consumers say a negative customer support experience defines your brand and the same amount say support quality is just as important as product quality. If your automation misses the mark, you're gambling with your reputation, and you'll never know it without honest feedback. 

To take the guesswork out of your strategy, we’re sharing data-backed principles for sharpening your AI journey and putting your customers first. 

AI customer service isn’t going anywhere, but it’s telling that four in five Americans strongly prefer getting help from a human over an AI agent. In fact, they overwhelmingly want human support for issues like: 

This goes to show that AI might be perfect for things like quick password resets or order tracking, but it’s a potential liability when a customer needs help beyond the basics. 

Take a look at your service journey and flag the high-stress moments—like billing errors or security concerns. These are touchpoints where empathy is non-negotiable. If a customer hits one of these, make sure there’s a clear, intentional path to a real person.

Since nearly a quarter (23%) of people lose trust the moment they feel trapped in an AI loop, you need to watch these high-stakes moments closely. Otherwise, your AI tool might become a fast track to losing customers.

Don’t assume you already know which customers will love your AI and which will avoid it. Women are significantly more likely than men to strongly prefer humans over AI (70% vs. 60%). And while Gen Z is the generation most open to AI agents, their preference still tops out at just 14%. This means that even for digital natives, bots aren’t the default they're looking for. 

Preference for AI over human agents, by generation:

Just as you segment by purchase history or location for marketing, you should be segmenting your CSAT or NPS© data to see how different groups experience your AI, especially because different demographics likely have different breaking points. 

For instance, while a younger customer might shrug off an unhelpful bot, our data shows that Boomers are significantly more likely to be fazed if a customer support experience goes sideways (81% vs. 59%). Without looking at specific groups, you won't know which customers are embracing your automation and which ones are one tech hiccup away from leaving.

While only 8% of people currently prefer AI customer service agents over a human, the reasons they do are clear: availability (41%), speed (37%), and accuracy (30%). If you’re going to use AI, these are the three areas where you have to really deliver. 

Think of this as a stress test for your tech. If your bot requires five interactions to understand a simple request, it’s failing the speed test. If it provides a generic link instead of a specific answer, it’s failing on accuracy.

The goal is to make sure your AI is a genuine shortcut, not a digital gatekeeper. Don't just settle for functional—use follow-up surveys to ask specifically about these three pillars.

Brand trust is fragile, and it’s easy for automation to chip away at it if you aren't being transparent. Our Q4 2025 AI-sentiment study shows that 14% of consumers would lose trust if an AI agent didn’t clearly identify itself as a bot, while another 11% would feel the same if they received generic, scripted answers.

The stakes go even higher when it comes to privacy: 38% of Americans say an AI assistant storing or sharing their personal data without consent is a deal-breaker. 

To protect your brand, you have to be upfront from the very first interaction. Your AI should provide clear information about its capabilities and set honest expectations for what it can actually solve. 

Privacy is where most customers draw the line. As we’ve seen in recent headlines, people are rightfully wary of where their information ends up. Make your privacy prompts clear and easy to find—customers should never feel like their personal info is being vacuumed up in the background for reasons they don’t understand.

Automation is a massive opportunity, but it shouldn't be a shield between you and your customers. Over half (56%) of consumers have had a bad experience with a company’s new AI features, but that doesn’t have to be your story.

The trick is remembering that AI isn’t a set-it-and-forget-it tool. Keep the feedback loop open so you can build something your customers actually value. 

Net Promoter, Net Promoter Score, and NPS are trademarks of Satmetrix Systems, Inc., Bain & Company, Inc., and Fred Reichheld.