Forecasting Methods: Qualitative and Quantitative
A store manager looks at last year's Diwali sales, this week's footfall, a competitor's discount banner across the road, and a sudden rain forecast. The spreadsheet says one thing; experience says another. Good forecasting is not choosing between judgement and data - it is knowing when each deserves the driver's seat.
- Forecasting estimates future demand before actual demand is known, so supply, capacity, inventory and cash can be planned.
- Qualitative methods use expert judgement, market intelligence and structured opinions when historical data is weak or the future is discontinuous.
- Quantitative methods use past data, patterns and statistical or machine-learning models when enough reliable data exists.
- The method choice depends on four variables: data availability, forecast horizon, demand stability and cost of error.
- Use qualitative for new products, policy shocks, one-off events and early market entry; use quantitative for stable, repeatable demand.
- The best planning teams use a hybrid forecast: baseline from data, judgement for known future events, and accuracy tracking after actuals arrive.
- Interview-safe line: βI would not ask which method is best; I would ask which method is best for this data, horizon and decision.β
Big Picture: Forecasting Converts Uncertainty into an Operating Plan
Forecasting sits between market reality and business action. A demand planner is not trying to predict the future perfectly; they are trying to create a usable estimate that downstream teams can act on. That is why forecasting is the starting point for inventory, procurement, production, staffing and logistics decisions - the same logic you see in why demand planning decides everything downstream.
Core Explanation: Qualitative and Quantitative Forecasting
The simplest way to understand forecasting methods is this: qualitative methods ask informed people, while quantitative methods ask reliable data. A strong manager knows when each one is appropriate.
1. Qualitative Forecasting Methods
Qualitative forecasting relies on judgement, expert opinion, customer insight and market intelligence instead of mainly using historical numerical patterns.
Use it when:
The main qualitative methods are:
2. Quantitative Forecasting Methods
Quantitative forecasting uses historical data and mathematical models to estimate future demand based on patterns, relationships or probabilities.
Use it when:
The main quantitative methods are:
3. The Hybrid Forecast: Where Good Planners Actually Operate
In real companies, the winning answer is rarely βonly qualitativeβ or βonly quantitative.β The baseline forecast may come from a model, but planners adjust it for known events that the model cannot see yet - a festival promotion, competitor stockout, price change, local weather disruption or new distributor onboarding.
This is where demand sensing through signals and point-of-sale data becomes useful: it gives the planner fresher inputs than last monthβs shipment history.
4. A Small Worked Example: Moving Average vs Judgement Adjustment
Suppose a cafΓ© chain is forecasting May demand for a cold coffee SKU in one city.
Three-month moving average forecast for May = (1,100 + 1,300 + 1,200) / 3 = 1,200 cups.
Now suppose the manager knows a college festival is scheduled near two outlets in May. A purely quantitative forecast would remain 1,200. A hybrid forecast may add a judgement adjustment, say 10%, making the forecast 1,320 cups. The important interview point is not the exact adjustment; it is that the adjustment must be documented and later tested against actual demand.
If actual May demand is 1,250 cups:
- Moving average error = 1,200 - 1,250 = -50 cups, meaning under-forecast.
- Hybrid forecast error = 1,320 - 1,250 = +70 cups, meaning over-forecast.
The learning is sharp: judgement can improve a forecast, but only if the assumption is tracked. Otherwise it becomes invisible bias.
5. Metrics to Track Forecast Quality
Never discuss forecasting without saying how you will measure it. Forecast quality is not judged by confidence; it is judged by error, bias and decision impact. For a deeper revision of these measures, go next to measuring forecast accuracy and bias.
Definitions You Can Say in One Breath
- Forecasting: Estimating future demand using data, judgement and assumptions before actual demand is known.
- Qualitative forecasting: Forecasting based mainly on expert judgement, market intelligence and structured opinion when hard data is limited.
- Quantitative forecasting: Forecasting based mainly on historical data, statistical patterns and measurable demand drivers.
- Hybrid forecast: A model-based forecast adjusted with documented business knowledge about known future events.
- Forecast bias: A consistent tendency to over-forecast or under-forecast demand over time.
Ferns N Petals: Forecasting a Festival-Spike Business
Ferns N Petals is a useful Indian case because gifting demand visibly spikes around occasions, while many products are perishable and delivery-sensitive.

Consider the business problem for Ferns N Petals during occasions such as Valentineβs Day, Motherβs Day, Raksha Bandhan and Diwali. Demand is not evenly spread through the year. It comes in sharp spikes, varies by city, depends on delivery slots, and includes perishable categories where over-forecasting creates waste while under-forecasting creates lost sales.
A purely quantitative forecast would look at past orders by city, category, channel and occasion. That gives the baseline: which SKUs sold, when orders came in, what delivery windows filled fastest, and which locations saw peak demand. But the model alone may miss this yearβs changes: a new campaign, a pricing bundle, a competitor offer, local weather, courier constraints or a new fulfilment location.
The stronger move is hybrid forecasting. The planning team would start with historical demand by occasion, then layer qualitative inputs from marketing, category managers, local operations and vendor teams. The primary driver is the occasion-linked historical demand pattern; supporting drivers include campaign intensity, city-level capacity, vendor availability, product perishability and delivery-slot planning.
The lesson: for a festival-spike business, forecasting is not just a data science exercise. It is a cross-functional planning discipline where the forecast must be accurate enough to protect sales, service levels and perishability economics at the same time.
How AI Changes Forecasting Methods
AI does not remove the need for forecasting judgement. It changes what data can be used, how fast forecasts update, and how planners test assumptions.
- AI expands the signal set. Traditional models rely heavily on past sales. AI-enabled demand sensing can include point-of-sale trends, search interest, weather, local events, price changes, campaign data and competitor signals. This is especially useful when demand changes faster than monthly planning cycles.
- AI improves granular forecasts. Machine-learning models can forecast at SKU-store-day or SKU-pincode-day level when the company has enough data. That helps categories with local variation, short lead times and many demand drivers. For a deeper next step, revise machine learning forecasting where AI beats statistics.
- AI makes scenario generation faster. Planners can ask: βWhat if the promotion uplift is lower, supply is constrained, or demand shifts to a different city?β AI tools can help generate scenarios, but humans must still validate business logic and data quality.
Use NotebookLM: upload this lesson, a company annual report, and any public investor presentation; ask it to generate βfive likely forecasting interview questions for this company, with signals, method choice, risks and accuracy metrics.β Then verify every company-specific claim before using it.
Interview Relevance
βA new FMCG brand is launching a premium juice in three cities. It has no historical sales data. Which forecasting method would you use, and how would you improve it after launch?β
In interviews, always connect the forecast to the business decision. A demand forecast is not useful because it is mathematically elegant; it is useful because it improves inventory, service, capacity and cash decisions.
Common Mistake
The most common error is saying βquantitative is better than qualitativeβ as if forecasting methods sit in a hierarchy. That costs candidates because it ignores new products, market shocks and judgement-rich contexts. One-line fix: say, βI will choose the method based on data availability, forecast horizon, demand volatility and cost of error.β