When compliance kills CX: How to automate with judgement

I recently shared a LinkedIn post about a AU$0.02 refund email I received from my bank. Within 48 hours it had generated 78,000 impressions, 66 comments and one very clear signal: this struck a nerve.

Not because the amount was embarrassing. But because almost everyone who read it had their own version of the story. A $0.00 invoice. A compliance notice that cost more to open than act on. A triggered communication that arrived at exactly the wrong moment in the customer journey.

The $0.02 email wasn't an outlier. It was a symptom.

First, the compliance defenders have a point

When the post went live, one of the most liked responses came from a senior leader with financial services experience. The argument was straightforward: legally, organizations are required to notify all customers of errors regardless of amount. The effort to not send the email would likely exceed the effort to send it.

They're right. And that's exactly the problem.

When compliance logic becomes the ceiling of customer experience thinking, organizations stop asking whether they should and only ask whether they must. The process gets optimized. The customer gets forgotten.

Don't miss any news, updates or insider tips from CX Network by getting them delivered to your inbox. Sign up to our newsletter and join our community of experts. 

Learn More

Compliance is non-negotiable. How you execute it isn't.

It's not an AI problem. It's a decision logic problem.

Several commenters were quick to point out that this wasn't AI at all: it was rules-based audience selection. A filter, a trigger, a send.

Also correct. But also, beside the point.

Whether the system is powered by machine learning or a basic if/then rule, the failure is identical: no human judgement was applied to the question of customer impact before the communication went out.

Automation should make good decisions faster. Not make bad decisions at scale.
The technology isn't the issue. The absence of a customer lens in the decision architecture is.

That gap becomes harder to ignore as AI takes over more decisions. KPMG's 2025-26 Customer Experience Excellence research defines agentic AI systems as those that can "independently sense, reason, and act, dynamically assembling the right capabilities to deliver the right outcome at the right moment." 

When systems are making decisions at that speed and scale, a customer impact filter isn't a nice-to-have in the design. It is the design.

The contrast that tells the whole story

Working recently with a loyalty program, one of the most successful automations we built was a simple triggered email: a notification sent to members the moment their reward was paid into their account.

Same principle. Same technology. Completely different outcome.

Why did it work? Because it passed the only test that matters: it delivered genuine value to the customer at exactly the right moment in their journey. The automation served the relationship, not just the process.

A two-cent compliance notice does the opposite. Same channel. Same trigger logic. But zero customer value and a quiet erosion of trust. Across lifecycle and loyalty work at Medibank, Wesfarmers and Flybuys, the difference between automations that strengthen relationships and those that quietly degrade them comes down to a single question: was customer impact assessed before the trigger fired?

The four-step Customer Judgement Test

A $0.02 credit applied silently communicates competence. An email communicating a $0.02 refund communicates process. Before any automated communication enters production, it should pass four questions:

1. Is it required? Legal, regulatory or contractual obligation. If yes, it must send: but how it sends is still a design decision.
2. Is it valuable? Does it deliver something the customer would thank you for? Information, a saving, a reward, a resolution. If not, reconsider.
3. Is this the best channel? Email is not the default. Match the weight of the message to the weight of the channel.
4. Is this the best moment? Where is this customer in their journey? Is this communication additive or disruptive right now?

If a communication can't pass questions two, three and four and is only sent because it passes question one, you have a compliance execution problem, not a customer experience program.

Automating with judgement

The answer isn't less automation. Organizations that remove automation in the name of CX don't scale. The answer is building customer judgement into the decision logic before the trigger fires.

That means:

Minimum value thresholds: If the communication delivers less value than the cost of receiving it, it shouldn't send: regardless of whether it technically can.

Compliance plus experience review: Every automated communication should pass both a legal filter and a customer experience filter before it enters production. The ACTIVATE Framework, a model I developed for AI-enabled CX transformation, treats this as a mandatory triage step, not an optional one. One filter without the other is incomplete.

Feedback loops: If customers are ignoring, deleting or unsubscribing from a triggered communication, that's signal. Build the mechanism to read it.

The commercial consequence nobody is measuring

The cost of poor communication decisions doesn't appear in a compliance audit. It shows up six months later: in NPS scores, in churn rates, in disengaged members who stopped opening emails long before they stopped being customers. Trust erodes incrementally. And when you do have something genuinely valuable to say, the inbox is already compromised.

The real cost of the $0.02 email

One detail stood out in the responses: the email had no reply address. To raise a question about the refund, you had to call. It cost more to respond than it did to receive.

That's the metric that matters. Not the open rate. Not the deliverability score. The felt complexity of being a customer.

Every communication builds or erodes the customer relationship. The $0.02 email costs more than $0.02 to send. And it costs far more than that in trust.

In an automated world, the brands customers keep choosing are the ones that put customer judgement inside every decision they automate. Not just the decisions that are easy to govern.

Quick links 

• Fragmented organizational cognition is why CX programs stall
• What high-performing companies do to improve CX
• Cutting costs keeping customers: How CX can achieve the impossible

CX HORIZONS: The state of CX in 2026

Download our report to learn the results of CX Network's research into the top CX trends,  challenges, spending plans, and customer behaviors driving CX in 2026. Hear from thought leaders on what you should be doing now to ensure your organization is on track to become a strategic growth engine.

Download Now