The work product expectation in consulting is unambiguous: everything you put in front of a client — proposal, readout, situation assessment, change management plan, close-out report — has to be tight. Logically structured, clearly argued, no wasted words, no hedging where you should have a point of view. The standard isn't "good enough." It's "this looks like it came from us."
That standard is what makes the writing load in consulting so relentless. You're not just producing a lot of documents — you're producing a lot of documents that need to be right. The work plan, the stakeholder analysis, the post-meeting readout, the slide storyboard, the recommendation narrative — all of it gets scrutinized. By the client. By your principal. By the engagement manager who will rewrite it if it comes back below the line.
Claude is a large language model that's very good at one specific thing: taking structured input — notes, data, analysis, context — and producing polished, well-organized prose. It does not have access to your firm's systems. It cannot connect to your engagement management platform, pull from your proprietary databases, access client data, retrieve anything from your internal knowledge management system, or look up industry benchmarks. It cannot access McKinsey proprietary research, Gartner data, IBISWorld, or any licensed database. It is strictly a drafting and structuring tool. You supply the facts, the frameworks, and the client context. Claude handles the prose and the structure. The analytical judgment is still yours. The 90-minute blank-page problem is not.
That distinction matters in consulting more than almost anywhere else. A fabricated benchmark statistic in a client deck is a career event. Claude will flag uncertainty rather than fill a gap with a plausible-sounding number — which is the right behavior. Every use case below includes "paste in your notes" or "provide the key findings you've already developed" for exactly that reason. You own the analysis. Claude writes the document.
1. Client Proposal / Pitch Narrative
Proposals are where the "why us, why now, why this approach" story lives — and they're where the writing bar is highest because the client is evaluating you as much as the proposal. The thinking is done: you've had the discovery call, you've scoped the engagement, you've structured the approach, you know the team. The proposal document is the prose layer over a set of decisions you've already made.
Paste your engagement description, proposed approach, team credentials, and fee structure. Claude drafts the proposal sections.
I'm writing a client proposal for [client name or type] for an engagement focused on [engagement description — topic, scope, business problem]. Here's what I have:
Situation and client context: [paste your understanding of the client situation — problem they're trying to solve, what's at stake, any background on why this is urgent now]
Proposed approach: [paste your phased approach — phases, key activities, workstreams, deliverables per phase]
Team: [paste team member names, titles, and relevant credentials or past engagement highlights]
Fees: [paste your fee structure — total, by phase, retainer, or however structured]
Write polished proposal sections including:
- Executive Summary (150–200 words): situation, our recommended approach, why this is the right time to act, expected value at stake — written to the decision-maker who may read only this section
- Situation Overview: the business problem as we understand it — specific, not generic; frame the stakes
- Proposed Approach: phased engagement narrative — what happens in each phase, what we're doing and why, what the client can expect at each milestone; not a bullet list — a narrative
- Team Credentials Summary: a 2–3 sentence narrative per team member focused on relevant experience for this engagement, not a CV
- Value at Stake: the "why this matters financially / strategically" frame — quantified where the data supports it, directional where it doesn't
Tone: confident, peer-to-peer with a senior executive. Not salesy. Not hedged. Direct.
The "not salesy" instruction matters. Experienced buyers read through proposal language that sounds like it was written to close a deal. The best proposals sound like a senior partner talking to a CFO — direct, specific, and uninterested in impressing anyone. Claude defaults to that register when you ask for it.
2. Situation Assessment / Diagnostic Document
The kickoff readout is the first impression the client forms of how you think. If your situation assessment is sharp — key observations well-categorized, hypotheses clearly stated, issue tree defensible — you've earned credibility before the analytical work begins. If it's a bullet dump of interview notes, you're already managing expectations.
Paste your interview notes, data observations, and initial hypotheses. Claude produces the structured situation assessment.
I've completed the discovery/diagnostic phase for [engagement name / client]. Here's the raw material: [paste your interview notes, workshop outputs, data observations, any quantitative findings — as much detail as you have].
My initial hypotheses going into this are: [paste your working hypotheses — even rough ones].
Write a structured situation assessment document for the kickoff readout, including:
- Key Observations: categorized by theme (e.g., organizational, operational, financial, market — use the categories that fit this engagement), with a one-sentence "so what" after each observation cluster
- Emerging Hypotheses: 3–5 hypotheses we're testing, stated as assertions ("The root cause is X" not "It might be X"), each with the evidence from observations that supports it and the evidence that could refute it
- Preliminary Issue Tree: top-level problem statement broken into 3–4 MECE branches, each with 2–3 sub-issues; label which branches are data-supported vs. hypothesis-only
- Recommended Next Analytical Steps: 4–6 specific analytical workstreams with rationale — what question each answers and why it's the priority
Do not invent observations or data not present in the notes I provided. Flag any hypotheses where evidence is thin.
The instruction to flag thin evidence is load-bearing. You want to know before the readout which hypotheses are well-supported and which are directional — not find out when the client pushes back on slide 4. For the analytical work that feeds this document, Claude AI for Business Analysts covers the data synthesis and findings documentation layer.
3. Executive Meeting Readout
Every substantive client meeting needs a readout. Attendees, key discussion points, decisions made, open issues, action items with owners and due dates. The engagement manager sends it the morning after. It is a document that takes 30–45 minutes to write from meeting notes and that people rarely read with full attention — but will reference constantly when something is disputed three weeks later.
Paste your meeting notes, decisions, and action items. Claude produces the clean readout memo.
I need to write a readout memo for [meeting name/date] with [client name]. Here are my meeting notes: [paste your notes — attendees, discussion points, what was decided, what was raised but not resolved, who committed to what].
Write a clean readout memo including:
- Meeting Details: date, location/format, attendees with titles (client and consulting team)
- Key Discussion Points: 4–6 substantive points — what was discussed and the key perspectives raised; not a transcript, a synthesis
- Decisions Made: numbered list — only actual decisions, not discussions or open items
- Open Issues / Parking Lot: items raised but not resolved, with a note on whether they need to be resolved and by when
- Action Items: table format — Action | Owner (name and affiliation) | Due Date | Dependencies
Tone: factual, professional, written as a neutral record of what happened. This will be distributed to all attendees.
The separation of "Decisions Made" from "Open Issues" is the part that saves the most confusion downstream. Getting that right in the readout prevents three rounds of re-discussion in two weeks. For executive assistants who manage meeting logistics and readout distribution workflows, Claude AI for Executive Assistants covers the adjacent operational layer.
4. Change Management Communication Plan
Change management deliverables are some of the most structurally complex documents consultants write — and among the most frequently templated badly. A real comms plan needs a stakeholder table that reflects actual concerns and actual behavior change targets, not generic categories. That table is time-consuming to build from scratch.
Paste your project overview, stakeholder map, and timeline. Claude produces the structured comms plan.
I'm building a change management communication plan for [project name] at [client]. Here's the context: [paste your project overview — what's changing, why, timeline, what the desired end state looks like].
Stakeholder groups: [paste your stakeholder map — groups, their current state, what you need them to do differently, any known concerns or resistance signals].
Timeline: [paste key milestones — go-live dates, decision gates, training windows, announcement dates].
Write a structured change management communication plan including:
- Stakeholder Groups Table: columns — Stakeholder Group | Key Concerns | Desired Behavior Change | Recommended Channel | Frequency | Message Owner. Populate based on the stakeholder notes I provided.
- Communication Milestones: timeline of key communication moments mapped to project milestones — what gets communicated, to whom, when, and in what format
- Resistance Risk Flags: 3–5 specific resistance scenarios based on the stakeholder notes, with recommended mitigation for each
- CEO/Leadership Messaging Guidance: the 3–4 key messages leadership needs to reinforce consistently, written in plain language (not consulting language) — the talking points for the sponsor's town hall
Do not invent stakeholder concerns not present in the materials I provided.
The "in plain language" instruction for leadership messaging is worth emphasizing. The messages leadership actually delivers to employees are shorter, plainer, and more personal than anything consultants write in a deliverable. Claude knows the difference when you specify the register.
5. Slide Storyboard Narrative
The deck isn't the problem. The logic architecture is. Every experienced consultant knows the feeling: three days into build, you realize the deck doesn't have a spine — the slides are good individually but the argument doesn't land as a whole. The storyboard is the thing that prevents that.
Paste your core argument, supporting points, and data bullets. Claude builds the written storyboard — the logic that the deck expresses.
I'm building a presentation for [audience — e.g., CEO and C-suite, board, client steering committee] on [topic]. Here's the core argument I'm making: [state your "so what" — the single thing you want the audience to believe or decide after this deck].
Supporting points: [paste your 4–6 supporting points — the building blocks of the argument].
Data and evidence: [paste your key data points, findings, and examples that support each point].
Write a slide storyboard including, for each slide:
- Slide Title: the assertion, not the topic (e.g., "Operational costs are 23% above benchmark" not "Cost Analysis")
- One-sentence "So What": the single takeaway this slide needs to leave
- Supporting Points: 3–4 bullet-level points that prove the slide title
- Transition: one sentence connecting this slide's "so what" to the next slide's argument
At the end, write a 2–3 sentence spine statement: the through-line of the whole deck, the argument in plain language.
This is the storyboard a principal reviews before the team builds anything — written logic architecture, not slide titles.
The assertion-not-topic instruction for slide titles is the single biggest structural habit that separates strong decks from weak ones. Claude will apply it consistently when you ask for it — and will call out titles that are still topics rather than assertions if you ask it to review the storyboard critically. For product managers and operations leaders who receive strategy decks as stakeholders, Claude AI for Product Managers covers the narrative communication layer from that side.
6. Project Close-Out Report
Close-out reports are the document everyone knows matters and no one has time to write well by the time the engagement ends. The client is already asking about next steps. The team is halfway to the next project. The report gets produced under time pressure and often reads like it.
A good close-out report is worth writing well — it's the document that informs the follow-on proposal, gets referenced in new business credentials, and shapes the relationship conversation for the next twelve months. Give Claude the engagement facts and it produces the structured report.
I'm writing a project close-out report for [engagement name] at [client]. Here's the engagement overview: [paste your engagement summary — objectives, scope, timeline, team, fees].
Deliverables produced: [paste your deliverable list — what was built, presented, or handed off].
Outcomes: [paste what actually happened — did objectives get met? What results were achieved? Any measurable impact?].
Lessons learned: [paste your honest read — what worked well, what would you do differently, what the client did well, where the engagement had friction].
Write a structured close-out report including:
- Engagement Summary: scope, timeline, team, fee summary — factual
- Objectives vs. Outcomes: table — Objective | Target | Actual Outcome | Status (Met / Partially Met / Not Met) — honest
- Key Deliverables: table — Deliverable | Date Delivered | Recipient | Status / Notes
- Lessons Learned: two columns — What Worked (keep doing) | What to Do Differently (process, scope, client management, team structure)
- Recommended Follow-on Work: 2–3 specific opportunities for continued engagement, grounded in what you learned during this engagement — not generic upsell language
- Client Relationship Status: honest 3–4 sentence read on where the relationship stands, who the key champions are, and what the next conversation should focus on
Tone: factual and direct. This is an internal document — write it like a partner debrief, not a client deliverable.
The "write it like a partner debrief" instruction matters. Internal close-out documents that are written in client-facing language don't get read. The lessons learned section especially needs to be honest — and Claude will write it honestly if you give it honest input. For finance teams who analyze engagement economics and project ROI, Claude AI for Finance Professionals covers the numbers side of that post-engagement review.
Why Claude Over ChatGPT for Consulting Work
100K+ context window. Paste a full set of interview transcripts — all twelve stakeholder interviews, 15,000 words of notes. Paste the full prior-phase deliverable. Paste the complete data file. Claude processes the entire document in a single session. No "please summarize first," no context truncation mid-analysis, no having to break a long engagement document into pieces. The full context is in scope for every output. That matters when you're synthesizing across a full diagnostic.
Conservative and attribution-honest. This is the one that matters most in consulting. Claude will not fabricate a benchmark statistic to support a hypothesis. It will not invent industry data you didn't provide. It will not produce a specific percentage or dollar figure that sounds authoritative but has no basis in what you gave it. When the data is thin, it says the data is thin. In a field where a fabricated number in a client deck is a career-level event, that conservatism is a feature. Claude vs ChatGPT: Which AI Is Better for Work covers this distinction in detail.
Projects per engagement. Build one Claude Project per client engagement. Load the proposal, the work plan, the client org chart, the interview guides, and the interview notes at the start. Every document you produce from that point — situation assessment, meeting readout, change management plan, close-out report — is produced with the full engagement context available. Context compounds across the lifecycle. By the time you're writing the close-out, Claude knows the original objectives, the Phase 1 findings, and the stakeholder dynamics you documented in week two.
Structured output fidelity. Multi-column stakeholder tables render correctly. Slide storyboards with slide-by-slide titles, "so whats," and transition sentences stay structured across 20-slide decks. Close-out reports with four separate tables don't collapse into prose. Claude holds complex document architecture across long outputs — which is what consulting deliverables actually require.
Practical Tips for Consultants
One Project per engagement. Load the proposal, work plan, client org chart, and all interview notes into the project system prompt at the start of the engagement. Every document you produce from that point benefits from full engagement context. Context compounds — the close-out report gets better when Claude has been tracking the engagement from the beginning.
Data in / narrative out. Claude structures what you supply. You are still the analyst. Never paste a prompt without the underlying facts, notes, or findings. The analytical judgment is yours. The blank-page problem is Claude's. For operations managers who use Claude for process documentation and analytical work, Claude AI for Operations Managers covers that parallel workflow.
Specify the reader. "This is going to the CFO who is skeptical of the initiative and will push back on the cost assumptions" produces meaningfully different output than "this is for the project team." Claude adjusts the framing, the level of supporting evidence, and the tone when you specify who's reading and what their prior is. Always include the reader and their context.
Use it for the blank page, not the final polish. The value is in going from nothing to a structured 80% draft in 20 minutes — not in micro-editing prose that's already good. The blank-page problem is where the time goes. Once there's a structured document on screen, your judgment and edits are the value-add. Claude gets you to the screen with something on it.
The Playbook
If you want every prompt above — refined, tested, and organized by consulting workflow — plus prompts for work plan development, client interview guides, hypothesis testing frameworks, board presentations, and the full library of professional use cases across every major business function, that's what The Complete Claude Playbook is.
$27. Instant access. The prompts your client work actually needs.