[Shocking] "Accurate Sales Forecasting" is not a dream! Practical techniques that change everything dramatically by simply classifying products and customers into "just 3 categories"

"My forecast was off again..."
"The field data and the numbers don't match at all!"

If you are struggling with such sales forecasting issues, this article is for you.

Modern business is an era of intense change, like a roller coaster, where the future is hard to see. In such a highly uncertain situation, if the "sales forecast," which serves as a company's compass, is shaky, management strategy cannot function either.

But don't worry!

Actually, there is a special secret to dramatically improving the accuracy of sales forecasting. It is a simple yet super-powerful method of classifying products and customers into "just 3 categories" each and applying a forecasting approach tailored to their respective characteristics.

"Wait, just by doing that?" you might think, but this simple shift in perspective will elevate your company's sales forecasting to a level that is more practical and convincing, making you say, "I see!" Now, let's jump into the world of "accurate sales forecasting" together!

1. Why is your sales forecast "inaccurate"? ~The pitfall of "the same forecast for everyone"~

In many companies, I see scenes like this.

"Whether it's a new product, a long-seller, or unsold stock, let's try to forecast everything using the same forecasting tool and the same method!"

...This is actually the biggest reason why sales forecasts are off.

Think about it. It's like applying the same growth forecast model to a newborn baby (a new product) and a veteran with a wealth of life experience (an existing stable product). It's only natural that it's unreasonable, right?

Product life cycles vary widely. Relationships with customers are not uniform either. For long-loved staple products and new products that have just arrived with great fanfare, the information required for forecasting and the uncertainty involved are completely different.

Nevertheless, applying a "uniform forecasting approach" to targets with such diverse characteristics is not only inefficient, but it is also like lowering the forecast accuracy yourself.

Deeply understanding the "individuality" of the target, "what are we forecasting?", and choosing the optimal approach accordingly. This is the first step toward improving forecast accuracy and the most important "realization."

2. Classifying products into "3 faces"! ~The compass for forecasting strategy~

Let's divide your company's products into the following 3 categories based on their sales performance and continuity. This classification will be your compass that clarifies each forecasting strategy.

2-1. Category 1: The backbone of the company! "Existing stable products" ~The future told by data~

"Oh, this one! It's our staple product."

Products that you can say that about fall into this category. It is characterized by abundant past performance data and relatively stable sales. Sales patterns such as seasonal fluctuations and trends can also be grasped clearly, like tree rings. Truly, the "honor student" of forecasting!

[Forecasting Approach] For this category, "quantitative forecasting methods" that make full use of past sales data are highly effective.

• Time series forecasting models: Using methods like ARIMA models or exponential smoothing, we predict future trends and seasonality from past sales performance. It's as if a skilled fortune teller is predicting the future from the past movements of the stars!

• ABC analysis: Rank products according to their sales contribution. Apply more detailed forecasts to "A-rank products" that support the company's sales, and apply efficient forecasting to other products; let's be selective.

• Machine learning models: It is also possible to build highly advanced forecasting models, like an AI secretary, that take into account complex factors such as promotions, competitor movements, and weather.

[Expected accuracy] By applying these methods, you can expect surprisingly high forecasting accuracy. Since it is based on quantitative data, the objectivity of the forecast results is also perfect!

[Operational points] The appeal of this category is that once a model is built, it is easy to automate and streamline. After building the forecasting model, focusing on detecting outliers and analyzing the causes when discrepancies between forecasts and actual results occur will lead to further accuracy improvements.

2-2. Category 2: Lessons from the past! "Discontinued/Underperforming Products" ~Failure is the mother of success~

"We sold it last year, but it's not here this year..."

This category refers to products that have been discontinued or whose sales have dropped significantly and are excluded from the sales forecast for the current term. At first glance, people tend to think it has "nothing to do with forecasting," but in reality, a "treasure map" to the future is hidden here.

[Forecasting approach] We do not forecast future sales, but the main purpose is to thoroughly "dissect" past sales performance and the circumstances that led to discontinuation.

• Analysis of the causes of sales decline: "Why did it stop selling?" The emergence of competitors? Market changes? Quality issues? Or a failure in promotion? Like a master detective, let's analyze the factors from multiple angles.

• Minimizing inventory disposal/waste loss: From past sales data, we consider the optimal disposal method for remaining inventory and strategies to minimize waste loss.

[Expected effects] This analysis is not just a look back at the past. It serves as valuable feedback for next-term product development and acts like a "vaccination" to prevent future losses.

[Operational points] The perspective of "learning from failure and applying it to future strategies" is more important than anything else. Bitter experiences are also valuable data for future success. Never let them go to waste.

2-3. Category 3: A challenge to the future! "New Products" ~A story from scratch~

"A new product we've put our heart and soul into, launching with great anticipation this year!"

This category is a challenging area where there is no or very little past sales data, making it the most uncertain in sales forecasting and "where the forecaster's skills are put to the test." Using the same methods as for existing products will likely make an accurate forecast impossible.

[Forecasting approach] In forecasting new products, the focus is not just on relying on quantitative data, but on gathering information from multiple angles and using a "qualitative approach." It's like an artist creating the future, imagining painting a picture by combining various pieces of information.

• Focus on qualitative forecasting: Interviews with sales staff and members of the product planning department are truly a "lifeline." Raw voices such as "which customers, how much, and how we want to sell" are important inputs for forecasting.

• Utilizing Supplementary Information:

  • Past Sales Performance of Similar Products: Data on "similar products" released in the past is extremely useful as a "benchmark" for initial forecasts of new products. Select products with similar price ranges, target demographics, and features.

  • Market Trends and Competitor Analysis: Thoroughly analyze the trends in the market where the new product will be introduced and the actions of competitors to estimate the potential market size and share.

  • Sales Plans and Marketing Strategies: The "goals" set by the sales department and the "promotional activities" planned by the marketing department serve as the "prerequisites" for the forecast.

  • Results of Preliminary Test Sales and Questionnaire Surveys: The results of small-scale test sales and customer surveys are valuable "leading indicators" for gauging initial demand.

  • Hypothesis-Based Consideration: An effective approach is to form a "hypothesis" such as "If we set this price and run this promotion, we can expect this much sales," and then revise the forecast while verifying it.

[Expected Accuracy] Although forecast accuracy tends to be low in the initial stages, by continuously collecting and verifying information as described above, you can gradually improve accuracy. Just like a newborn baby growing up little by little.

[Operational Points] "Close coordination" with related departments such as sales, product planning, and marketing is essential. The key to success is to build a "flexible system" that allows for multifaceted information gathering, tolerates fluctuations in forecasts, and enables early course correction.

3. Capture Customers with "3 Faces" Too! ~Forecasts Change Depending on Who You Sell To~

Similar to product classification, you can further increase the resolution of your sales forecasts by classifying customers into "3 categories" based on their transaction history.

• Continuing Customers: These are your "loyal customers" who have had stable transactions for many years. Based on past purchase history and transaction volume, quantitative forecasting is possible with relatively high accuracy.

• Customers who had transactions last year but not this year: "Huh? That company hasn't placed an order recently..." These are customers who have stopped doing business for some reason. It is useful for analyzing the reasons for the cessation of transactions and for evaluating the potential for re-approach or the impact of lost orders.

• New customers who started trading this year: Whether they are new products or existing products, orders from new customers are an area where forecasting is difficult because there is no past data. In particular, quantity forecasting for new products from new customers has the highest uncertainty and requires careful interviews and hypothesis testing.

4. [The Ultimate Combination] "Explosively increase" forecast accuracy with cross-analysis of "Products x Customers"!

By combining product and customer classifications, the "conquest zone" for sales forecasting becomes clear. This will make it as clear as a map where you should focus your efforts and what kind of forecasting methods you should use.

• "Existing Products x Continuing Customers" zone: Truly the "Golden Zone"! This zone is the area with the highest forecast accuracy where stable quantitative forecasting is possible. Because there is abundant past data and clear patterns, you should actively promote automation and efficiency through data analysis. Focus on fine-tuning to further increase the accuracy of the forecast model and early detection of outliers. Example: Delivery forecast for standard seasonings to a supermarket with which you have been doing business for many years. You can see the sales trends by day of the week and season from past data as if you were looking at the palm of your hand.

• "New Products x New Customers" zone: The "Challenge Zone" that opens up the future! This zone is the most uncertain and challenging area. Because there is almost no past data, qualitative forecasting, hypothesis testing, and the "intuition and experience" of the sales floor are essential. It is important to allow for a certain margin of error in the forecast, provide rapid feedback on actual results after sales begin, and build a management system based on early course correction. Example: Proposing a newly developed AI-equipped gadget to a startup company you have never traded with before. Adjust the forecast while closely watching the enthusiasm of the sales representative and the market's reaction.

Besides this, there are zones such as "Existing Products x New Customers" and "New Products x Continuing Customers," and forecasting approaches tailored to each can be considered. The important thing is to build an optimal forecasting strategy for all combinations according to their characteristics. It is like moving shogi pieces one by one, building your forecast strategically.

5. Special operational points to "explosively increase" forecast accuracy!

There are several "secrets" to incorporating this "3-category approach" into your business and improving forecast accuracy.

• The "ultimate tag team" of data analysis and on-site perspective: Quantitative data analysis is very powerful, but the background of "why?" that cannot be seen from that alone, the raw voices of customers, and the atmosphere of the market cannot be captured without on-site knowledge. By fusing the results of data analysis with the "living perspective" of the field, such as sales and product planning, more realistic and convincing forecasts that make you say "I see!" are born.

• The "art of using" flexible forecasting models: It is important not to apply the same forecasting model to all categories, but to flexibly use the optimal method according to each characteristic. Just as a chef changes cooking methods according to the ingredients, choose the best tools, such as advanced quantitative models for existing stable products and qualitative interviews and hypothesis testing for new products.

• "High-speed rotation" of the PDCA cycle: Forecasting is not over once it is done. It is essential to establish a "PDCA cycle" that regularly analyzes the divergence between forecasts and actual results, identifies the causes, and continuously improves forecasting models and information gathering processes. Forecasts are living things. Keep brushing them up constantly.

• The "magic word" is communication!: Sales forecasting is a "team battle" involving many departments such as sales, product planning, production, and purchasing. Close cooperation and information sharing with each department not only increases the accuracy of the forecast but also fosters a sense of conviction in each department regarding the forecast, leading to smooth business execution.

Summary: Your sales forecast will now be "accurate"!

Improving the accuracy of sales forecasting was an unavoidable challenge in corporate management strategy, but one that many companies struggled with.

However, as introduced in this article, classifying products and customers into three categories: "continuing, disappearing, and new," and taking an approach tailored to each characteristic will be an extremely effective "game changer."

Streamline existing stable areas with quantitative data analysis, and make maximum use of qualitative information and on-site knowledge for new or highly uncertain areas. This "balanced approach" will not only chase numbers but will lead to building a convincing sales forecasting model that is in line with the realities of business, and will dramatically increase the quality of your company's management decisions.

Now, try incorporating this "3-category approach" into your company's sales forecasting starting today. By combining data analysis and on-site perspectives, you should be able to increase forecast accuracy, realize more practical and highly reliable sales forecasting, and evolve into the "ultimate business person" who predicts the future!