INTRODUCTION
Forecasts employ a variety of techniques to anticipate future market changes using historical data as input1 for making informed decisions today. In most businesses, decisions have traditionally been based on subjective opinions. Though quantitative decision methods ages more than thousand years, they have only been applied in business circles from the twentieth century. But when employed, they improve the cost-cutting and revenue growth initiatives we often cry for. Let's see how.
Cost Savings at Taco Bells
A large fast-food chain specializing in Mexican food saved $16.4 million in labour costs for its 1996 fiscal year. Labour (30% of sales) was the major cost component managed through scheduling: Too few workers resulted in lost sales and poor service, as too many also resulted in excessive costs and reduced profits.2
Experiencing high variations in demand during the day, Taco Bells grouped sales in 15 minutes periods to notice 52% of sales occurring between 11AM and 2PM. Feeding such inputs into its six weeks weighted average forecast model, it generated sales estimates leading to employee scheduling arrangements by time bands.
TIME HORIZONS
At Taco Bells, the forecast horizon is the upcoming weekdays schedule for waitresses. But a six weeks data trend is in use to achieve an optimal forecast result. Business forecasts can be made over a decade, years, quarters, months, weeks or days. We provide examples of such forecasts under the below headings:
Strategic [Year/Years]
Tactical [Quarterly/Monthly]
Operational [Weeks/Days]
Effective strategic deliberations are not devoid of preliminary economic, demographic and competitor forecasts. Forecast assumptions are scrutinized and refined, as others play the devil's advocate to develop alternative scenarios. The closer the strategic forecasts are to reality, the more resources are effectively used. This is where forecasts determine the capacity costs (human, machinery and facility space) to engage for funding (investments) and profit considerations.
Tactical and operational forecast models help to manage the predetermined commitments at the strategic level. Relevance of forecast models to operational needs is critical. Business operations can evolve leaving its forecast models far behind the present. As one inspects the model, it is apparent that new operational developments have not been formally incorporated.
Besides, underlying assumptions within the forecast model can steer results away from accuracy. As such, there is need to periodically evaluate forecast accuracy3. The evaluation approach depends on the forecast detail sought: deterministic or probabilistic, discrete or continuous4. The greater the accuracy required, the higher the cost of implementing the forecasting model. However, if the costs arising from forecasting errors are manageable, we can consider the model as optimal.
EARLY DETECTION SIGNALS
With the right models, forecasts can highlight specific risks events analysts must watch. Exposures to interest-rates, exchange rates, material prices and within labour markets when evaluated can highlight issues that need addressing. By evaluating financial results through the lenses of the Altman Z-score, companies can anticipate early bankruptcy.
Demographic forecasts juxtaposed with customer preferences can help predict revenue levels devoid of marketing campaigns. If revenue needs are higher, managers will seek campaign resource maximization to reach specific segments with product, price and place adjustments. Co-ordination with operations is required to curtail queues and baulking likely to reduce the chance of reaching these targets.
Within operations, raw materials, interest and exchange rates price forecasts are key to avoiding market shocks. If operations can produce at low cost, using pre-arranged future contracts to protect sales margins, campaign programs will go on unperturbed. Investments in business models do yield significant paybacks above their design and training costs.
ACCURACY TECHNIQUES
Understanding these clear-cut advantages, we explore the barriers for greater accuracy. It starts with qualitative is highly reliant on expert opinions and are most beneficial in the short-term. Examples include: on-site visits, interviews and polls. The quantitative employs time-series or an econometric inference. Machine learning has become.
References
1 Forecasting: What It Is, How it's Used In Business and Investing, September 2022. Article by Tuovila Alicia
2 Introduction to Management Science (10th Edition), pp 674, 2006. Author Bernard W. Taylor
3 How Do I Measure Forecast Accuracy? August 2020, Article by Forecast Pro
4 Evaluating Forecast Quality, pp 3, International Research Institute For Climate Predication. Article by Simon J. Mason
