Benchmark insights: Why your happiest sessions might be from frustrated customers.
Sep 28, 2026
A longtime MasOrange customer visits a product page. No errors. No slow loads. No rage clicks.
Then they come back. And come back again. Three visits, and still no purchase.
On paper, it's a healthy session. In reality, something is clearly wrong. The traditional next step would be to start guessing. Is it the layout? The pricing display? The call to action?
As David Doogs, who leads go-to-market for Quantum Metric's Voice of Customer solution, put it: "You spin those assumptions up into three or four experiments, run them for a month, and if you're lucky, one of them was the right hit."
MasOrange didn't guess. They asked.
As María explains, the answer came back immediately. The customer wanted the new iPhone, but the pricing options they were seeing didn't work for them.
That's not a bug. It's not a broken funnel. It's a missing option only the customer could name.
Yet Quantum Metric's new Customer Voice Benchmark shows while brands are investing more in feedback channels than ever, half of consumers have stopped giving feedback altogether because they don't believe it matters.
That's the customer listening gap. Closing it starts with rethinking how customer listening can be closely tied to immediate action.
Digital teams are measuring the experience, not the customer's reality.
Digital teams have long treated error rates as the clearest signal an experience needs fixing. By that measure, 2026 should look like a win.
It doesn't.
Our aggregated platform findings show:
- Error rates did decline 23% between 2025 and 2026
- But, frustration rates grew 77% year-over-year.
This was a surprise, because it breaks the traditional patterns of how digital experiences are optimized. Brands are fixing what they think is broken, and customers are more frustrated anyway.
This is another illustration of that customer listening gap: the widening space between how much brands invest in understanding customers and how heard those customers actually feel. Behavioral data is exceptional at showing what happened, but not how a customer feels. When the data looks fine but the customer isn't, teams either miss it entirely or spend hours combing through session replays to reverse-engineer a feeling.
AI is making the gap harder to see.
Now, AI traffic does something interesting. While frustration climbed across traditional traffic, AI-referred traffic moved in the opposite direction:
- Frustration rates for AI traffic dropped 32%.
- AI sessions saw half as many page views as traditional traffic, despite similar abandonment and engaged time.
It's tempting to read that as AI shoppers being happier. They aren't.
They arrive knowing exactly what they want, take it, and leave. They have more options than ever and less loyalty to any single brand. When something frustrates them, they don't stay long enough for it to show up in your metrics. They just don't come back.
Now imagine sending them a survey asking generally about their experience. How likely do you think you’ll receive a response?
To an AI customer, who's not really loyal to your brand, anything that doesn’t feel like it will support their experience, won’t be worth their time.
The solution to the listening gap isn’t more surveys.
When behavior can't explain a problem, the instinct is to ask more questions. That’s the wrong instinct here. And it's not just surveys. Consumers are losing faith in feedback channels altogether. In fact, 1 in 10 saying no channel (survey, chat, phone, or review) actually gets a brand to listen.
Part of the problem is that traditional programs ask customers to repeat what brands already know.
"You don't need to survey people on what you should already know," said Mark Treschl, GM of VoC at Quantum Metric. "You get a random sample, it's six or seven questions. The brand should already know who you are, what you've done, and what you're likely to do next."
The survey should then be tied to action. Brands can send every survey possible and make the greatest investments in those channels, but if they have no way to tie it to action, they’ve failed.
- Fewer than 20% want a survey when something goes wrong. They want support.
- 51% expect acknowledgement of their feedback.
- 50% want visible proof their issue was fixed before they'll keep trusting the brand.
Trust is a major point to call out here. It shows that customer feedback is no longer just about sentiment or satisfaction, it determines how likely that customer will come back in the future. If they’ll browse upsells or just come for what they need. If they’ll suggest your brand to others.
Trust matters, and a single NPS score can't tell you why it's rising or falling.
Listening has to change what happens next.
The brands getting this right treat feedback as a step in their customer experience strategy.
Go back to MasOrange. They didn't just ask the right question at the right moment. When a customer left a low rating after an otherwise error-free experience, MasOrange routed them straight to support.
Coach applied the same thinking to product releases. As new site features roll out, they drop in micro-surveys using what they call the Netflix model: a simple thumbs up or thumbs down, right in the flow. That signal goes directly to the product team, who can adjust in days instead of waiting weeks for a survey cycle.
The new job is connecting the signals.
Customer feedback comes in three forms:
- Direct: what customers tell you when you ask.
- Indirect: what they say about you elsewhere, like Reddit, app reviews, and social.
- Inferred: what their behavior reveals without them saying a word.
Most brands run these as separate programs owned by separate teams. That's why an NPS score can drop two points and no one knows what that means to the business.
The solution? Connect each of these signals within one source of truth and use every piece of experience context to take immediate action. So, what does that mean for brands?
- Stop asking what you already know. Use behavioral context to shorten surveys and target customers in the moment.
- Pair low scores with support. When feedback signals a problem, route customers to help.
- Test with the customer, not just the metric. Add lightweight feedback to feature releases to learn why, not just what.
- Show customers you heard them. Acknowledge feedback and make fixes visible.
Want to see how brands are putting this into practice? Watch our Voice of Customer launch event for more customer stories and an in-depth demo: https://www.quantummetric.com/events/quantum-metric-voc-launch-2026.