Skip to content
seasonality-forecasting logo

Seasonality Forecasting

seasonality-forecasting

Identifies, quantifies, and forecasts seasonal demand patterns for CPG and retail products to optimize inventory, promotions, and assortment timing. Use when the user wants to understand seasonal trends, build seasonal demand forecasts, plan seasonal inventory, or optimize the timing of product l...

writer/skills0installs8stars

SKILL.md

Full skill instructions

Seasonality Forecasting

Overview

Seasonality Forecasting decomposes historical demand into trend, seasonal, and residual components to produce accurate forecasts that account for predictable demand fluctuations. In CPG and retail, seasonality is the most reliable and actionable demand signal—it repeats annually with high fidelity and can be planned for months in advance.

Failure to properly account for seasonality is the single largest source of preventable stockouts and overstock in retail. Seasonal peaks missed by 1–2 weeks result in lost sales; seasonal inventory not cleared by markdown timing erodes margin. This skill delivers precise seasonal profiles and actionable timing recommendations.

When to Use

  • Annual demand planning and open-to-buy budgeting
  • Inventory pre-build decisions for seasonal peaks
  • Promotion calendar timing aligned to seasonal demand cycles
  • Seasonal assortment transitions (e.g., summer → fall planogram reset)
  • New product launch timing optimization
  • Markdown and clearance timing for seasonal inventory
  • User provides historical sales data and asks about seasonal patterns

Required Inputs

InputRequiredDescription
Historical sales dataYesWeekly or monthly units/​revenue; minimum 2 full years (104 weeks), ideally 3+ years
Product hierarchyYesCategory, subcategory, brand, SKU for aggregation levels
Calendar dataRecommendedHoliday dates, promotional calendar, store events
Weather dataOptionalRegional temperature, precipitation for weather-sensitive categories
Macro indicatorsOptionalEconomic data, school calendars, cultural events for context
New product flagsOptionalLaunch dates for items without full history

Methodology

Step 1: Data Preparation and Cleaning

  1. Aggregate to appropriate level: SKU-level for high-volume items, subcategory or brand level for slower-moving items (need ≥ 50 units/​week for reliable SKU-level seasonality)
  2. Align to consistent calendar: Convert to ISO weeks or 4-5-4 retail calendar for comparable periods
  3. Adjust for anomalies: Remove or flag COVID-period data (2020-2021), extreme weather events, supply disruptions, stockout periods
  4. Normalize for distribution: If store count changed, convert to per-store-per-week metrics before decomposition
  5. Handle holiday shifts: Easter, Ramadan, Lunar New Year, and other movable holidays must be aligned by event, not calendar date

Step 2: Time Series Decomposition

Decompose the demand signal into three components:

Y(t) = Trend(t) × Seasonal(t) × Residual(t)   [Multiplicative model]
Y(t) = Trend(t) + Seasonal(t) + Residual(t)    [Additive model]

Choose multiplicative when seasonal amplitude grows with trend (most CPG categories). Choose additive when seasonal amplitude is constant regardless of trend level.

Decomposition methods (in order of recommendation):

  1. STL (Seasonal-Trend decomposition using LOESS): Robust, handles outliers, flexible seasonal period
  2. X-13ARIMA-SEATS: Gold standard for official statistics; handles trading-day effects and holiday effects
  3. Classical decomposition: Moving-average based; simple but less robust to outliers
  4. Prophet: Bayesian approach; good with multiple seasonality (weekly + annual) and holiday effects

Extract the Seasonal Index for each period:

Seasonal Index(week w) = Average Seasonal Component(week w) / Grand Mean of Seasonal Component

An index of 1.20 means demand in that week is typically 20% above average.

Step 3: Seasonal Profile Construction

Build a seasonal profile (52-week or 12-month index) for each product/​category:

WeekSeasonal IndexClassificationPlanning Implication
W1–W4 (Jan)0.85TroughPost-holiday clearance; lean inventory
W5–W8 (Feb)0.90Below averageValentine's spike in confection/​floral only
W18–W22 (May)1.15ShoulderPre-build for summer peak
W23–W35 (Jun-Aug)1.35PeakMaximum inventory; increase promotional activity
W44–W48 (Nov)1.40Holiday peakThanksgiving/​Black Friday; maximum coverage
W49–W52 (Dec)1.55Maximum peakHoliday gifting; risk of overstock post-Dec 25

Peak-to-Trough Ratio: Measure of seasonality intensity:

PT Ratio = Max Seasonal Index / Min Seasonal Index
  • PT < 1.3: Low seasonality (staples: paper towels, dish soap)
  • PT 1.3–2.0: Moderate seasonality (most CPG food & beverage)
  • PT 2.0–4.0: High seasonality (sunscreen, allergy meds, grilling)
  • PT > 4.0: Extreme seasonality (holiday candy, pool chemicals, de-icer)

Step 4: Demand Forecasting

Combine trend projection with seasonal profile:

Forecast(t) = Projected Trend(t) × Seasonal Index(t) × Adjustment Factors

Trend projection methods:

  • Linear trend: Adequate for stable categories
  • Exponential smoothing (Holt-Winters): Captures level and trend with adaptive weighting
  • Growth curves: For categories in growth or decline phases

Adjustment factors:

  • Planned promotional lifts (from promotion calendar)
  • Distribution changes (new store openings, channel expansion)
  • Known external events (Olympics, World Cup, elections)
  • Macro-economic adjustments (inflation impact on volume)

Forecast accuracy metrics:

MAPE = (1/​n) × Σ |Actual − Forecast| / Actual × 100
WMAPE = Σ |Actual − Forecast| / Σ Actual × 100
Bias = Σ (Forecast − Actual) / Σ Actual × 100

Target: MAPE < 15% at weekly category level; < 25% at weekly SKU level.

Step 5: Inventory and Timing Recommendations

Translate the forecast into operational decisions:

Safety Stock by Season:

Safety Stock = z × σ_demand × √(Lead Time + Review Period)

Increase z-score (service level) during peak season to prevent stockouts on high-velocity items.

Pre-Build Timing: Start inventory build Lead Time + 2 weeks before the seasonal ramp begins.

Markdown Timing: Initiate seasonal clearance when:

Weeks of Supply > Remaining Weeks in Season × 1.5

Assortment Transition Windows: Recommend dates for:

  • Seasonal set introduction (e.g., summer grilling set goes live Week 18)
  • Core assortment flex (expand/​contract shelf space to match seasonal index)
  • End-of-season clearance start and target completion dates

Step 6: Category-Specific Seasonal Drivers

Map external drivers to seasonal patterns for causal understanding:

DriverAffected CategoriesIndex ImpactData Source
Temperature > 80°FBeverages, ice cream, sunscreen+25–60%Weather API
Back-to-school (Aug)Snacks, lunch items, school supplies+15–30%School calendar
Cold/​flu season (Nov-Feb)OTC medicine, tissues, soup+40–80%CDC flu data
Grilling season (May-Sep)Charcoal, condiments, hot dogs, buns+50–120%Temperature data
Holiday baking (Nov-Dec)Flour, sugar, butter, baking mixes+80–200%Calendar
Super Bowl (early Feb)Chips, dips, beer, frozen appetizers+100–300% (event week)Event calendar

Output Specification

1. Seasonal Profile Report

52-week (or 12-month) seasonal index chart and table for each product/​category analyzed.

2. Demand Forecast

Point forecast with prediction intervals (80% and 95%) for the planning horizon (typically 13–52 weeks).

3. Seasonal Classification

Categorization of each product/​category by seasonality intensity (low/​moderate/​high/​extreme) with peak-to-trough ratio.

4. Operational Calendar

EventCategoryTimingActionInventory Implication
Summer ramp beginsBeveragesWeek 18Increase orders+35% vs. baseline
Peak weekSunscreenWeek 27Maximum coverage4 weeks forward cover
Markdown startSummer seasonalWeek 34Begin clearanceTarget 90% sell-through by W38

5. Forecast Accuracy Assessment

Backtesting results showing MAPE, bias, and hit rate for prior-year seasonal forecasts.

Analysis Framework

Key Metrics

  • Seasonal Index: Ratio of expected demand in a period to average demand (1.0 = average)
  • Peak-to-Trough Ratio: Measures seasonality amplitude
  • MAPE / WMAPE: Forecast accuracy
  • Forecast Bias: Systematic over/​under-forecasting tendency
  • In-Stock Rate During Peak: Target ≥ 98% for top seasonal items
  • Seasonal Sell-Through Rate: % of seasonal inventory sold at full price (target: ≥ 75%)
  • Markdown Depth on Seasonal: Average markdown % required to clear (lower is better; target: < 30%)
  • Lost Sales Rate: Estimated revenue lost due to seasonal stockouts (target: < 2% of peak revenue)

Forecasting Accuracy Benchmarks

LevelGoodAcceptableNeeds Improvement
Monthly, category< 8% MAPE8–15%> 15%
Weekly, category< 15% MAPE15–25%> 25%
Weekly, SKU< 25% MAPE25–40%> 40%
Event weeks (holidays)< 12% MAPE12–20%> 20%

Examples

Input: "Analyze seasonal patterns for our Grilling & Outdoor Cooking subcategory. We have 3 years of weekly POS data for 85 SKUs across charcoal, lighter fluid, grilling sauces, and outdoor cooking accessories. Build a seasonal forecast for the next 12 months."

Output:

  1. Seasonal Profile: Extreme seasonality (PT Ratio 5.2). Season runs Week 14–Week 38 (early April through late September). Peak: Week 26–28 (early July, around July 4th). Secondary peak: Week 22 (Memorial Day).
  2. Seasonal Indices: Jan 0.18, Feb 0.22, Mar 0.55, Apr 1.05, May 1.65, Jun 1.85, Jul 2.10, Aug 1.45, Sep 0.90, Oct 0.45, Nov 0.25, Dec 0.15
  3. Forecast: Total category forecast $14.2M (+3.8% vs. prior year driven by trend growth in premium grilling sauces). Peak month July: $2.98M. MAPE on backtested 2024 forecast: 11.4%.
  4. Calendar: Pre-build starts Week 10 (early March) with initial orders for charcoal. Full seasonal set in-store by Week 14. First markdown Week 34 (late August) for accessories; charcoal clearance Week 36. Target 85% sell-through before Week 38.
  5. Weather risk: If average June temperature is > 5°F above normal, add 8–12% to June-July forecast.

Guidelines

  • Minimum 2 full years of data for basic seasonal patterns; 3+ years for reliable indices and confidence intervals
  • At the SKU level, aggregate very low-volume items to subcategory level before decomposition—noise overwhelms signal below ~50 units/​week
  • Always check for structural breaks: consumer behavior shifts (e.g., year-round demand for formerly seasonal products), new competitors, or category redefinition
  • Movable holidays (Easter, Ramadan, Diwali, Chinese New Year) must be event-aligned, not calendar-aligned, for accurate seasonal indices
  • Separate the promotional signal from the seasonal signal — a promotion during peak season inflates the seasonal index if not controlled
  • Consider climate change trends for weather-sensitive categories; historical averages may understate warming trends
  • For new products without history, inherit the seasonal profile from the most similar existing product or subcategory
  • Cross-validate forecasts by comparing top-down (category level) with bottom-up (sum of SKU forecasts); reconcile differences
  • Flag any year used in the seasonal average that had material disruptions (COVID, supply chain, weather extremes)
  • Include prediction intervals in forecasts — point estimates alone are insufficient for inventory planning

Validation Checklist

  • Minimum 2 full years of clean, comparable data are used
  • Decomposition method is appropriate (multiplicative vs. additive) and stated
  • Anomalous periods are identified and handled (excluded or adjusted)
  • Seasonal indices sum/​average to approximately 1.0 over a full year
  • Peak-to-trough ratio is calculated and seasonality is classified
  • Holiday and event effects are separately identified and aligned
  • Forecast accuracy is backtested against at least one holdout year
  • MAPE and bias are within acceptable ranges for the aggregation level
  • Operational recommendations include specific dates and inventory implications
  • Weather and external driver sensitivities are addressed for weather-sensitive categories

More skills from writer

Validate informed consent documentation for completeness, regulatory compliance, and clinical appropriateness by assessing required elements against CMS Conditions of Participation, Joint Commission standards, and state-specific consent laws. Use when auditing consent records, preparing for accre...

8 0
View

Conduct M&A financial due diligence analysis for bank acquisitions including accretion/dilution modeling, fair value marks, cost synergy validation, and pro-forma capital impact. Use when evaluating bank acquisition targets, preparing deal committee materials, analyzing merger economics, or asses...

8 0
View

Interpret and explain fraud detection signals, model outputs, and suspicious patterns across payment, identity, and account fraud typologies. Use when analyzing fraud alerts, explaining model scores, investigating fraud rings, or documenting fraud case findings for consumer and commercial banking.

8 0
View

Analyze policy exceptions for risk patterns, concentration risks, and control gaps in financial institutions. Use when reviewing policy exception requests, analyzing exception trends, assessing exception backlogs, or evaluating whether exceptions indicate systemic control weaknesses under COSO, O...

8 0
View

Explain portfolio drift from target allocations in client-friendly language with actionable rebalancing guidance. Use when preparing client review presentations, explaining allocation changes caused by market movements, justifying rebalancing trades, or helping clients understand why their portfo...

8 0
View

Decompose and explain healthcare cost growth into contributing factors including utilization changes, unit price shifts, case mix evolution, and population changes. Use when analyzing PMPM cost trends, explaining cost variance to leadership, preparing actuarial summaries, or identifying cost redu...

8 0
View

Analyze operational risk events and produce structured loss event narratives, root cause analyses, and OpRisk reports. Use when documenting operational loss events, conducting RCSA analysis, preparing OpRisk committee reports, or supporting Basel III operational risk capital calculations under th...

8 0
View

Predict provider capacity shortages and surpluses using demand modeling, wRVU productivity analysis, panel-size optimization, and workforce pipeline analytics to inform strategic recruitment and scheduling decisions.

8 0
View

Assess internal control testing effectiveness for financial institutions using COSO 2013 framework and SOX 302/404 requirements. Use when evaluating control test results, assessing control design and operating effectiveness, analyzing control deficiencies, or preparing management's assessment of ...

8 0
View

Diagnose supply delays by decomposing lead time into constituent stages, identifying variance sources, quantifying downstream impact on inventory and service levels, and recommending corrective actions.

8 0
View

Attribute AUM growth to its component drivers — market performance, net new assets, client acquisition, attrition, and fee impact. Use when analyzing practice growth, preparing management reporting, evaluating advisor productivity, forecasting AUM trajectory, benchmarking organic growth rates, or...

8 0
View

Generate timeline-based patient summaries that synthesize clinical events, diagnoses, treatments, and utilization across multi-year care histories. Use when preparing care transition summaries, complex case reviews, care coordination briefs, or longitudinal health records for clinical decision su...

8 0
View

Popular AI tools

Kaiber logo
Video

Kaiber

Generate, edit, and beat-sync AI video with leading models in one workspace.

Paid
View
Vimcal logo
Productivity

Vimcal

The world's fastest calendar for remote work

Free
View

Transform Your Design with AI Designer by ImgCreator.ai

Freemium
View
Akool AI logo
Content & writing

Akool AI

Revolutionizing Video Production with AI-Powered Creativity

Paid
View

Extend an image past the frame and let AI fill the new aspect ratio.

Freemium
View
StarByFace logo
Security

StarByFace

Discover your celebrity doppelgänger with StarByFace!

Free
View
C

ChainClarity explains 700+ crypto whitepapers in plain English, with layered summaries, comparisons, research tools, alerts, and a $4.99 Pro plan.

Freemium
View
Opus Clip logo
Coding & apps

Opus Clip

Opus.ai: Revolutionize Your Web Experience

Free
View