So
Sonia.

At work, we have a staff recognition programme called Project Braveheart. A customer gives an unsolicited commendation saying someone delivered great service. That employee becomes a Braveheart. Their name goes on the Braveheart wall. They appear in the newsletter. They receive a gift card. Every quarter, there is a reception where the people who earned it get to stand in a room and be thanked out loud.

Staff chase it. Customers notice it. The wall is evidence that good service is seen here.

The next step, now that we are one Sagicor Caribbean, is to take it to Jamaica. This is where I want to stop and say something that sounds obvious but isn't: you cannot simply transplant a successful programme into a different market and expect the same result.

Every market has its own history, culture, leadership dynamics, and attitudes towards recognition. The customer expectations may differ. The employee experience may differ. Even something as simple as public recognition can mean something different. The programme may have the same name, but the experience of the programme should not necessarily be.

That is where change management earns its place. I completed my Prosci certification this year, and ADKAR gave me a useful frame for this specific problem.

Prosci's ADKAR Model describes five building blocks an individual needs to adopt and sustain change successfully: Awareness, Desire, Knowledge, Ability and Reinforcement. When one of those elements is missing, adoption can stall at that barrier point. For a recognition programme crossing a border, ADKAR gives us a useful question at each stage: what needs to be true for people in this market actually to adopt this programme and make it their own?

Here are five ways to get that wrong.

1. Lead with the prize instead of the reason

The temptation is to announce the rewards. Gift cards. The wall. The reception. The photographs. It is the fun part. But ADKAR Awareness is not telling people that a change is happening. It is helping them understand why the change is needed.

So before announcing the prize, answer the harder questions. What are customers currently telling us about their experience? What service behaviours are we trying to recognise and make repeatable? In a newly merged operation, employees may already be interpreting corporate announcements through the lens of what the change means for their roles and future. A recognition programme that arrives reward-first can easily be interpreted as another corporate initiative being imposed from somewhere else. Start with the customer. Explain the purpose. Then introduce the prize.

2. Assume motivation travels

Recognition is cultural. A public wall may feel like a reward in one office and a social liability in another, where being elevated above colleagues carries a cost you pay quietly for months. The value of a gift card may also mean something different in a different economic context. You cannot design any of that accurately from Trinidad. You learn it by involving the people who will actually live with the programme.

Sit with Jamaican team leaders. Talk to employees. Ask a deliberately blunt question: if we wanted to honour someone here for excellent customer service, what would that look like? The programme's spine can stay the same. The way we express it may need to change. The objective is not to make Jamaica's Braveheart look exactly like Trinidad's Braveheart. The objective is to create the same desired service behaviour in a way that employees in Jamaica can own.

3. Leave the criteria vague

If employees cannot explain how someone got onto the wall, they will supply their own explanation. A credible recognition programme needs a clear line of sight between the customer experience and the recognition. Employees should understand what qualifies as a commendation, how a nomination reaches the programme, who validates it, who makes the final decision, and most importantly: what behaviour are we actually trying to reinforce?

4. Reward behaviour the system will not allow

Ability is where change becomes real. It is not enough for employees to understand what good service looks like. They need to be able to deliver it consistently. So before celebrating a behaviour, ask whether the operating environment allows people to perform it. If we want to recognise ownership, can employees actually take ownership? If we want to reward problem-solving, do they have the authority to solve the problem?

You can tell employees to go the extra mile. But if the process, policy, system, or approval structure makes that extra mile almost impossible, the programme turns a structural problem into an employee performance problem. Do not reward people for swimming against the current and then congratulate yourself for creating a culture of service. Fix some of the current.

5. Treat launch as the finish line

Most programmes have plenty of energy at launch. There is sponsorship. There is communication. There are photographs. Everyone is excited. Then comes quarter two. The energy drops. By quarter three, nobody is quite sure who is collecting nominations, whether the programme is still active, or whether anyone is paying attention.

Reinforcement keeps change from becoming another initiative people remember fondly. That means Braveheart needs more than an annual calendar. It needs ownership. Someone local should know who is responsible for collecting nominations, validating them, communicating the results, and keeping leaders engaged. The programme should also measure more than the number of gift cards distributed. Are nominations continuing? Are leaders using the examples in coaching? Do customers still notice the behaviours we say matter?

What it buys you when it works

Every Braveheart nomination contains a real customer describing what good looked like in that moment, in the customer's own words. That gives you something training slides cannot: real evidence of the experience customers value. Customer commendations can inform coaching. Recognition can also help employees see that customer experience is not simply another metric on a dashboard.

In a newly merged organisation, the first question employees may ask is what the new organisation expects from them. A recognition programme can answer a different question: what does the new organisation value? If the first visible change is another process or reporting requirement, employees experience the merger through what is being asked of them. If one of the first visible signals is genuine recognition, they also experience it through what is being appreciated. It is one way culture gets built during change.

A programme can cross a border. Culture cannot be copied and pasted. So before your next CX programme crosses a border, ask yourself the honest version of the question: are we designing this for them, or are we simply exporting what already makes us proud?

Most organisations focus on their ranking. However, a more important question is emerging: what if you never make the shortlist?

Recently, I developed a straightforward audit tool to assess how AI systems respond to a common customer question: “What is the best car insurance company in Trinidad and Tobago?”

I posed the same question to Claude four times and compared the responses against the full list of fifteen licensed motor insurers in the market. This approach reflects how these systems typically operate, whether a customer is using Claude, ChatGPT, Gemini, or Perplexity.

The results surprised me. Only two insurers appeared in every response. Two others appeared in three of the four responses. Three insurers appeared only once. The remaining insurers never appeared at all. Depending on the response, between 67% and 73% of the licensed market was effectively invisible to customers.

The purpose of this exercise was not to assess the accuracy of the answers. My interest lay elsewhere. The same question generated different shortlists, rationales, and visibility for competing companies, all within minutes.

That matters because the customer journey is changing. For years, customers searched for information on Google, reviewed websites, compared options, and made decisions. Now, more customers are asking AI for recommendations. “Which insurer should I choose?” “Which bank has the best customer service?” “What’s the best option?” Instead of receiving multiple links, they now receive a direct answer.

This shift may appear minor, but it has significant implications. In the search era, organisations focused on being found. In the AI era, organisations increasingly need to be recommended.

AI-generated recommendations may draw from company websites, reviews, complaints, news coverage, comparison sites, public discussions, and many other sources. Some information is current, while some was incorporated into the model months or years ago and cannot be easily updated. Your organisation still influences certain inputs, but your narrative now competes with others and with previous versions of itself.

The company never fully controlled the conversation. It controlled its own channels: the website, advertisements, and official messaging. The conversation itself always belonged to customers. Previously, this conversation was scattered and quiet. Now, it is aggregated, summarised, and delivered as a single answer. The company does not own it. It participates in it.

When customers read multiple reviews, they filter the information themselves. They weigh extreme opinions against more balanced ones and notice when a low rating is unrelated to service. While this process is inconvenient, it also offers protection, as scepticism arises from evaluating information independently.

AI eliminates this effort. The synthesis occurs before the customer begins researching, and the conclusion is presented as an authoritative answer rather than a collection of opinions. An answer is much harder to dispute than a set of reviews. Whether fair or not, that summary can quietly become the prevailing perception.

We have seen a version of this before. When online reviews first appeared, many organisations dismissed them. People won’t trust strangers on the internet, the thinking went. That held right up until reviews quietly became part of how customers decided, at which point businesses scrambled to manage something they had ignored for years.

AI discovery appears to be at a similar stage. While most are aware that answers exist, not everyone recognises that these tools are becoming a primary discovery channel, where consideration begins, rather than a novelty. Organisations that recognise this early will be seen as forward-thinking.

For customer experience leaders, this is not a marketing issue. It is a customer experience issue. Every review, complaint, resolution, and interaction contributes to the public record from which future recommendations may be formed.

A well-resolved complaint does more than save a customer. It changes the story. A review answered thoughtfully does more than protect a rating. It demonstrates how your organisation behaves when things go wrong.

The organisations that benefit most from AI recommendations will not necessarily be the loudest. They will be the ones who consistently create experiences worth talking about.

For years, organisations worried about being found. Increasingly, they may need to worry about being considered. Customers cannot choose a company that never makes the shortlist.

A customer walks in and gives an incomplete instruction. This happens. Customers do not always give instructions in perfect sequence. They pause. They remember something. They get distracted. They are human.

The team member assumes the instruction is complete and announces the total. The customer points out that they had not finished giving their instructions. The team member responds: "I already completed the transaction, so you have to pay." The customer responds: "Get me a manager."

This was interesting, not because it escalated, but because a team member believed they had no better option than to complete the transaction and defend it. For me, another question arises: can customer-centricity be taught?

The first failure: the assumption

An incomplete instruction is not a problem. It is an opportunity to ask a clarifying question. "Will that be everything?" costs about three seconds. We skip it for speed, only to completely blow up a manager's afternoon.

A clarifying question is the most cost-effective tool in customer service. Get curious.

The second failure: the belief at base

The second failure was not the sentence itself. It was the belief behind it: that they could do nothing. That belief is seldom true. What it usually means is, "I did not see an easy option," or "I did not want to deal with the available option." That is why this one escalated.

Customer-centricity is not only behaviour. It is judgment in the moment. Do I pause? Do I clarify? Do I reverse? That kind of judgment is practised and coached in real scenarios.

Could the system reverse the transaction? If not, that is a design failure — it means we built systems that penalise customers for ordinary human error. If yes, did the team member know how? If not, that is a training failure. If they knew how and still did not act, that may be a willingness failure.

Even willingness deserves a closer look. Sometimes it truly is attitude. More often than not, what looks like attitude is adaptation. People learn quickly what gets praised, punished, or rushed. If speed is rewarded over accuracy, people rush. If questions are treated as inefficiency, people assume. If initiative has been punished before, people default to the safest sentence available: "I had no choice."

Same sentence. Three different root causes. Three completely different fixes. Coaching the team member only addresses one of them.

So, can it be taught?

Yes, but only if you build three layers. I think of it as Taught, Equipped, Permitted.

Taught: The unconditional rules. Always ask before you assume. Always confirm by repeating the full instruction back to the customer before moving on. "So that is X and Y, correct?" The customer hears their own instructions out loud; any gaps surface, and the correction happens before the transaction rather than after.

Equipped: The rehearsed judgment for the pressured moments can be practised with if/then statements: If the customer says I got it wrong, then I say "Let me fix that" before I explain anything. If I am not sure the system can reverse it, then I say "give me one moment to check", not "I can't." Each of these behaviours is observable; therefore, each can be coached.

Permitted: What is permitted in the environment and culture? What do systems enable, what do leaders measure, and what does the culture tolerate? Permission is granted every time a leader chooses accuracy over speed, every time a supervisor backs good judgment, every time a system makes the right action possible.

The team member failed the customer. However, when your frontline says "I had no choice," the real question is not whether they care. It is whether they have been taught, equipped, and permitted to choose differently.

Customers value speaking with someone in person. They want reassurance, guidance, and human support. What they do not want is a process that feels longer than it needs to be.

The finding is not the full diagnosis.

When customers say the in-office process takes too long, that statement needs to be broken down. Which part takes too long? The waiting time? The documentation review? The payment process?

The customer describes the issue as one thing. Operationally, the delay may be coming from several points in the workflow. Without structure, the team risks fixing the wrong thing.

Fair warning: I love a model. You are welcome!

Five Whys

You do not always need a full toolkit to find a root cause. In many service scenarios, one method will get you there: Five Whys. It works by refusing to accept the first answer. You ask why until the issue becomes specific enough to act on.

That is the root cause. The workflow was designed for the organisation's convenience, not the customer's.

The recommendation is no longer "work faster." It becomes: redesign the documentation stage to pre-populate existing customer information, removing the need for manual re-entry and duplicate confirmation. That is a specific, actionable, solvable problem. Five Whys was the girl who got us there.

When you need more structure, there are more models!

Five Whys works best when the team agrees on what the problem is and is ready to investigate it honestly. Sometimes that is not the case. If the workflow is complex or undocumented, a process map or service journey map makes the full picture visible before any analysis begins. If there are multiple competing problems, a Pareto chart identifies where the greatest customer impact sits. If the team keeps shifting the goalposts, the Is/Is Not method creates useful boundaries around what is actually being solved. If the problem spans several departments, a fishbone diagram can open up thinking. These tools earn their place in the Root Cause Analysis Hall of Fame. Knowing when to reach for them is what separates a useful investigation from a very long meeting.

Customer feedback tells us what customers are experiencing. Root Cause Analysis (RCA) tells us why it is happening and where to act. In this case, 92% of customers still want to be in that office. They want the guidance, the reassurance, the human interaction. What they do not want is a process that wastes their time. That is a solvable problem.

No one signed the email. Nearly a month later, I am still waiting.

The problem is not that government services are going digital. The problem is that digitisation can make bad service look modern while leaving the underlying experience unchanged.

Trinidad and Tobago has committed to digital government. The DigitalTT National Digital Strategy 2023–2026 and the UNDP's active Digital Transformation Project share a stated mission to improve efficiency and reduce transaction costs for citizens. That push is necessary.

Digitisation changes the channel. It does not change the culture behind it. In my case, what moved online was the submission. What stayed exactly where it was — vague, unaccountable, unresponsive — was everything that followed. That is what tends to go unspoken in conversations about digital transformation: we do not only move processes online. We move our existing service culture online with them. Where there was a long line and no clear timeline, there is now a web form and no clear timeline.

In person, even a difficult service interaction had a human face. You could ask someone their name. You could see who was responsible. That basic form of human accountability, however imperfect, existed. Behind a no-reply email signed by a department rather than a person, it does not.

In a physical office, a frustrated citizen was impossible to ignore. An angry face, ah "cuss out", a person standing in front of you demanding an answer — that discomfort created a form of accountability. A public servant had to respond to it in some way. Online, that dynamic shifts. The citizen is still frustrated, still waiting, still unserved. But there is no witness. An unanswered email generates no notable fuss. Digital channels, without deliberate design, not only deliver poor service but can also shield the institutions that deliver it from ever having to confront the human cost of that service.

While Trinidad and Tobago does not appear to have a widely published citizen satisfaction benchmark for digital public services, the global pattern is instructive. A 2022 Deloitte survey found that satisfaction with online government services lags the private sector by more than 20 percentage points. T&T is not uniquely behind. The lesson is that digital access alone does not guarantee a better citizen experience.

Digital government services should include:

• A visible service owner
• Clear service-level timelines
• Status updates at key stages
• Human support when the process stalls
• Follow-up communication that tells citizens what happens next

There is a difference between digitising a transaction and transforming a service. Digitising a transaction moves a form online. Transforming a service redesigns the experience around the citizen. If we do not take that opportunity now, we will spend the next decade digitising the same problems we have always had. We will just have a website for them.

I am not exaggerating.

Before I even walked in, I had to find a photocopier off-site. The office doesn't have one available to the public. Fine. I came anyway. Ready.

I joined a manual queue, received a number, and was told absolutely nothing about what came next. No wait time. No process explained. Just a number.

Then I waited for five hours. In a room where eating wasn't allowed, the restroom was out of order, and leaving wasn't really an option because names were being called, and you had no idea when yours would come up. No updates. No acknowledgement. Just silence and a sign telling me what I couldn't do.

At 1:30 PM, I was seen, photographed, and told to return in six weeks. No date given. Just six weeks.

So I did all of that, and I still had to come back.

Here's the thing about working in CX: you can't turn it off. You don't just have a bad experience. You map it.

What I experienced wasn't a bad day or an understaffed office. It was a system that was never designed with the customer in mind, and it showed at every stage.

I mapped the whole journey. See it for yourself below.

The product itself won't likely wow you. The experience around it might.

Looking at the data year after year, across multiple channels, the things that move customers from neutral to advocate are almost always people-centred. The rep who knew the answer. The agent who made a stressful claim feel easy. The person on the other end of the line, who you reunite with once a year, who knows your whole life story.

Customer service. Empathy. Ease of communication. Knowledge. These are the NPS drivers.

Add the external economic environment layer. Premiums are up. Cost of living is up. When a customer answers an NPS question, they aren't just rating their service experience. They're rating how it felt to pay that bill in this economy.

As we move toward digital, if we remove the human touchpoints that have historically been our score drivers in favour of faster, self-service interactions, are we designing for a better experience? Or are we engineering a channel that, in a compulsory category, will consistently produce emotionally neutral responses?

A digital-native insurer like Lemonade can serve as a counterexample, but they're solving a different problem. Traditional insurers aren't building digital-native. They're digitally migrating customers who came through a branch, an agent, a call centre, or a trusted person.

Digital can reduce friction. But friction reduction is not the same as emotional advocacy. If human trust has historically driven NPS, then digital migration must intentionally recreate trust, reassurance, and guidance — not just speed.

Has digital transformation positively moved your NPS, or has it just redistributed it?