Demand forecasting for new products? Many heads are better than one!
August 25, 2015
How does social forecasting work? Senior Business Consultant Ieke le Blanc and Managing Partner and Academic Director Freek Aertsen explain how in this column.
Forecasting the demand for new products is often a challenge. This was, for example, illustrated by Samsung in early July. They see the misjudged demand for their high-end smartphones as an important cause of the decline in operating profit. Samsung launched the Galaxy S6 and its curved screen variant, the Galaxy S6 Edge, in April 2015. The expected distribution of sales was 80% for the straight variant and 20% for the curved Galaxy S6 devices. The production was set to this ratio. In reality, however, the demand was 50% for straight phones and 50% for the curved variant. That created a surplus of the normal S6 and a shortage of the curved S6 Edge with a 100-euro higher sales price. Sales opportunities were thus missed (source: WSJ).
How can this be prevented?
Forecasting the demand for new products is difficult. The assessment of sales and/or product marketing is often taken as a starting point (judgmental forecast). This is often still compared to historical sales by using the predecessor-successor relation. This is often only partially successful. The main problems are:
- The estimates are often too optimistic (biased). Every product manager, after all, has a firm belief in his own products.
- Historical data is only useful to a certain extent. The extrapolation of the past does not work well in new/innovative products. In addition, predecessor-successor relations are ambiguous.
The crux is to use the capabilities and the data that are available. In a project for a competitive producer of smartphones, the demand for different colors/designs is forecasted through the use of various inputs. During the process of introducing a new product, opportunities and data sources become available, each with their own advantages and disadvantages:
(Extrapolating from the past)
- Data availability
- Data reliability (note: this is not the same as forecast reliability)
- Limited by the past: "Past results are no guarantee for the future"
- Predecessor-successor relations are often ambiguous
(Estimates by marketing and sales professionals)
- Easy to set up
- Market and product knowledge is used
- Often biased
- Based on a small group
3.Company internal survey
(Staff panel/survey: every employee is also consumer, ask for preference)
- Internal and therefore easy to set up
- Based on data
- Biased: employees believe in their own company
(Targeting potential buyers on social media and asking for their preference)
- Reliable data can be obtained by targeting potential buyers
- In dialogue with potential buyers
- Cooperation marketing/sales with supply chain planning
- Intention-driven (No real sales yet)
(Presale or subscription for consumers to get product delivered first)
- Real sales data of the new product
- Maybe already too late?
- Data from "early adopters"
Each option has its added value. Therefore, the predictions are combined with weightings on the basis of the reliability from the past. This method proved to provide reliable forecasts for real consumer demand for the new designs and colors (not limited by availability). The real preference of consumers based on the targeted marketing profile proved very valuable and is used for new product development.
Social forecasting: the power of a group
Many heads are better than one! In a "company internal survey" the expert panel is extended to all employees using a structured (online) survey. Social forecasting goes one step further by approaching potential buyers by using social media. Together with an agency specializing in online marketing, a Facebook application has been built to collect responses. Participants were asked to indicate their first and second preferred color. This information was combined with data from the Facebook profiles. To enlarge participation and reliability, a smartphone in the favorite color was raffled among all participants. Targeted Facebook ads were used to select the right group of potential buyers according to the marketing profile. The origin could be traced with the help of Google Analytics. Using advanced statistics (data mining) and a reliable data set consisting of potential buyers only was assembled in this way.
This approach has many similarities with the use of consumer panels. The main difference is the use of social media to appeal to targeted potential buyers. Due to the low threshold, the sample size is large and it provides sufficient data in order to predict the demand for the new product.
Implementation requires expertise
Responsive supply chains also have limits in respect to flexibility - that is why a good forecast is always important. Social forecasting is a useful technique for new products. The knife cuts both ways. Firstly, the new products can be announced on social media and aimed at potential buyers. Secondly, reliable data is collected at the same time about the real preference of the potential buyer.
The application can range from new colors and designs to flavors. A good implementation, however, is the work of specialists and requires good knowledge of data analysis techniques to prevent drawing wrong conclusions. But even then it remains a forecast and the market demand is always right: "De gustibus non est disputandum."
This article has been written by Senior Business Consultant Ieke le Blanc and Managing Partner Freek Aertsen from EyeOn. Aertsen is the Academic Director of the TIAS Master of Operations and Supply Chain Excellence. The blog was first published on Logistiek.nl.
Forecasting customer demand?
Forecasting is discussed in the TIAS Master of Operations and Supply Chain Excellence program.Would you like to prepare properly for the demand of your customers? Download the brochure.