The AmplifierReal cases · Banking · Branch manager
Suresh, 48: a bank branch, new digital targets, and a team that has stopped listening
Twenty-two years of knowing every customer by name, a dashboard he does not understand, and a young RM who moves faster than him.
Planned for release phase 2. The page is complete as a draft; interactive parts run in the browser and do not yet save to an account.
In brief
The situation. Suresh manages a public-sector bank branch in a district town in Maharashtra. Twenty-two years in the bank; he knows the customers, their families, their harvests. Head office now measures him on digital adoption, cross-sell and a dashboard of things he did not learn. A 26-year-old relationship manager on his team is fluent in all of it, and Suresh has noticed himself avoiding her. He recognises, reading a description of the Avoider, a man he does not want to become.
Step 1 — Define the End State
"By the next quarter, I understand every number on the dashboard well enough to explain it to my staff, and I use what I know about customers — which the dashboard doesn't — to hit the targets in a way head office would not have thought of. The young RM sees me as someone learning, not someone in the way." Audiences: his team, the RM, the regional head, and a farmer named Patil who has banked here for thirty years.
Step 2 — Map the Flow
His list: learn the dashboard, pick two targets, map which customers each fits. AI review adds: ask the RM to teach him one metric a week — a human task the tool cannot do and he had been avoiding; and a customer-segment list built from his own knowledge before looking at the system's segments, then compared. It suggests automating outreach messages. He takes it in a smaller form: drafts, sent by staff who know the customer, never automatically.
Step 3 — Assign Tasks
Which customers are right for which product: his — twenty-two years of context in his head. Explaining each dashboard metric in plain language: AI as tutor, then confirmed with the RM. Drafting outreach messages in Marathi: AI drafts, staff personalise. Deciding who is called and by whom: his. The weekly lesson: the RM.
Step 4 — Select the Tool
A conversational tool for the metric explanations and message drafts; the bank's own system for everything with a number in it. He uses the tool on his phone, in the evening, in Marathi as often as English.
Step 5 — Prepare and Direct
Criteria: I can explain a metric to a clerk in two sentences before I count it learned; no message goes out that names a product before it names the customer's situation; every outreach list is one I built from my knowledge and then checked against the system, not the other way round.
Steps 6 and 7 — Execute and Evaluate
The metric explanations are clear and, twice, subtly wrong about how his bank calculates them — the RM catches both, which is the point of the weekly lesson. His customer list for a loan product overlaps only 40% with the system's segment; the 60% the system missed are his, and they convert. Message drafts are generic until he adds three real situations to the brief.
Step 8 — Integrate and Stress-Test
End of quarter, assembled. Small fixes: a message template in the wrong dialect for one taluka. Structural: the RM is teaching him but he is not teaching her — his end state said "learning, not in the way", and the flow had nothing going the other direction. He adds a task: one customer visit a week with her, where he does the talking. A Step 2 gap; the fix is a car ride.
What changed
He hits the targets, partly with customers the system did not know to look at. He can explain the dashboard. The RM, asked by a colleague what Suresh is like to work for, says: "He knows things the system doesn't, and he's started asking." The stack — twenty-two years at the bottom, a middle layer added at 48 — held.
Build your own list from what you know before you look at the system's; the difference is your value. Let the person who is ahead of you teach you one thing a week, and teach them one back. A principle: the customer's situation before the product's name.
A composite case, not a client engagement; the person is invented, the situation is not.