We consider a monopolist grocer selling a single perishable product with a fixed shelf life. The product can be replenished daily, based on the demand forecast and a safety factor. The grocer wants to evaluate which pricing strategy performs best in terms of total revenue, waste and stock-outs. In terms of the dynamic pricing problem dimensions provided by Elmagraby & Keskinocak [20], this research problem is an R-I-M dynamic pricing problem. Important to note is that in our case the expiry dates of the replenished products differ from the products that were already in the store. This is different from for example fashion, where products are also perishable, but their expiry date is the end of the season. When replenishments are ordered during the fashion season, their expiry date is also the end of the season. Since the grocer sells a single product only, substitution or complementariness effects with other products are not (explicitly) taken into account. Examples of other effects that influence customer behaviour which were also not taken into account include the reference price effect, where customers would adjust there willingness to pay based
CHAPTER 7. RESEARCH METHOD 63
on the prices they recently observed, and the presentation effect, where customer demand would be influenced by the amount of remaining inventory.
7.2.1
Pricing Strategies to Evaluate
Four main pricing strategies for perishable products will be evaluated:
PS1. Fixed price, no price change at all as items deteriorate.
PS2. Single fixed price change, with D% fixed discount on last day before expiration.
PS3. Multiple fixed price changes, withD% fixed discount spread linearly over the lastS days before expiration.
PS4. Single dynamic price change, with a dynamically determined discount on the last day before expiration.
The selected pricing strategies represent a balanced mix of strategies that are already used by grocers in practice and promising new strategies that were identified in the literature review. For the new strategies, an additional inclusion criterion was that they had to be simple enough to be applied in practice.
PS1 and variations of PS2 are currently being applied by Dutch grocers in prac- tice. Jumbo applies PS2 with a 100% discount, meaning that they give products away for free on the last day before expiration. Albert Heijn also applies PS2, but with a 35% discount on the day of expiration. Some supermarkets don’t change their price at all and hence apply PS1. PS3 and PS4 were included based on the literature review from section 6.3. PS3 was included because the results from sim- ilar simulations by Chung and Li [12] indicated that gradually applying the fixed discount, so spreading it over the last few days before expiry, could improve per- formance. PS4 was included because dynamically determining the discount based on the inventory levels and expected demand in each period is more flexible than applying a fixed discount percentage in all periods and intuitively this extra flexi- bility might result in improved performance. PS4 determines the optimal discount for the products that have one day left until expiry by estimating the revenue for that set of products for all discounts in a set of allowed discounts. It relies upon demand forecasts for next period to estimate revenue and chooses the discount that provides the highest expected revenue.
7.2.2
Performance Measures
The performance of these pricing strategies will be evaluated based on several key performance indicators (KPIs). These KPIs include the total revenue, the percent- age of inventory wasted and the percentage of customers that wanted to buy a product but faced stock-outs. PS2 to PS4 will be benchmarked against the default pricing strategy, which is PS1. For the comparison, the percentages change in each of the KPIs will be calculated relative to PS1.
Stock-outs (%)= # Stock-outs
# Sales + # Stock-outs∗100 Waste (%)= # Products Wasted
CHAPTER 7. RESEARCH METHOD 64 imposes constraints on Forecaster Pricer Customers Product Type Grocer forecasts demand for
Products Shelf Space
of type
stored on
select product from determines
prices for
Figure 7.1: Simplified overview of simulation model components
Before systematically evaluating the different strategies, some hypotheses can already be drafted when comparing PS2-4 to PS1. It is expected that PS2 with very high discount percentages result lowest waste by far, since customers like (almost) free products. However, one could argue about how valuable this waste reduction is, since customers might then pick products solely because they were (almost) free without really wanting or needing them, thereby increasing the risk of waste at the customers’ homes. A clear downside of PS2 with high discounts is that even though the products are ‘sold’, the impact on the profit is the same as when they would have been wasted because they generate (almost) zero revenue. That is why PS2 with lower discounts seems more attractive, since then more revenue will be generated for the products that are sold and customers are still attracted with a discount. It is expected that spreading out the discount such as with PS3 will result in higher revenues, as this was suggested in previous research [12]. PS4 is expected to result in the highest total revenues, since it has the ability to give discounts only when necessary (e.g. when excesses are very large),
At first a truly dynamic pricing strategy with multiple price changes over the products’ lifetime (with a maximum of 1 price change per day) was discussed as well, however since that is difficult to implement at brick-and-mortar grocery stores it was left out of the research scope for now. With a truly dynamic price strategy, it no longer seems feasible to let an employee change the product price sign or apply new discount stickers on a daily basis.