Why this is hard to get right
Imagine you're a COO at a 700-person enterprise software company. Last quarter's numbers came in 12% below plan. The board has asked leadership to cut $8M in operating expenses before the end of the fiscal year. You and the CFO have a plan: freeze all non-backfill hiring, suspend international travel, and renegotiate your top 10 vendor contracts. No layoffs — at least not yet.
You need to tell 700 employees what's happening before the rumor mill does it for you.
You open a blank document and stare at it. You know what you want to say — roughly — but every sentence you draft sounds either too alarming or too dismissive. "We're making some adjustments" feels dishonest. "The company faces significant headwinds" sounds like a lawsuit. You spend 40 minutes writing and deleting the same three paragraphs.
This is the scenario most leaders face with cost communications. The content itself is not the problem. The structure is. Without a clear framework, the message meanders: it buries the key facts, hedges on accountability, and fails to answer the questions employees are actually asking — "Does this affect my job?" "Is this temporary?" "What do I do differently starting Monday?"
So you open ChatGPT and type: "Help me write a message to employees about cutting costs." The AI generates a three-paragraph essay that reads like a press release written by a committee. It's polished but completely wrong for your people, your tone, and your specific situation.
The problem isn't the AI. The problem is the prompt.
Without the context that makes your situation specific — the company size, the type of cuts, the timeline, the emotional register you need to hit — the AI has no choice but to produce something generic. And generic cost-reduction communications don't just fail to reassure employees. They actively damage trust, because employees can tell the message was written by nobody for nobody.
A properly constructed prompt changes the output entirely. It tells the AI exactly what kind of company this is, what the reductions actually are, what's not on the table, and what emotional tone will land with your workforce. The result is a draft that sounds like it came from a human leader who thought carefully about the moment — because the prompt forced that careful thinking upfront.
Common mistakes to avoid
Omitting What Is NOT Changing
Employees always assume the worst. If your prompt doesn't instruct the AI to explicitly cover what's off the table — compensation, benefits, current headcount — the output will leave those gaps open, and employees will fill them with fear.
Using a Generic Audience Label
Prompting for 'an employee message' without specifying company size, stage, or employee mix produces one-size-fits-none output. A 60-person startup and a 6,000-person enterprise need radically different registers, detail levels, and accountability structures.
Skipping the Business Rationale
Asking the AI to announce cost cuts without providing the actual reason — market slowdown, missed plan, strategic pivot — forces it to invent a rationale. Invented rationales sound hollow, and employees notice. Always supply the real one-sentence reason.
Requesting 'Positive Tone' Without Defining It
Telling the AI to 'keep it positive' without further instruction produces toxic positivity that minimizes real impact. Specify instead that the tone should be calm, honest, and forward-looking — which is different from reassuring employees that everything is fine when it isn't.
Forgetting the Accountability Close
Most AI-generated cost communications end with vague promises to 'stay in touch.' Without prompting for a specific accountability mechanism — a named cadence, a named owner, a next communication date — the AI will generate the same empty closing every time.
The transformation
Help me write a message to employees about cutting costs at the company.
**Act as a senior executive communications strategist.** Draft an all-employee communication announcing a company-wide cost reduction initiative for a 600-person B2B SaaS company. The initiative will reduce operating expenses by 15% over two quarters through a hiring freeze, travel restrictions, and vendor contract renegotiations — with no planned layoffs. **Audience:** All employees, including individual contributors and managers **Tone:** Direct, honest, and calm — acknowledge difficulty without catastrophizing **Format:** 1. Opening: brief context on why action is needed (market conditions) 2. What we're doing and what we're not doing 3. What this means for each employee day-to-day 4. How leadership will stay accountable and communicate progress 5. Closing: reaffirm confidence in the team **Constraints:** Under 500 words, no corporate jargon, no passive voice. Avoid euphemisms like "right-sizing."
Why this works
Specificity
Naming the exact scale of reductions (15% OpEx), the specific levers (hiring freeze, travel, vendor renegotiation), and the explicit boundary (no layoffs) gives the AI concrete facts to work with rather than generic placeholders. Specific inputs produce specific — and credible — outputs.
Structure
The five-section format acts as a content contract. Each section has a defined job: context, scope, employee impact, accountability, and close. This prevents the AI from blending all five into an undifferentiated block of corporate prose that answers nothing clearly.
Constraint
Hard word limits, banned phrases, and voice requirements are negative instructions — they define the failure modes the AI must avoid. Banning 'right-sizing' and passive voice produces writing that sounds human and direct, not HR-sanitized.
Role Priming
Assigning the AI the role of a senior executive communications strategist shifts its default register from generic assistant to domain expert. This one instruction changes vocabulary choices, paragraph structure, and the assumed audience sophistication across the entire output.
Audience Precision
Distinguishing between individual contributors and managers within the same audience forces the AI to layer meaning — writing a message that gives ICs enough to feel informed while giving managers enough to lead their teams through the Q&A that follows.
The framework behind the prompt
Cost reduction communications fall within a well-established body of organizational change communication theory. The foundational framework here is Kübler-Ross change curve thinking applied to organizational announcements — understanding that employees move through stages of reaction (shock, resistance, exploration, commitment) and that the message must meet them where they are, not where leadership wants them to be.
The SCARF model (Status, Certainty, Autonomy, Relatedness, Fairness), developed by David Rock in neuroleadership research, is particularly useful for structuring cost communications. Cost reduction announcements threaten all five SCARF domains simultaneously — they reduce certainty about job security, threaten status through resource constraints, reduce autonomy through spending controls, and create perceived fairness questions when cuts are uneven. Effective communications explicitly address each domain.
From a structural standpoint, the pyramid principle — developed by Barbara Minto at McKinsey — argues that executive communications should lead with the conclusion, then support it with reasoning. Most leaders do the reverse: they build the case before delivering the verdict, which forces employees to sit through the rationale before learning whether they should be worried.
The most important principle is specificity as a trust signal. Research on organizational trust during uncertainty consistently shows that employees interpret vague communications as concealment. Specific, bounded information — even when it's difficult — is processed as honesty, which is the precondition for the behavioral change cost programs require.
Prompt variations
Act as a startup executive communications advisor.
Draft an all-hands message from the CEO of a 120-person Series B SaaS startup announcing a 15% workforce reduction due to an extended fundraising timeline.
Tone: Honest, human, and accountable — the CEO owns the decision fully Format:
- What happened and why (one paragraph, no jargon)
- Who is affected and what support they receive (severance, references, outplacement)
- What the path forward looks like for those staying
- A direct close that acknowledges the emotional weight without deflecting
Constraints: Under 400 words. First-person voice throughout. No passive voice. No phrases like 'this was a difficult decision' without explaining what made it difficult.
Act as a public company investor relations and executive communications specialist.
Draft an internal leadership communication for a 3,000-person publicly traded company that has just revised its full-year revenue guidance downward by 8% and is announcing a corresponding cost reduction program.
Audience: Top 200 leaders (VPs and above) receiving this before the public earnings call Tone: Candid, strategic, and confidence-building — this group needs to lead others through the news Format:
- What the guidance revision means and why
- The cost program: scope, timeline, and governance
- What leaders should and should not say to their teams before the earnings call
- Q&A prep: the five questions their teams will ask
Constraints: Under 600 words. Include a brief FAQ section. No financial jargon without plain-English translation.
Act as a nonprofit executive communications director.
Draft a staff-wide message from the Executive Director of a 90-person nonprofit announcing a $1.2M budget shortfall due to a federal grant not being renewed. The organization must reduce program spending by 18% over the next fiscal year through a combination of program scope reductions and a pause on two open positions.
Tone: Mission-grounded, honest, and resilient — connect the hard news to the organization's long-term sustainability Format:
- What happened and how leadership learned about it
- The specific budget actions being taken
- What this means for program delivery and staff
- How the organization is working to close the gap (new grant pipeline, fundraising campaign)
- How staff can support and stay informed
Constraints: Under 500 words. Avoid language that distances leadership from the decision.
When to use this prompt
CFO-Led Cost Programs
CFOs partnering with HR and communications teams can use this prompt to align the financial narrative with the employee experience message, ensuring the two don't contradict each other.
Startup Founders in a Down Round
Founders who have just closed a down round or missed a fundraise need to communicate operational changes honestly without triggering talent flight. This prompt structures that difficult conversation.
Division Leaders Managing Budget Cuts
A VP or GM whose division has been handed a 20% budget reduction can use this prompt to communicate the constraints to their team before the cuts are visible in day-to-day operations.
HR Leaders Drafting Manager Talking Points
Chief People Officers can adapt this prompt to generate manager talking points that align with the all-employee message, giving front-line managers the language to handle direct reports' questions.
COOs Managing Efficiency Programs
Operations leaders launching a formal efficiency program need to frame cost discipline as strategy, not crisis. This prompt helps separate the financial rationale from the emotional subtext.
Pro tips
- 1
Specify what is NOT changing, not just what is. Employees anchor on the worst-case scenario when information is absent. Telling the AI explicitly to cover what's off the table forces it to write the reassurance your team needs to hear.
- 2
Include the timeline and sequence of communication events. If managers are being briefed 24 hours before all-staff, tell the AI that context so it can craft language appropriate for each stage of the cascade.
- 3
Name the accountability mechanism. Vague promises to 'keep you updated' erode trust. Prompt the AI to include a specific cadence — weekly leadership Slack updates, a monthly town hall — so the commitment is concrete.
- 4
Provide the business reason in one sentence before asking for the draft. The AI writes better context paragraphs when you give it the actual facts, not a placeholder like 'market conditions.'
A single all-employee message is rarely enough. Effective cost reduction communications run in a cascade — each layer of the organization receives a message calibrated to their role and their audience.
The three-layer cascade:
-
Leadership briefing (24-48 hours ahead): VPs and directors receive the full picture first — including financial details that won't appear in the all-employee message. They need this to answer questions, not to discover the news at the same time as their teams.
-
Manager talking points (same day as all-employee): Front-line managers need a one-page reference document covering the five most likely questions from their teams and the approved language for answering them. This prevents managers from going off-script in ways that create inconsistency.
-
All-employee message: Broad, honest, structured around the employee experience — not the financial mechanics.
When prompting for a cascade, build each layer separately. Reference the previous layer's message in each subsequent prompt so the AI maintains consistency across all three outputs. You can paste the leadership briefing into the manager talking points prompt and say: 'Using the context from this briefing, draft manager talking points for front-line managers whose teams will have these questions...'
This approach prevents the most common failure in cost communications: three different messages that seem to contradict each other.
Research on organizational communication during uncertainty consistently points to the same structural elements that distinguish trusted leaders from those who lose credibility during hard moments.
The five elements that build trust in difficult communications:
-
Acknowledgment before action. Before explaining what the company is doing, the message must acknowledge the emotional reality of the moment. One sentence that names the difficulty — 'I know this news is unsettling' — is not weakness. It's what separates human communication from corporate communication.
-
Specificity over vagueness. 'We're making some adjustments' destroys trust. '15% reduction in operating expenses through a hiring freeze and travel restrictions' builds it, even though it sounds harder. Employees trust specifics because vagueness signals there's more being hidden.
-
What's off the table. Employees default to catastrophic assumptions. Explicitly naming what is NOT happening — no layoffs, no compensation changes, no benefit cuts — reduces anxiety more than any positive framing.
-
An accountability mechanism with a name and a date. 'We'll keep you informed' means nothing. 'I'll send a monthly update on the first Monday of each month, and we'll hold a Q&A town hall in 30 days' means something.
-
A forward-looking close that doesn't minimize. The close should acknowledge the hard work ahead without pretending it isn't hard. 'I believe we can navigate this together' lands better than 'I'm confident we'll emerge stronger' — one sounds earned, the other sounds like a script.
When you build your AI prompt, map each of these five elements to a section instruction. The AI will produce structurally sound output if your prompt makes these elements non-optional.
Before you send the AI-generated draft to your communications team or legal review, run it through this checklist.
Accuracy
- [ ] All specific numbers (percentages, timelines, dollar amounts) match your approved plan
- [ ] No details appear that haven't been confirmed internally
- [ ] The message doesn't imply decisions that haven't been made yet
Tone
- [ ] The opening acknowledges difficulty without catastrophizing
- [ ] No passive voice constructions that obscure who made the decision
- [ ] No euphemisms that employees will read as spin (right-sizing, streamlining, optimization)
- [ ] The close sounds like a human, not a press release
Completeness
- [ ] Explicitly covers what is NOT changing
- [ ] Names a specific accountability mechanism with a date
- [ ] Answers: 'What does this mean for me personally?'
- [ ] Answers: 'Is my job safe?'
- [ ] Answers: 'What do I do differently starting tomorrow?'
Alignment
- [ ] Consistent with what your manager-level cascade will say
- [ ] Consistent with any public statements (press release, investor communication)
- [ ] Legal and HR have reviewed any statements about severance or job security
Final read
- [ ] Read it out loud. If it sounds like a robot wrote it, revise the prompt and regenerate.
- [ ] Would you feel respected receiving this message? If not, it needs more work.
When not to use this prompt
This prompt pattern is not the right tool when cost reductions are still undecided or subject to board approval. Communicating prematurely — even with hedged language — creates expectation and rumor that's harder to correct later. In those cases, use a "preparing for the communication" prompt to develop your messaging framework in advance, without triggering the communication itself.
This prompt is also not a substitute for individual conversations with directly impacted employees. Use it to build the all-company or all-team layer of the communication cascade, then write separate, personalized language for the hardest individual conversations.
Troubleshooting
The AI output sounds too formal and corporate, not like the leader's actual voice
Add a voice instruction to your prompt: 'Match this writing style:' and paste in 2-3 paragraphs of the leader's previous communications — a past all-hands email, a Slack message, a published post. The AI will calibrate its vocabulary, sentence length, and register to match the sample. Also explicitly ban phrases like 'in these challenging times' and 'going forward.'
The draft buries the most important information in the middle or end
Add a structural constraint: 'The single most important piece of information — [what it is] — must appear in the first 50 words.' Leaders instinctively soften hard news by warming up to it. The AI mirrors this pattern unless you override it. Front-loading the key fact respects your employees' time and builds credibility.
The AI generates generic accountability language with no specifics
Add explicit accountability parameters to your prompt: 'Include a named communication cadence — specify the frequency, the channel, and the owner by title.' If you want the AI to commit to a monthly update email from the CEO, say that explicitly. Vague accountability instructions produce vague accountability language every time.
How to measure success
A strong AI output from this prompt will have five measurable qualities. First, the key information — what is changing, what is not changing — appears in the first 100 words, not buried in paragraph four. Second, the word count falls within 10% of your specified limit without the message feeling truncated. Third, no passive voice constructions appear — every action has a named subject. Fourth, the accountability mechanism names a specific cadence, channel, and owner. Fifth, reading the draft aloud takes under two minutes at a normal speaking pace, which signals appropriate length and sentence complexity for a broad employee audience.
Now try it on something of your own
Reading about the framework is one thing. Watching it sharpen your own prompt is another — takes 90 seconds, no signup.
a cost reduction all-employee communication
Try one of these
Frequently asked questions
Yes — and you should adjust the audience specification accordingly. Change 'all employees' to your specific team and add context about how your department's situation fits the broader company picture. Department-level communications often need more operational detail and less macro context.
Be honest with the AI about what's confirmed versus in progress. Prompt it to draft the message in a way that acknowledges some decisions are still being finalized, with a specific date when you'll share more. This is actually more credible than pretending all details are settled.
Add an explicit section to the format instructions covering affected employees: what notification process they'll receive, what severance applies, and the timeline. Tell the AI to separate the layoff information from the forward-looking operational changes so the two messages don't blur together.
Absolutely — and it's worth running a separate prompt for that. Managers need different content: answers to the hardest questions, language for 1:1 conversations, and guidance on what to escalate versus handle locally. Reference the all-employee message in that prompt so both outputs stay aligned.
They will if you don't constrain the tone explicitly. Add instructions like 'write in the CEO's first-person voice,' 'avoid corporate jargon,' and 'do not use passive voice.' You can also paste in a short sample of the leader's writing style for the AI to match.