Customer Service: Delivering the News

Friday 13 March, 07:15. Siân Pritchard has a spreadsheet open with roughly 200 orders on it, and behind those orders about 400 named people — a buyer, a head chef, a site project manager, a bakery owner. Every one of them has an oven coming from Rockford, and none of them knows anything is wrong.

That is not an accident. Ravi’s instruction has held for ten days: no customer hears a date until Calder has a date worth giving. It is about to be spent. By this evening all four hundred will have been told something, and the quality of the telling will matter to the year’s revenue roughly as much as the allocation did.

What lands on Siân’s desk

From Marcus comes the protected list and a dated commitment schedule — which customer gets how many units, in which tranche, in which week. Harlow’s 180 Meridians re-phased around which floors actually close. Nordfoods’ 240 Compacts spread across a longer window. Bellwether’s 60 early, because the dry dock cannot move. St Chad’s 40 before a kitchen handover. Acme Bakery’s single oven, still on the list. Those five accounts need 521 boards against the 520 that exist, and Marcus has spent his week deciding which name comes off.

From Grace come the only dates anyone may use: last board-fed oven off the line Friday 17 April, build-ahead 20 April to 22 May, full-rate assembly resuming Tuesday 26 May. So for anything outside the protected list the earliest honest date is the first week of June, and for much of the tail, later.

None of it is hers to change. She cannot move a board, improve a date or add a customer. She decides only how four hundred people find out — which turns out to be a larger lever than it sounds.

What a customer service lead actually does

In an ordinary month the job is order status, delivery slots, spare parts and complaints — reactive, transactional, invisible when it works. A shortage changes its shape. Sales owns the relationship and the commercial concession; customer service owns the telling and the record. Marcus decides that Nordfoods is protected; Siân decides that Nordfoods hears it on Friday morning from a named human being rather than from a delivery that fails to arrive in May. And because hers is the only function that will speak to every affected customer — sales talks to the accounts worth talking to, finance sees the ledger — it is the only desk that ends the fortnight having heard back from all four hundred.

The software on Siân’s desk

The service desk turns a protected list into a work queue with an order and an owner, which is how four hundred conversations happen in a fortnight without anyone important being forgotten. The CRM view is what a rep opens before dialling, because the fastest way to make a bad call worse is to be told something the customer already knows better than you do.

The two screens that matter most are the recording ones. Logging each conversation in a structured way — flexible, at risk, escalating — rather than as free text in an inbox is what lets the last screen exist: a report that hands recovered flexibility back to Tomasz and at-risk accounts back to Marcus while there is still time to act. Most companies discover they needed that structure only after the crisis has ended.

The decisions

Who hears first, and how

Sequence is a decision, not an administrative detail, and Siân makes it before anything else.

The accounts with the worst news and the deepest relationships get a telephone call from a named person, before a word goes out in writing. Bellwether at 08:00, because their harm is soonest and irreversible. Then Harlow, whose contract carries liquidated damages and whose sub-contractors are booked against dates. Then St Chad’s, where lateness triggers a formal supplier review. Then Nordfoods, which Marcus takes himself. The flexible middle — Peninsula and the larger orders in the tail — gets a call from its own account manager between 11:00 and 14:00, from someone the customer already knows. The long tail gets a clear written notice with a specific new date, a named person and a direct line, released at 14:00 and not before, once the calls above it are confirmed done.

The rule underneath is that nobody learns this from the wrong mouth. The worst sequence is a customer hearing it from their own project manager, who rang despatch about a delivery slot and got the truth from someone with no context. The second worst is a competitor, because the sector knows control boards are short and a good salesperson is already ringing Calder’s customers to ask, sympathetically, whether their dates have been confirmed. The third, and the most common, is a lorry that does not turn up.

One date, and what may be promised around it

The single rule that protects everything else: one date, honestly derived, never improved to end a difficult call.

The failure mode happens in every shortage. A representative is forty minutes into an unpleasant conversation, the customer is genuinely distressed, and offering two weeks earlier than the schedule allows would end it. So they offer it. The customer plans around it — books an installer, closes a kitchen, hires agency staff, schedules a store opening. The date fails, because it was never real. And the second broken promise costs far more than the first: the first is a supply problem, which customers forgive because they have suppliers of their own; the second is a credibility problem, after which everything Calder says must be independently verified. A relationship in which every statement has to be checked has already begun to end.

So the boundary is drawn in advance. Without asking anyone, the team may confirm the revised date, offer phased delivery so some units land earlier, arrange a loan unit from the demonstration fleet, and waive delivery and installation charges. Anything touching price or paper — discounts, contract variations, waived penalty clauses, altered payment terms — goes to Marcus, with a callback promised inside two hours and actually made.

Two details make that envelope work rather than merely exist. The fleet is finite: eleven demonstration and training machines, six realistically movable, so Siân holds the allocation herself rather than letting the first four calls spend all of it. And Peninsula will ask for a discount, because Peninsula always asks and this time has a decent case. The answer on the call is not no; it is that is a conversation for Marcus, and I will have him ring you today. A representative with no authority stalls, and a stalling representative sounds like a company with something to hide.

What a good bad-news conversation contains

Siân drills that last question hardest, because it does two jobs. It routinely surfaces flexibility nobody knew existed — customers whose building work is behind anyway, sites that could not take a delivery in April if you begged them. And occasionally it surfaces a problem far more serious than the one you rang about.

Acme Bakery is the human end. One Compact, an independent, ordered six weeks ago, no leverage of any kind — and still protected, because one board is $1,560 of margin and one independent baker telling this story publicly is worth rather more than that. Grace’s resequencing moves their delivery by nine days, which is trivially small news. Siân makes the call herself anyway, and asks the question. Nine days turns out not to be trivial: the owner has a fitter booked, has given notice on the old oven’s service contract, and has a bank drawdown dated against the installation. Calder can fix the first, help with the second and do nothing about the third. Knowing costs eight extra minutes.

Nordfoods is the commercial end. 240 Compacts for in-store bakeries, no penalty clause, price-driven, openly dual-sourcing. The buyer does not get angry, and that is what makes the call dangerous. Anger is engagement; polite acceptance ending in thanks for telling us, we’ll manage is often the sound of an account being re-planned around somebody else. The question earns the real answer: a store-opening programme with fixed launch dates, and a second supplier already qualified on the smaller format. Which stores, and by when, is the difference between phasing that holds the account and phasing that loses it — and it reaches Marcus that afternoon rather than sitting in a spreadsheet until May.

The feedback loop nobody builds

Because customer service is the only function talking to every affected customer, it is the only place two things surface, and both are worth more than the calls.

The first is genuine flexibility. Every customer who volunteers a later date frees a board for somebody who cannot move, and they only volunteer if asked. By the middle of the second week the log shows around thirty orders in the tail where the customer offered a date later than the one Calder was about to promise, several in the critical April weeks. That goes straight back to Tomasz, and it is the only source of extra supply in this whole story that costs nothing.

The second is departure risk. Leavers rarely shout; shouting is engagement. The signals are quieter and quite specific — a request for the full technical specification, a question about serial numbers already delivered, a mention of “reviewing our supply arrangements”, a buyer who stops negotiating. Those go to Marcus the same day, because a customer who is deciding can still be reached and a customer who has decided cannot.

Neither is possible if what customers say lives in individual inboxes. So Siân spends ninety minutes on the Thursday building a shared form with six fields — date given, accepted or not, concession offered, flexibility volunteered and how many weeks, risk signal, follow-up owed — and makes it non-optional. It is crude and it works. Why it did not already exist is worth noticing: Calder’s systems record exhaustively what the company said to the customer, and almost nothing of what the customer said back. Most companies discover that asymmetry only when a crisis makes the second half load-bearing.

The people making the calls

Seven people on Siân’s desk and four account managers borrowed from Marcus cover roughly four hundred conversations over ten working days — about four difficult calls each per day on top of the normal work, which does not sound like much until you have done it. Nobody there chose a sole-source supplier, set the stock level or wrote the allocation, and every one of them will spend a fortnight being held personally responsible for all three.

Her support runs on four things. A fifteen-minute briefing every morning, covering what the team knows including what is still unknown, because a representative who learns from a customer that something changed overnight has lost the room and will not get it back. Scripts as a floor, not a ceiling: the script guarantees the six elements and the boundaries of the envelope, and is not to be read aloud, since a read script is instantly audible and lands as contempt. A named escalation route — her mobile for anything hard, Marcus for anything commercial, with that two-hour callback honoured. And explicit, advance permission to end an abusive call, with an agreed form of words and a guarantee that nobody will be second-guessed for using it. The permission matters more than the usage: people who know they may leave a conversation stay in it far longer.

She also takes the worst call of each day herself — partly because it is what leading looks like, and partly because the worst call of the day is where the most useful information is.

Where this goes wrong

The common failure is not a bad first conversation. It is a good one followed by three weeks of silence, because there is nothing new to say and nobody wants to ring a customer without news. From the customer’s side, silence after bad news reads as indifference or as a further problem being concealed, and both are worse than the truth. The defence is dull: a scheduled re-contact whether or not anything has changed, because nothing has changed and your date still holds is a message, and delivering it is what makes the next date believable.

The second failure comdollars the first. When dates move again in week six, as some will, nobody can reconstruct what each customer was told the first time. The second conversation quietly contradicts the first — different tranche, different concession, different reason — and the customer experiences that as being lied to rather than as poor record-keeping. Which is why the promise log matters more than it looks, and why its absence is only ever noticed at the moment it is needed.

What Siân hands on

To Tomasz goes the recovered flexibility: a dated list of customers who volunteered later delivery, with the weeks each releases, so boards can move to people who cannot wait. One warning travels with it — this flexibility is perishable. A customer who offers on Monday to wait a fortnight has re-planned around the original date by Friday and cannot offer it again.

To Marcus goes the at-risk register: accounts showing departure signals, each with the signal itself, the date it was heard and who heard it. Not a feeling, and not a ranking of who was rudest. Evidence, dated, while there is still time to act on it.

To the record — and so to Claire, to Ingrid and to the reckoning in module twelve — goes the promise log: every commitment made this fortnight, to whom, by whom, with what concession attached. It changes the status of everything upstream. Until Friday morning, Grace’s dates were a production plan and Marcus’s schedule an internal document. By Friday evening they are four hundred promises, made by name, to people now planning their own businesses around them.

Spot the decision

Read each situation and decide how a customer service lead should handle it, then tap a card to check.