Manufacturers of fast moving consumer goods face a challenging task. Their products not only have to appeal to consumers, they also need to offer a range with a strong presence and clear position in their category. Retailers also need assurance that these products contribute to the overall success of the category if they are to achieve presence on the shelf.
Much can be learned from analysis of sales data, but this has a number of limitations. It can only give information on what has happened in the market, and not what could happen (e.g. the introduction of a new product or a major change in price). It is also difficult to pull apart the value of individual products to consumers, especially in categories that are heavily promoted or where distribution is very uneven. This in turn makes it challenging to identify the optimum portfolio and what each product brings to it.
To assist manufacturers and retailers, market researchers can use information from consumer surveys to describe the characteristics, usage patterns and attitudes of category shoppers. These descriptive methodologies enhance understanding of the detail behind sales data and provide the basis for modelling future consumer behaviour. However, they still lack sufficiently detailed information about shoppers’ purchase intentions when faced with new product offerings with a variety of prices and features.
This is where category simulation, based on conjoint analysis, comes into play. Conjoint is a powerful survey technique that has been around for decades; however, market researchers tend to use it to assist new product development and ignore its wider applications, especially in simulating purchase behaviour. Category conjoint uses the principles of experimental design and choice modelling to present consumers with Pointof-Sale type choices and uses this information to produce realistic models of consumer purchasing within a category.
Optimising category strategy
Some examples of questions that clients ask are: “What is the optimal range that will significantly increase category share and grow the category as a whole?”, “If we expand our product line, will overall revenue grow, or will we suffer too much cannibalisation?” or simply “How should we price our new product to maximize adoption?” Category simulation addresses all of these.
Consumers make buying a product look simple, but they are in fact processing information at an astonishing speed. Much of this process is intuitive and consumers are unable to explain the complex trade-offs they are making between price, brand, product features etc. ‘Category conjoint’ addresses this by seeking to replicate the point-of-purchase decision across all the elements that drive product choice – brand, variant, pack size, price, promotion. The challenge for researchers is to understand consumer sensitivity to these factors in a realistic but controlled context.
Taking a lead from behavioural economics, we further enhance the realism of the simulations by incorporating additional information from the survey. This includes price sensitivity measurement (so that we can pick up psychological price points), past switching behaviour and information on the context of their choices (eg setting the conjoint exercise in relation to a recent shopping occasion and reminding them of the needs / channels / etc that applied at that time).
Combining market data to enhance business decisions
The ‘share of preference’ outputs from conjoint reflect what consumers say they will do, given a clear set of choices, and this alone can be a good guide to how consumers will react to changes to the category. From these outputs, we can observe gains vs losses and potential cannibalisation of a new product (fig. 1) and the potential volume of sales and value at each price point of a product (fig. 2). However, in the real world, consumer choices are constrained by lack of information, habit, advertising, distribution and other factors. The most useful models therefore combine the strengths of conjoint (sensitive measures of relative consumer preferences) with those of actual market data (absolute changes in purchasing behaviour).
Combining suitable market data (e.g. POS), can widen the scope of conjoint analysis to simulate market share, allowing market researchers to gain greater insight into potential future changes to the category. Further benefits can be achieved if information on current distribution and awareness are also incorporated. More challenging is the calibration of price movements – where market information exists relating to the introduction of a new product or a change in price (together with changes to competitor products), the analyst can represent these changes in the simulator and observe the changes that actually occurred in the market.
Category simulation fills a useful gap where ‘usage and attitudes’ market research and sales data cannot capture the real dynamics of the category in relation to new products, prices or promotions that are not currently present, without the expense of other proprietary forecasting methods. The simulator tool that category conjoint produces allows manufacturers to optimise their offering; at the same time it enables them to do this in a way that also effectively serves the needs of retailers.
Recent Case Study: Fast-Moving Consumer Goods
A manufacturer of a variety of FMCG categories regularly uses proprietary research techniques that produce solid estimates of take up for individual new products or line extensions. However, these research approaches are less helpful when guidance is needed on how a whole portfolio of products should be managed. We worked with this client to develop a suitable category conjoint methodology for a number of their categories (detergents, air fresheners) and produced Excel-based simulators designed for range planning. They have subsequently played an important role in helping this manufacturer to develop suitable strategies and go to retailers with clear arguments for adopting those strategies.
The manufacturer had their own sophisticated internal category pricing models, so the role of category conjoint in this case was to provide information that would feed into these models. It has been sufficient therefore to report in our simulators consumers’ ‘shares of preference’ weighted by frequency of purchase, which are then entered into the internal models. On the basis of this information, they have made some bold moves in terms of introducing new products and charging premium prices and these have proved very successful.