Your Metrics Say Everything Is Fine. That Might Be Exactly the Problem.

Your metrics say everything is fine. And that might be exactly the problem.

Dr. Vanja Ljevar has a claim that should make every marketing team a little uncomfortable: most dashboards are lying about customer loyalty. Not because the data is wrong. Because the questions behind the data are wrong.

She's spent over a decade as a global consultant. Now Co-Founder and Chief Data Scientist at Kubik Intelligence, she sees the same pattern repeat across companies and industries. A CMO opens their dashboard in the morning. NPS looks fine. Sentiment scores are positive. Retention curves are stable. Everything is operationally healthy. But emotionally, something is quietly eroding underneath.

"Customers are not angry. They are not leaving loudly."

That's the problem. The quiet exit doesn't show up on your dashboard until it's too late. By the time the metrics dip, the relationship has been failing for months. And all that time, the dashboard was telling you everything was fine.

The question you're not asking

Vanja's critique isn't that we lack metrics. It's that we have too many of the wrong ones. Metrics built for simpler, slower relationships, that haven't evolved to match how customers actually behave today.

Take Net Promoter Score. The question: "Would you recommend us?" Her version: "Do you trust us? Do you feel proud choosing this brand?" These are not the same question. One measures a behavior. The other measures a relationship. And for brands that want lasting loyalty, the relationship is what actually matters.

That insight is what led her to build the Emotional Market Behavior Index, the EMBI.

What the EMBI actually measures

The EMBI isn't another dashboard metric. It's a framework for reading the signals that hide inside customer language: specifically in reviews, comments, and feedback that brands already have access to but rarely interrogate at this level.

Vanja gives a case that stuck with me. Her team analyzed a luxury brand and found that overall sentiment was positive. On the surface, good news. But when they dug deeper, the dominant emotion wasn't delight or trust. It was relief. Relief that the product arrived on time. Relief that the stress of uncertainty resolved.

Positive score. Wrong emotion. And an entirely different foundation to build loyalty on.

The fix, in this specific case, was straightforward: more frequent delivery updates. Reduce the anxiety. Convert relief into trust. A small operational change with a real psychological impact.

"These are very small nuances. But these nuances mean life or death long term for brands."

She also dissected Will Smith's Oscar apology as an illustration. He wrote "I want to apologize," not "I am sorry." Subtle, but not trivial. One is a statement of desire. The other is a statement of accountability. Customers pick up on these differences, often unconsciously. The EMBI was built to surface exactly these kinds of signals at scale.

The data you already have

The data Vanja uses isn't proprietary or expensive. It's customer reviews. The kind sitting in your CRM, your Google reviews, your app store ratings. The unlock isn't more data. It's better questions, and a trained eye for what the language is actually revealing.

The AI problem nobody's saying out loud

The last part of this conversation was the most thought-provoking.

Vanja has held a consistent position on generative AI in brand communication since before it became the topic of every conference. She hasn't changed it. AI is a powerful tool. But leaning too heavily on it comes with a cost she thinks is still largely underestimated.

The academic research, she says, is confirming what many suspected: we're becoming lazy. More reliant on LLMs to generate our content. Less of ourselves, our voice, our perspective, ends up in the work.

"We risk letting our own voice sound like everybody else's."

This matters specifically in brand communication because Vanja believes the relationship between a brand and its customers is psychologically similar to a relationship between two people. When that voice gets averaged out, processed through the same models as every other brand, the trust-building mechanism weakens in ways that won't show on any dashboard.

And then she said something worth sitting with: "I now consciously introduce grammatical errors in my writing, to signal this is written by a human." It sounds strange. Maybe a bit sad. But it's rational. When polished, error-free prose starts to read as a signal of AI authorship, imperfection becomes a credibility marker.

The bottom line

Loyalty doesn't always erode loudly. It erodes quietly, underneath metrics that look fine, through a relationship that's slowly losing warmth without anyone noticing until it's gone.

"We don't see on dashboards that customers are not angry. They are just not choosing us anymore."

The brands that learn to read both data science and behavioral psychology will have an edge that doesn't show up in any competitive analysis.

Find Vanja on LinkedIn: linkedin.com/in/vanjaljevar