Demand Forecasting & Capacity Planning
Build demand forecasts with confidence intervals and scenario analysis, calculate capacity requirements, identify bottlenecks, and model capital needs for scaling infrastructure to support growth targets.
Use This When
Planning, analysis, client strategy sessions, decision support.
Inputs Needed
Business model, goal, constraints, market, competitors, budget, timeline, internal capabilities.
Expected Output
Executive summary, diagnosis, options, risks, recommended path, implementation plan, KPIs.
The Workflow Prompt
You are a business strategist and operator. Objective: Demand Forecasting & Capacity Planning Context: Build demand forecasts with confidence intervals and scenario analysis, calculate capacity requirements, identify bottlenecks, and model capital needs for scaling infrastructure to support growth targets. Original task: **Act as a demand forecasting analyst. Historic demand data: [TIME_SERIES_DATA]. Current demand drivers: [DRIVERS]. Future plans: [PRODUCT_LAUNCHES, MARKETING_SPEND, SEASONALITY]. Your task:(1) Build a demand forecast for the next [PERIOD] accounting for seasonality, trends, and growth drivers(2) Calculate confidence intervals around forecast(3) Identify key variables that have highest impact on demand(4) Model scenarios: base case, bull case, bear case(5) Determine required capacity (staff, infrastructure, inventory) to meet demand(6) Identify constraint points (bottlenecks) under different scenarios(7) Calculate capital requirements for capacity expansion. Use statistical methods: time series analysis, regression, scenario modeling. Account for: historical patterns, market trends, competitive actions, internal initiatives. For each scenario, calculate:(1) Resource requirements(2) Capital needs(3) Timeline for decisions/investments. Present as: Historical Analysis → Forecast Methodology → Base Case Forecast with Confidence Intervals → Scenario Analysis (Bull/Base/Bear) → Capacity Requirements → Bottleneck Analysis → Capital Planning → Recommendation on Capacity Investments → Key Assumptions & Sensitivities. Make it practical for budgeting and planning cycles.** Inputs I may provide: Business model, goal, constraints, market, competitors, budget, timeline, internal capabilities. Operating instructions: - First, restate the objective in one clear sentence. - If critical information is missing, ask up to 5 focused questions. If there is enough information to proceed, make practical assumptions and label them. - Use a Detailed response style. - Be specific to the business, audience, channel, and constraints provided. - Avoid generic AI advice. Give concrete recommendations, examples, templates, copy, or steps I can use. - When current facts, competitors, laws, prices, policies, or market claims matter, use current research and cite sources. - Do not expose hidden chain-of-thought. Provide a concise rationale or decision summary instead. - End with a short QA checklist that helps me verify the output. Required output: Executive summary, diagnosis, options, risks, recommended path, implementation plan, KPIs. Caution: Do not treat output as professional legal, medical, financial, or compliance advice; verify with a qualified expert. Use live web research or source documents before finalizing claims.
QA Follow-Up Checklist
After the AI returns its output, verify against:
- Output is specific to the provided business/context.
- Assumptions are clearly labeled.
- No unsupported claims without source checks.
- Next actions are clear and usable.
Follow-Up Prompt
Now turn the result for 'Demand Forecasting & Capacity Planning' into a client-ready version: tighten wording, remove fluff, add missing assumptions, and provide the next 3 actions.
Avoid / Cautions
Do not treat output as professional legal, medical, financial, or compliance advice; verify with a qualified expert. Use live web research or source documents before finalizing claims.
How Different Verticals Use This Workflow
Restaurant & Hospitality
A 14-location fast-casual brand planning a new commissary kitchen needs to know if 8,000 sq ft handles 2027 growth or if they should sign for 12,000. They load three years of unit-level sales by daypart, a CapEx model for both build sizes, and the planned eight new openings. The output shows the 8K kitchen tops out by Q3 2027 unless they shift catering production off-site — so they sign 12K with a sublease clause and stop re-litigating the decision every leadership meeting.
Retail & E-commerce
A home goods DTC brand doing $18M is staring at Q4 and trying to decide whether to commit to 3PL capacity for 2.5x or 3.5x peak. They feed in three years of weekly orders, the marketing spend ramp, and current SKU velocity. The bear case shows 3.5x is a $400K waste; the base case says 2.5x leaves them stocking out from Black Friday through Dec 15. The answer is 2.8x with a pre-negotiated overflow agreement at premium rates — they ship the trigger memo in the same week.
Professional Services & B2B
A 40-person consulting firm is debating headcount for FY27. They feed in win-rate trends, current pipeline by stage, average project duration, and partner-level utilization. The model surfaces that the bottleneck isn't sellers — it's senior delivery, and the bull case requires hiring two principals nine months before revenue confirms it. The output becomes a hiring trigger document: 'hire principal #1 when Q3 pipeline crosses $4.2M weighted.'
Beauty & Personal Care
A clean beauty brand launching into 240 Sephora doors needs to forecast door-level sell-through to avoid the classic indie failure of running out in month two. They input current DTC velocity, comparable brand launch curves, and Sephora's tier-store mix. The model produces a per-door weekly forecast with confidence intervals plus the inventory build needed to hit 95% in-stock through month four — and flags the three SKUs likely to overshoot so they don't tie up working capital on slow movers.
Local & Trade Services
A regional pool installation company is debating whether to hire two more crews for next season or push prices and stay at three. They feed in inquiry-to-close conversion, install duration by pool type, weather-pattern data, and the marketing pipeline. The forecast shows demand supports four crews in May-August but only three for the rest of the year — recommendation is one full-time hire plus a seasonal sub-contractor agreement, not two full crews carrying overhead through January.
Frequently Asked
What level of historical data does this actually need to produce a real forecast?
At minimum 18 months of weekly data so seasonality is visible. With less than that, you're guessing trend slope from noise. If you only have 6-12 months, force the model to label everything as 'directional, not predictive' and pair it with at least two leading indicators (search volume, ad-account spend trajectory, partner pipeline) to triangulate.
What's the most common way capacity planning blows up after the forecast lands?
Treating bull/base/bear as equally likely. They're not. Most ops teams over-build for bull case because nobody wants to be the bottleneck on a good month, then carry 35-40% excess capacity through a normal quarter. Force the forecast to assign explicit probabilities (60/30/10 base/bear/bull is realistic for most B2B), then build for P70 demand with documented overflow plans for the upside.
How do I make the output useful for a CFO budget cycle vs. internal ops?
For the CFO: scrap the methodology section, lead with the capital ask split across quarters and a single sensitivity table showing what happens if demand undershoots by 20%. For ops: keep the bottleneck analysis, week-by-week staffing curves, and the decision triggers for when to hire vs. when to delay. Same data, two completely different documents — don't ship the same PDF to both.
Should I use this for a brand-new product launch?
No — there's no time series to forecast from. Use the analogue method instead: find three comparable launches in your category, model demand off their first 24 months, and run this prompt only after you have 90 days of your own actuals. Using time-series methods on a launch with three data points produces a number that's worse than your gut.