A journal for customer experience leaders.
Inside this edition: how to make every complaint, dispute and failed transaction leave the operation stronger than it found it, and where AI earns its place in that.
Welcome to the first edition of
The Compounding Edge, Nigeria Edition. Like a good praline, the magic is in the layers and how they come together. We believe the same is true of customer interactions.
A single interaction may seem small. But thousands of interactions, compounded over time, can reveal patterns, shape better decisions, and create a meaningful advantage. We look forward to collaborating soon.
Last year, complaints to Nigerian banks led the CBN to order ₦19.12 billion in refunds. The money went back to customers. But the problems that caused those complaints stayed where they were, ready to cost the same again the following month.
Fraud went differently. Over the same twelve months, reported digital payment fraud value fell 51%, because institutions shared what each breach taught them. Every attack made the next one harder to pull off.
Both outcomes came from the same raw material: what customers and criminals reveal in ordinary interactions. On fraud, the sector kept what it learned and stacked it. On complaints, it paid and moved on.
Nigerian institutions pay twice for the same problem: Once at resolution, and again when it returns the following month.
Customer operations across Nigerian banking handle a remarkably stable set of problems. Failed transfers and disputed debits recur month after month, yet each one lands on the desk of somebody solving it for the first time. Nothing retains how it was resolved before, so the unit cost of resolution never falls.
That carries a measurable price. Complaints climbed sharply through 2025, prompting the CBN to order ₦19.12 billion returned to customers. KPMG's 2025 West Africa survey still ranks Resolution, meaning proactive correction of the customer's problem, last among the six experience pillars.
Every refund is a lesson the operation has already forgotten.
Compounding describes the alternative, where each interaction leaves the team and the underlying product measurably better than it found them.
Electronic payments across 300 million+ active accounts.
Nigerian institutions hold the richest behavioural dataset in the economy, and most of it evaporates the moment a ticket closes.
Does last month's leading complaint reach product as a written spec?
Can a new agent inherit the best agent's resolution playbook on day one?
Is recurrence measured after a ticket is marked resolved?
Does dispute language reach compliance without anyone forwarding it?
Four nos describes an operation that is well run and going nowhere.
Worth ranking above CSAT and handle time, because it answers whether the same issue keeps resurfacing, which is the clearest available signal of an operation that learns.
Growth is exciting but it's also diagnostic. More customers arrive, teams expand, work becomes more complex, and decisions have to be made faster and with better intelligence. Somewhere in that, the informal operating model of goodwill, memory and a few capable people knowing what to do begins to fail under the weight of scale.
Across transformation programmes, one truth keeps surfacing. Growth increases revenue and it increases visibility, exposing what an organisation has been managing quietly, informally or inefficiently.
Unclear responsibilities become missed actions. Poor communication becomes rework. Weak processes become delays. And when critical knowledge sits with one person, even the most committed employee becomes a bottleneck.
The instinct is often to push people to work harder or be more proactive.
Effort is not a service improvement strategy.
But at a certain point, the issue is no longer the people. It is the operating system around them. Growth teaches us that structure is not bureaucracy, because structure designed well reduces confusion, strengthens accountability and frees people to perform at their best.
AI can amplify this by removing friction, surfacing insights and enabling teams to focus on higher-value work. But technology alone will not fix an unclear operating model, because the processes, responsibilities and decision paths around it still need to work.
So when an organisation begins to grow, the question should not only be how to win more business. You must ask: what must evolve internally, in our systems, processes, roles and use of AI, so that growth does not overwhelm the people delivering it?
Sometimes the employee labelled not proactive was never given true ownership. Sometimes a team looks disorganised because the workflow around them is disorganised. And sometimes what appears to be resistance to change is exhaustion from too many changes introduced without enough clarity, sequencing or support.
Each one widened conversation coverage first, then spent what it learned on making people better. The returns followed from the learning rather than from the software.
Intel Connect went from reviewing under 1% of conversations to analysing every monthly call, over 600,000 of them. Supervisors who had spent their time spot-checking compliance began coaching against evidence, and agents doubled how often they made an offer during a service call.
The coverage was the input. The compounding came from what supervisors did with it every week.
Deutsche Telekom worked with McKinsey's QuantumBlack to build a personalised AI coaching engine, replacing the annual training calendar with feedback that arrives inside the daily workflow.
A major financial services provider ran voice analytics across its contact centre to establish why customers were calling, then decided what to automate. The same diagnostic mapped a path to a 45% reduction in operating cost.
Gartner projects the cost per GenAI resolution passing $3 by 2030, above what many offshore human agents cost, and expects "right to speak to a human" rules to lift assisted-service volumes 30% before 2028. Automation bought to strip out headcount runs into a ceiling and a rising unit price. Money spent making the people you keep measurably better carries on compounding.
Establish goal clarity with the executive team, then isolate the single highest-volume complaint journey and baseline its recurrence rate with finance in the room.
Close one complete loop, running from frontline conversation through to a permanent product or process fix. A single loop that ships teaches more than a transformation programme on a slide.
Measure the drop in recurrence, quantify handle cost saved, and take that number to the board before requesting further budget. Then repeat on the next journey.
Insight7 is an AI-native platform for CX teams. We QA every customer interaction, coach your agents with roleplay practice, and surface what your customers are really telling you.
Every conversation should leave the operation better than it found it.
Central Bank of Nigeria, Annual Report 2025 (complaint volumes, ₦19.12bn refunds, regulatory mandates). NIBSS payments and fraud data 2025 (₦1.07qn payment value, 51% fraud decline). KPMG, 2025 West Africa Banking Industry Customer Experience Survey (Resolution pillar). Gartner press releases 2025–26 (87% human-access expectation, GenAI cost per resolution to 2030). McKinsey & Company (Entel Connect and QuantumBlack case work, banking handle-time diagnostic). PwC Customer Experience Survey (38% report worse service). Field observations come from Insight7 discovery conversations with Nigerian CX leaders, reported in aggregate and never attributed.