Building a CX function that protected $505M in lost sales.

The situation

The organisation had 55,000 customers and no dedicated customer experience function. But the deeper problem was that nobody could even see the customer.

Customer details were scattered across six different platforms and touchpoints — the mobile app, Facebook, Intercom, Zendesk, WhatsApp, and Instagram — each managed by different people with no central focal point. Customer issues sat pending for months. Customers were lost after several complaints. There was no way to track what had been resolved and what hadn't. Issues lived or died at the hands of whichever support staff member picked them up and whatever knowledge that person happened to carry.

There was no voice of the customer. No single place where the organisation could see and validate the issues it was having and resolving month on month. The business was scaling; the customer operation was invisible.

The diagnosis

The problem was structural.

There was no operating model to route, prioritise, and resolve customer issues against defined standards. Without a central platform, nothing was trackable. Without SLAs, nothing was measurable. Without compliance systems, nothing was auditable. Without data infrastructure, nothing could be personalised. And without a single accountable unit, every improvement was an adhoc project instead of a process.

I also couldn't fix what I couldn't see. Before I could set standards, I needed a baseline. Before I could set SLAs, I needed to know what was actually breaking, where, and how often. The fix had to be built in sequence — each step unlocking the next — not retrofitted onto existing workflows.

What I built

1. The CX business unit

First, I set up the customer experience unit from scratch — defining its remit, staffing model, reporting lines, and escalation paths. The unit owned the full customer journey end to end, from first contact through resolution, with clear accountability for every stage.

This had to come first. Everything else needed a home and an owner.

2. The platform rollout (Freshdesk)

With the unit in place, the next problem was the data. I needed one place where every customer interaction lived, regardless of which channel it came through.

I researched the options and chose Freshdesk because it could integrate with all our touchpoints — the mobile app, mobile shops, drivers' apps, and every social commerce channel. The goal was to track all customer support issues in one place, eliminate leakages, and ensure no issue existed without an owner or without transparency.

I rolled out the platform and migrated all existing customer data from the scattered platforms into it. I integrated it with every social media touchpoint and every local and digital touchpoint. I onboarded new customer data and established a central point for all customer support interactions.

Then I defined and automated a process that allowed customers to raise tickets directly through their channels of choice — email, dedicated phone lines, SMS, WhatsApp, and social media. The same process enabled support staff and any other employee to raise tickets on behalf of customers or themselves.

After the automation, I trained all support staff within the CX unit and ran a company-wide training so every business unit knew how to reach us directly. The internal goal was simple: every staff member is first a customer experience agent. The rollout and training reinforced that.

The result was that for the first time since inception, the organisation had data on customer complaints. It gave us a baseline to work with. It gave us visibility into where we'd been losing money through different touchpoints and what we could do to fix it.

This is what unlocked everything else. Without the platform rollout and the data it gave us, we couldn't have set SLAs, built compliance, or justified the CDP migration. We would have been setting standards against nothing.

3. The SLA framework

Now that we could see the data, the next natural step was to set standards against it.

I designed and stood up the service-level agreement framework — tiered response and resolution targets by issue severity, escalation triggers, and a measurement system that made performance visible in real time. SLAs weren't aspirational; they were operational, with owners and consequences.

• Tiered response and resolution targets by issue severity and customer segment — so a critical issue for a key account wasn't held to the same clock as a routine query
• Escalation triggers that automatically routed at-risk tickets before they breached — prevention, not just reporting
• Real-time dashboards giving leadership live visibility into resolution performance across the 55,000-customer base
• Weekly review cadence turning SLA data into corrective action

This was the framework that reduced $505M in lost sales across two quarters. Once you can see what's breaching, why it's breaching, and who owns it, you stop losing sales you didn't need to lose. Quarterly revenue and CSAT each increased by over 10% as a direct result.

4. Compliance systems

With the SLAs in place, I built the compliance systems to make the unit audit-ready — documented processes, data handling standards, and a controls framework that could withstand internal and external review. Compliance was designed in from day one, not bolted on after a finding.

• Documented processes for every customer-facing workflow — nothing operated on tribal knowledge
• Data handling standards aligned to regulatory requirements — customer data treated according to the rules, not the convenience
• Controls framework with evidence trails — every control had an owner, a check, and a record

5. The CDP migration

The Freshdesk rollout proved something: we could consolidate scattered data and make it usable. That gave us the confidence to go further.

The goal of the CDP migration was personalisation — but personalisation requires knowing everything you can about your customers from every place their data lives. So I led the migration of 60,000 customer records into a Customer Data Platform by pulling from Mixpanel, Vicidial (the platform we used for inbound and outbound calls), Freshdesk (which now held all their aggregated data and every conversation we'd had with them over the years), and the organisation's existing software stack.

This laid the groundwork for data manipulation that enabled personalised services at scale. For the first time, the business could see who its customers were, what they needed, and where the experience was breaking down — all in one place. The migration completed with zero downtime.

The people transformation

This wasn't a separate workstream. It happened side by side with everything else, because the transformation was in the daily doing and managing alongside all our other work.

I took the lowest-performing employees across various teams and trained them on customer journey management — not as a one-off workshop, but as an ongoing coaching practice tied to the SLAs and the new workflow. I re-engineered the resolution workflow itself: eliminating handoff delays, introducing triage at first contact, and automating routine resolution paths.

The people who had been the weakest links became customer champions. Resolution times dropped by 70%. The unit achieved a 75% Customer Satisfaction score. And the transformation stuck because it was tied to a system they could actually operate, not a motivational speech they'd forget by Friday.

What made it work

The unit succeeded because it was built in sequence, as a system. The CX unit gave everything a home and an owner. The Freshdesk rollout gave us the data — the baseline we needed to see what was actually happening. The SLAs gave us the standard to measure against. The compliance systems gave us the foundation to defend it. The CDP gave us the depth to personalise. And the people transformation, running through all of it, gave us the capability to deliver.

Each step unlocked the next. I couldn't set SLAs without data. I couldn't build compliance without standards to comply with. I couldn't justify the CDP migration without proving we could consolidate and use scattered data first. The sequence wasn't a preference — it was a dependency chain.

When I walk into an organisation where the customer operation is stuck, I don't start with a team or a tool. I start with the question: what's the system this needs to become, and what's the order that makes each step possible? Data, process, standards, people — but only once you can see what you're working with. Because without the data, you're setting standards against nothing. And without the sequence, you're building pillars with no foundation.

Building from zero meant there was no legacy to negotiate. But the same logic applies to inherited operations — you just have to diagnose what exists before you decide what stays. The question is always the same: does this make the customer's problem go away faster, measurably, and defensibly? If the answer is no, it gets rebuilt.