The AI paradox: better numbers, worse experiences

Is it time to introduce Key Human Indicators to your dashboards?

Let’s face it, businesses are obsessed with their KPIs. Every dashboard’s jammed with metrics that the whole company lives or dies by - despite their obvious limitations.

Now AI has arrived and fundamentally changed how digital experiences are designed, forcing us to rethink how we measure success in the first place.

Why? While everyone’s obsessing about KPIs, we're losing something fundamental: our humanity.

That’s where Key Human Indicators (KHIs) come in - not as a replacement, but as an additional level of insight. Instead of just tracking what people do, KHIs try to measure how people actually feel: trust, empathy, whether customers feel safe and respected by your brand.

So, why do KHIs matter right now? Because the more we hand over to AI, the easier it is to forget about people altogether.

---

Dark Patterns: The Hidden Cost of Chasing KPIs

AI driven manipulative designs dressed up as helpful features are everywhere. They push users into decisions they’ll regret and regular KPIs don’t sound the alarm early enough to be useful.

Maybe you squeeze out a 10% uplift in conversion using these shady tricks. But what’s the real cost?

Whilst you’re celebrating that “win”, customers feel drained or manipulated, leading to silent churn, bad press or regulators start asking questions.

David Mannheim nailed it on the From A to B podcast, claiming these hacks often backfire, damaging trust and creating what Mannheim calls “disimpact”.

Here’s a few examples of how this manifests (I'd be surprised if you haven't experienced at least one of these):

Guilt-Trip Copy / Confirm-shaming: Instead of a simple “Cancel subscription,” you get options like “Yes, I want to save money” or “No, I prefer to waste money”.

A friend of mine shared an example while changing grief counsellors. On the account page, the ‘Delete account’ button was labelled "Quit Therapy" which can be pretty triggering for vulnerable users.

The “Roach Motel” Trap: Easy to enter, hard to leave. Amazon Prime-style cancellation flows use friction to suppress churn. AI scales it. Customers feel trapped. Trust erodes, leading to the ‘disimpact’ Mannheim talks about.

Call-to-Cancel: Ever sign up online, but have to call someone to cancel? The New York Times is famous for this. One click to join, but leaving means endless hold music and pushy reps. 

None of this is by accident.

In the age of AI-driven design, companies test UX flows and messages relentlessly to keep people hooked. But they rarely stop to ask if they’re actually hurting trust along the way.

The best companies aim for real win-wins instead of tricking people with shady tactics - that’s how you build something that lasts.

---

Building KHIs: How to Track What Your Customers Really Feel

KHIs aren’t here to replace KPIs. They’re about keeping your customer front and centre by understanding the signals that actually tell you how your customers feel at any given time.

One of the best ways to uncover those hidden emotions? It's actually a fairly established tool in the UX toolkit: Empathy maps.

Take a four-quadrant map (Says, Thinks, Does, Feels) and walk through the customer journey from their perspective, especially in high-stakes moments like cancellation.

Let's take the “Quit Therapy” example from earlier. An empathy map might reveal:

Feels: triggered, guilty, misunderstood

Thinks: “This wording is cold…do they even care?”

Does: reluctantly stays, cancels but feels shitty, vents online

Says: “what shitty language to use when I'm feeling vulnerable” 

Suddenly, those emotional pain points become visible and actionable - and form the basis for your KHIs.

Codifying these insights into something measurable is key. This is something AI is excellent at doing… if we give it the right inputs.

For example:

Sentiment Analysis: Use Natural Language Processing to detect frustration, guilt, or relief in reviews and social posts. Watch for sudden drops after manipulative flows.

Quick Emotional Surveys: Post-interaction, ask, “Did you feel supported, pressured, or triggered?” Track an “Emotional Safety Score.”

Behavioral Signals: Keep an eye on re-engagement, referrals and long term churn. Sometimes, short-term wins hide long-term losses.

Feedback Loops: Tag support calls and chats for an “Empathy Perception Index.” Do people feel heard or trapped?

---

TL;DR

If you only obsess over KPIs, you end up with experiences that work on paper but feel soulless or manipulative. Bring in KHIs and empathy-driven tools, and you’re building a brand people actually care about.

Start measuring what actually makes customers feel seen and valued, not just what keeps them in the short term. 

So, what KHIs or empathy map insights would you track? Drop your thoughts below - I’d love to hear them.