The Six Laws of Customer Experience: Lessons for Humans and AI Agents
Customer experience in Europe operates under visible constraints.
Regulation is tighter. Trust is harder won. Customers are less forgiving of hype and more sensitive to fairness, transparency, and tone. These conditions make the Six Laws of Customer Experience easier to observe in practice.
They reveal themselves not in strategy decks, but in how organisations behave under pressure.
This article serves as my take on a modern-day reminder to Bruce Temkin’s The Six Laws of Customer Experience, first introduced in 2008 and updated in 2010 on his Customer Experience Matters blog and later compiled into a short ebook that I printed, dog-eared, and kept as a touchstone for years. These six laws endure not because they describe best practice, but because they describe gravity, not aspiration. You can disagree with gravity, but you still fall. I still have the original 2028 version printed out sitting in a plastic wallet on my bookshelf and I’ve used this copy as my base for this article.
Law 1: Every Interaction Creates a Personal Reaction
Human truth: Experience is subjective.
System truth: Subjectivity still matters, even when automated.
Consider disruption in European air travel.
When British Airways cancels flights, its rebooking systems via the app prioritises operational efficiency. For some passengers, proactive rebooking feels helpful. For others, it creates stress when the new flight breaks personal constraints such as childcare, rail connections, or work obligations.
The system solved a logistical problem.
It did not account for personal consequence.
Low-cost European carriers often automate this without meaningful human override, which pushes emotional fallout into complaints and social media escalation.
Practical lesson:
AI-driven recovery flows must include escape routes. Speed without choice creates resentment, not relief.
Law 2: People Are Instinctively Self-Centred
Human truth: Customers care about outcomes, not internal logic.
System truth: Systems mirror internal logic unless actively corrected.
European telecoms provide a familiar example.
Customers contacting support often encounter chatbots or Interactive Voice Response (IVRs) that reflect billing systems, contract types, and product silos. The customer’s issue is singular. The organisation’s response is fragmented.
Contrast this with Monzo.
Monzo structures support around customer intent rather than internal departments. Whether an issue is technical, financial, or behavioural, the experience feels owned end to end.
Practical lesson:
If your AI agent explains how your organisation works, it is serving the company, not the customer.
Law 3: Customer Familiarity Breeds Alignment
Human truth: Shared understanding reduces conflict.
System truth: Shared definitions reduce model drift.
European ecommerce shows this clearly.
At companies like Zalando, customer insight is used to align merchandising, logistics, and personalisation. Recommendation systems and promotional logic operate from a shared understanding of customer value and intent.
Where this breaks down, AI agents optimise locally. Marketing pushes offers that logistics cannot support. Personalisation engines optimise clicks while returns spike.
Practical lesson:
Customer insight must shape how systems are trained, not just how teams report.
Alignment is informational, not hierarchical.
Law 4: Unengaged Employees Don’t Create Engaged Customers
Human truth: Belief precedes behaviour.
System truth: Automation amplifies organisational confidence or doubt.
Public sector digital services in the UK illustrate this well.
GOV.UK redesigned services by empowering multidisciplinary teams with clear authority and purpose. Technology followed capability, not the other way around.
The result is not flashy. It is trusted.
Contrast this with organisations that deploy AI chatbots to deflect demand without giving employees authority to resolve underlying issues. The agent becomes a barrier, not a bridge.
Practical lesson:
AI should extend employee capability. It should not compensate for its absence.
Law 5: Employees Do What Is Measured, Incented, and Celebrated
Human truth: Behaviour follows reward.
System truth: Optimisation follows objective functions.
European banks provide a consistent pattern.
When contact centres are measured on speed and cost reduction, both human agents and AI assistants optimise for closure, not resolution. Customers experience efficiency without care.
Wise (formerly TransferWise) the financial technology company takes a different approach. Transparency, clarity, and customer outcome metrics govern both human interactions and system design. This alignment reduces the gap between promise and experience.
Practical lesson:
If AI agents are optimised differently from humans, the customer will feel the inconsistency immediately.
Metrics are moral documents.
Law 6: You Can’t Fake It
Human truth: Customers detect insincerity.
System truth: Inconsistency surfaces faster at scale.
European consumers are particularly sceptical of overclaim.
When companies market “AI-powered personalisation” without changing service policies, customers notice the mismatch. The language feels modern. The experience feels old.
Spotify and Netflix work because its recommendation systems reflect real investment in user experience, not just messaging. Discovery improves because the organisation committed operationally before promoting it.
Practical lesson:
Do not automate values you have not operationalised.
Marketing confirms reality. It cannot replace it.
What the Six Laws Show Us in a European Context
These laws do not compete with AI adoption.
They govern whether it works.
They explain why some European organisations scale trust while others scale friction. Why AI agents feel respectful in one context and evasive in another. Why customer experience becomes more visible, not easier, as systems mature.
AI does not fix customer experience.
It reveals organisational truth.
And gravity remains consistent, whether the decision-maker is human or machine.
So based on the Six Laws of Customer Experience, how can you plan for this?
I’ve put together a simple checklist before you design, while you build, and before you scale an AI agent.
If you cannot confidently tick an item, pause. The law is already operating, whether you acknowledge it or not.
Law 1: Every Interaction Creates a Personal Reaction
Design assumption to test:
“A correct response is not the same as a good experience.”
Checklist
- ☐ Have we defined user intent, not just task completion?
- ☐ Do we capture contextual signals such as urgency, history, and constraints?
- ☐ Can the agent recognise when confidence is low and slow itself down?
- ☐ Is there a clear path for human escalation when emotional stakes are high?
- ☐ Are we measuring variance, not just success rates?
Red flags
- Optimising for average outcomes only
- Treating sentiment as cosmetic rather than causal
Practical rule:
If the agent cannot adapt, it should not pretend to personalise.
Law 2: People Are Instinctively Self-Centred
Design assumption to test:
“The user’s goal is not aligned with our internal structure.”
Checklist
- ☐ Does the agent frame responses around user outcomes, not system states?
- ☐ Can a user complete their goal without understanding internal categories?
- ☐ Are internal terms, schemas, or policy language hidden from the user?
- ☐ Does the agent own the issue end to end, even if resolution is external?
- ☐ Have we tested journeys without exposing organisational logic?
Red flags
- “I can’t help with that, but…” responses
- Explanations that reference departments, queues, or processes
Practical rule:
If the agent explains how the company works, the company has failed the design.
Law 3: Customer Familiarity Breeds Alignment
Design assumption to test:
“Shared understanding must exist before optimisation.”
Checklist
- ☐ Is there a single, agreed definition of a successful customer outcome?
- ☐ Are training signals aligned across teams and systems?
- ☐ Do humans and agents use the same customer metrics?
- ☐ Can customer evidence resolve internal disagreements?
- ☐ Is customer insight shared outside the AI team?
Red flags
- Competing KPIs feeding the same model
- Teams interpreting “success” differently
Practical rule:
If teams disagree about the customer, the agent will amplify the disagreement.
Law 4: Unengaged Employees Don’t Create Engaged Customers
Design assumption to test:
“AI will amplify our organisation, not replace it.”
Checklist
- ☐ Can employees do what the agent promises?
- ☐ Are humans empowered to override or correct the agent?
- ☐ Does the agent reduce friction for staff, not add new work?
- ☐ Have employees been trained on why the agent behaves as it does?
- ☐ Is the agent positioned as support, not deflection?
Red flags
- Agents deployed to compensate for lack of authority
- Staff blaming the system, or hiding behind it
Practical rule:
If humans cannot deliver the experience, neither can automation.
Law 5: Employees Do What Is Measured, Incented, and Celebrated
Design assumption to test:
“Agents optimise exactly what we reward.”
Checklist
- ☐ Are agent objectives aligned with customer outcomes, not just efficiency?
- ☐ Do human and agent KPIs reinforce the same behaviours?
- ☐ Have we stress-tested for performance ?
- ☐ Are negative incentives explicitly removed?
- ☐ Is quality measured as more than speed or closure?
Red flags
- Handle-time obsession
- Deflection framed as success
Practical rule:
Your metrics are your values, whether you admit it or not.
Law 6: You Can’t Fake It (My favourite part)
Design assumption to test:
“Trust cannot be simulated.”
Checklist
- ☐ Does the agent reflect real operational capability?
- ☐ Are we honest about what the agent can and cannot do?
- ☐ Is marketing language aligned with lived experience?
- ☐ Have we resisted deploying before organisational readiness?
- ☐ Would we be comfortable using this agent ourselves?
Red flags
- “AI-powered” as a substitute for change
- Polite language masking rigid policy
Practical rule:
Do not automate promises you cannot keep.
Example: AI Agent Handling a Subscription Billing Issue
Context:
A customer has been charged after cancelling a subscription. They are frustrated, time-poor, and sceptical of chatbots.
Conversation
Customer:
I cancelled my subscription last month, but I’ve been charged again. This keeps happening and it’s really annoying.
AI Agent:
Thanks for explaining. I can see why that would be frustrating, especially if you already cancelled.
Let me check what happened and focus on getting this sorted for you.
Before I do, can I confirm whether you cancelled online or through support?
What’s happening here
- Law 1: Acknowledges emotion before task resolution
- Law 6: Honest, calm tone. No fake cheer or corporate apology
Customer:
I cancelled online. I even got a confirmation email.
AI Agent:
Got it. I’m checking your account history now.
I can see the cancellation confirmation from last month. You’re right. The charge shouldn’t have gone through.
Here’s what I can do immediately:
- Reverse the charge today
- Make sure no further payments are taken
- Share what caused the error, if you want
Would you like me to go ahead with the refund now?
What’s happening here
- Law 2: Focus on the customer’s outcome, not internal process
- Law 5: Optimised for resolution, not deflection or speed alone
Customer:
Yes, please. And I’d like to know why this happened.
AI Agent:
I’ve processed the refund. You’ll see it back in your account within 3–5 working days.
As for what happened:
Your cancellation was successful, but the payment system didn’t sync correctly before the billing run. That’s on us, not you.
I’ve flagged this so it doesn’t happen again on your account.
If you want, I can also pass this to a human support specialist to review the system issue more broadly.
What’s happening here
- Law 6: Takes responsibility without hiding behind systems
- Law 3: Uses customer feedback to improve alignment, not just close the ticket
Customer:
Thanks. That explanation actually helps. No need to escalate.
AI Agent:
You’re welcome. Before you go, one quick check.
Is there anything else about your account that feels unclear or unresolved right now?
If not, I’ll step out and let you get on with your day.
What’s happening here
- Law 1: Respects emotional closure, not forced engagement
- Law 2: Centres the customer’s time and intent
Why This Conversation Works Against the Six Laws
Law 1: Every Interaction Creates a Personal Reaction
The agent treats frustration as real input, not noise. It slows down where emotion is present.
Law 2: People Are Instinctively Self-Centred
The agent never explains internal teams or systems as excuses. The problem is owned end to end.
Law 3: Customer Familiarity Breeds Alignment
The agent uses customer evidence to validate reality and improve future behaviour.
Law 4: Unengaged Employees Don’t Create Engaged Customers
The agent reflects confidence and authority that must also exist in the organisation behind it.
Law 5: Employees Do What Is Measured, Incented, and Celebrated
The interaction optimises for correct resolution and trust, not handle time or deflection.
Law 6: You Can’t Fake It
No exaggerated empathy. No “AI magic.” Just competence, honesty, and follow-through.
Nothing here is technically impressive.
That’s the point.
The agent succeeds because it reflects:
AI agents do not create good customer experience, they reveal whether one already exists