Delivering Bad News to Customers: Drill It
Chapters
TL;DR
Delivering bad news to customers goes well when the rep states the change before any recap, context, or apology, owns the impact without blaming a vendor or another team, and closes on one committed next step with an owner and a date. Train it as a drilled first response, score it against one rubric row per call stage, and pass reps only on what the transcript shows.
- Drill one agreed first response per type of bad news until reps say it without notes.
- Score three rows: headline first, impact owned, next step committed.
- Write exit criteria a manager can verify from the transcript alone, including a no-overpromise check.
- Debrief one flagged moment, let the rep diagnose it first, and book the re-run before the session ends.
Why do reps bury the bad news?
Reps bury bad news because the first sentence is the hardest one to say, and nobody has made them say it under pressure before the real call. Without rehearsal, they open with progress, context, and apology, and the customer hears the actual change minutes in.
The burying has a recognisable shape on a transcript: a warm opening, a recap of what went well, a passive-voice cause, and finally the news, softened. Each step feels like kindness to the rep. To a customer waiting for the point, it can sound like stalling.
The fair counterargument is that experienced account managers learn to deliver bad news on live calls, and some do. The question is who pays for the lesson. A fumbled service delay conversation with a live account spends trust with a real customer. A fumbled practice call costs the rep a low score and a re-run.
On a bad-news call, the rep's first job is to say the news, not to prepare the customer for it.
We treat bad-news calls as a high-stakes conversation that earns scored rehearsal before it reaches a live account, the same way we treat a security review or a renewal pricing call.
Drill one agreed first response per type of news
Give every rep the same first response for each kind of bad news, and drill it until it comes out without notes. The first response covers three things: what changed, what it means for the customer, and that a plan follows.
For a missed deadline, a version we recommend: I have bad news on the rollout. We will miss the go-live date we committed to; the new date is [new date]. I want to walk you through why, what it affects on your side, and what we are doing about it. Note what the line leaves out. No adjective about how hard the team worked. No vendor named as the cause in the first breath. No promise the brief does not already back.
Strong reps object that a fixed line makes them sound canned. The objection is real for the branches. A customer's reaction is unpredictable, and a scripted reply to anger sounds worse than an honest one. So we fix only the first response. Once the headline is automatic, the rep's attention is free for the customer's actual reaction, which is where improvisation belongs. Reps caught flat improvise the headline itself, and the improvised headline is the one that gets buried.
Rehearse the first sentence so the rep can improvise everything after it.
What does the rubric score at each call stage?
The rubric scores one observable behavior per stage: the headline lands first, the impact is owned, and the next step is committed. Each row has a pass description a manager can check against the transcript without having heard the call.
Load the rows into whatever scoring you use so feedback cites your own call stages, not a generic service rubric. Reps discount feedback graded against someone else's call structure, and they are rational to. On XL Roleplay, call stages and rubrics are loaded first and every session is scored against them; a manager with a printed rubric can do the same job more slowly.
If your org runs Sandler, set the up-front contract before the call, when you book it: purpose, time, both parties' agendas, and the acceptable outcomes, including that the customer may say no to the recovery plan. Its weakness on a bad-news call is what reps do with it afterwards. They restate the contract at length on the call, and the restatement buries the news. Score the opening row as passing only when the headline still lands before any recap or context.
Score the behavior a transcript can show, not the tone a manager remembers.
| Call stage | Rubric row | Pass looks like | Fail looks like |
|---|---|---|---|
| Opening | Headline first | The change and the new date or terms appear before any recap, context, or apology | Progress recap, context, or apology before the change |
| Impact | Impact owned | Rep says we for the cause and names the customer-side effect before the customer asks | Passive voice, blame on a vendor or another team, customer names the impact first |
| Close | Next step committed | One action, an owner, a date, and the customer agreeing in their own words | We will keep you posted, or a commitment not in the brief |
Build scenarios from three kinds of news
Write one scenario per type of news, pulled from a real account where possible, and give the persona playing the customer a specific stake. In customer success roleplay, the persona's stake is what makes the pressure real.
Missed deadline. Headline: We will miss the go-live date; the new date is [new date]. The overpromise trap is a recovery date the delivery team has not confirmed. Persona stake: an internal launch the customer has already announced.
Price-impacting change. Headline: Your renewal price will change, and I want to give you the new figure and what drives it. The trap is a hinted discount the rep has no authority to give. Reps who own renewals should also run our price increase conversation practice, which covers the negotiation that follows the news.
Feature sunset. Headline: We are retiring [feature] on [date], and there is a replacement path. The trap is promising the replacement matches every workflow. Persona stake: a team whose daily process runs through the retired feature.
Give each persona one pressure line to use right after the headline, such as Why am I only hearing about this now? The line tests whether the rep stays with the impact or retreats into explanation.
An overpromise on a bad-news call is a second piece of bad news, scheduled for later.
How do you write exit criteria a transcript can verify?
Write each criterion as something a manager could find by reading the transcript alone: a sentence, its position, and who said it. If a criterion needs the manager to have felt something on the call, rewrite it.
Weak: rep shows empathy. Verifiable: rep paraphrases the customer's stated concern before offering any explanation. Weak: rep is clear. Verifiable: the change and the new date or terms appear before any recap, context, or apology. Weak: rep sets a next step. Verifiable: the transcript shows one action, an owner, a date, and the customer agreeing to it.
Add one negative criterion: no commitment appears that is not in the scenario brief. It is the easiest fail to verify and the most expensive to miss on a live call.
Read results per rep per scenario. A rep who passes the missed-deadline scenario and fails the sunset scenario has a specific gap; a blended average hides it. Talk/listen ratio is worth reading as a diagnostic input. A rep who talks over the customer's reaction often shows it there first, and the transcript tells you why.
Experienced managers do catch obvious problems by ear. The trouble is unevenness. Confident reps sound ready on bad-news calls because they sound calm, and quiet reps who deliver the headline cleanly get underrated.
A confident voice is not a passed rubric row.
Run the drill this week
Pick one rep with an account manager's bad-news call coming up, and use that account. The run itself costs you 30 minutes, which we recommend splitting into 5 minutes of setup, 15 of drill, and 10 of debrief. Scenario prep of about 10 minutes happens beforehand and sits outside that run.
Before the run: prep, about 10 minutes. Choose the scenario from a live account with news pending, pick one rubric row to score, and write the brief: what changed, the new date or terms, and what the rep may and may not commit to. Setup, 5 minutes. Read the brief to the rep and confirm the agreed first response. Drill, 15 minutes. We recommend three attempts against the customer persona, with the pressure line delivered after the headline. Score every attempt on the chosen row. Debrief, 10 minutes. One behavior, covered in the next section.
Pass bar. We recommend two consecutive attempts that meet the row's pass description from the transcript alone, with zero commitments outside the brief.
What a failing attempt sounds like. Thanks for making time. Before anything else, the team has made real progress on the integration, and the data mapping is in great shape. There were some challenges with a third-party dependency that had a knock-on effect on timing. We are doing everything we can, and I am confident we can get you most of the way there by the original date. The headline never lands, the cause belongs to someone else, and the last sentence is an unapproved promise.
A rep passes when the transcript shows the behavior twice in a row, not once by luck.
Debrief one behavior, then book the re-run
Pick the single moment that cost the attempt its pass, usually the sentence where the customer should have heard the news and did not. Read or replay only that moment; a scored report that links flags to the exact transcript line saves the search.
Ask the rep to diagnose it before you say anything: Find the sentence where the customer first learns the date moved. What came before it? A rep who finds the sentence owns the fix in a way a rep who is told does not. Then give your verdict in one sentence and name the one behavior to change.
Leave the passive-voice cause and the soft close for another session, even if both were present. A debrief that covers three problems sends the rep away with none fixed. Our roleplay debrief guide covers the full script.
Book the re-run on the calendar before the rep leaves, on the same scenario and the same row. We recommend a 15-minute 1:1 built around that flagged moment, held in addition to the pipeline 1:1 and never folded into it. Deal reviews always feel more urgent than a re-run, so the re-run needs its own protected slot.
A failed drill without a booked re-run is a scored opinion, not coaching.
Frequently asked questions
Should the rep apologize before stating the bad news?
Apologize after the headline, once and briefly. An apology before the news delays it and makes the customer brace without knowing for what.
Can a rep explain the root cause on a bad-news customer call?
Yes, after the headline and the impact, briefly and with we as the subject. The cause is context for the customer, not a defense for the rep.
What if the rep does not have a confirmed new date yet?
The headline states that the date has moved and when the new date will be confirmed. The committed next step is that confirmation, with an owner and a date.
How many first responses should a team drill?
We recommend starting with the three types covered above: missed deadline, price change, and feature sunset. Add a new first response only when a new kind of news starts showing up in live accounts.
Who should play the customer?
An AI persona or a peer works, provided the player has the brief and the pressure line. A peer who knows the rep may go easy, so script the pressure line so every attempt faces the same test.