Exploring Walk-Out Rates with Shopper Journey Data

Introduction

Shopper Journey Data provides opportunities to analyze shopping trip patterns and complex consumer purchase behavior. One key challenge for CPGs and Retailers is understanding why shoppers choose to purchase products at certain Retailers while bypassing other Retailers. Some reasons for a shopper leaving a store with a purchased basket of goods and shopping at another store may be beyond the control of CPGs and Retailers (e.g., idiosyncratic shocks such as forgetting a shopping list, crying baby, needing to return to work, and etc). However, many factors influencing walk-out rates such as product assortment, pricing, and promotional strategies are within their control. By understanding walk-out rates and the factors that influence them, CPGs and Retailers can reduce these subsequent trips and enhance shopper satisfaction. This article delves into constructing walk-out rates using detailed Shopper Journey Data, which relies on precise timestamps and basket-level information, and explores the insights gained by studying walk-out rates at the Retailer, category, and brand levels.

Brief Description of Shopper Journey Data

The Shopper Journey Data we analyze is a subset of Numerator’s extensive consumer panel data, which contains data for over 1 million shoppers, and 1 billion shopping trips [1]. These data contain detailed shopping trip information including household ID to identify who went on a shopping trip, the Retailer where the household purchased items, the entire basket of items purchased at the specific Retailer, location of shopping trip, and, precise timestamps. These details allow us to track a shopper’s trip circuit to identify how many stops were made, which Retailers were visited, and which products were purchased in a given time period [1]. For example, we can see if a household shopped at Retailer X in the morning and Retailer Z in the evening on the same day, or if they visited Retailer X and then Retailer Y within the same morning.

The detailed timestamps and basket-level purchases enable us to observe whether shoppers are making subsequent trips to purchase specific products. However, a challenge we face is not knowing why a shopper didn’t purchase a particular product at Retailer X but did so at Retailer Y, which could be due to various reasons including product assortment or price/promotion.

For this study, we analyze a small slice of Numerator’s data that captures all the shopping trips for households located in the Ohio zip code area starting 453*. These data contain over 666,030 items, 16,000 shoppers, 1.7 million shopping trips, and 15.3 million purchases across 2,205 Retailers between 2022 and 2024 (the top 10 retailers capture more than 90% of the shopping trips).

Constructing walk-out rates

As previously mentioned, data on shoppers’ trip circuit helps us understand how many trips were made, Retailers visited, and products purchased by a shopper within a given time period. To construct walk-out rates, we use these data and exploit the timestamps of shopping trips to characterize shopping trips as either satisfactory or unsatisfactory. The main assumption is that an unsatisfactory trip occurs when a shopper makes an additional trip on the same day, indicating they did not obtain all their necessary items on their first trip. For each household, on any given day we can observe:

• Number of shopping trips
• Precise timing of shopping trips
• Detailed basket of products purchased for each shopping trip

With these data, we can identify the following scenarios and construct walk-out rates for a specific Retailer “k” and brand “i”:

• Satisfied shopping trip at Retailer “k”: Household purchased products at Retailer k and did not have subsequent shopping trips on the same day.
• Unsatisfied shopping trip (i.e., walk out) at Retailer “k”: Household purchased brand “i” at a different Retailer “j” given previous trip was at Retailer “k” earlier the same day.
  • Potential reasons for not making a purchase at Retailer “k”: product unavailability, price, package size, Retailer experience, unexpected circumstances, and etc.

The walk-out rate for Retailer “k” for brand “i” over a given time period is defined as:

$$\text{Walk-Out\ Rate} = \frac{\text{Number of Unsatisfied Trips}}{\text{Number of Satisfied Trips} + \text{Number of Unsatisfied Trips}}$$

Insights

For the specific subsample of data we analyze, there are over 1.7 million shopping trips and 390,000 instances where households went on multiple shopping trips the same day. Table 1 below presents aggregate statistics on walk-out rates across several Retailers (anonymized for data privacy). These walk-out rates represent the percentage of trips where a household initially purchased products at a specific Retailer and then purchased additional products at another Retailer on the same day. The average walk-out rate across all Retailers is 25.8% suggesting approximately a quarter of initial shopping trips result in a subsequent shopping trip the same day. However, these walk-out rates range between 24.0% to 28.8% depending on the Retailer. The variation in walk-out rates could be due to product assortment, availability, price, shopper experience, or household demographics. For example, it’s not too surprising that Dollar A has higher walk-out rates considering their stores tend to be smaller and their product assortment is more limited relative to other Retailers. Likewise, it makes sense that Grocery A would have lower walk-out rates considering they are larger stores with a broader product assortment and multiple pack size variations.

Table 2 below illustrates walk-out rates by product category. For example, column two shows the percentage of trips where shoppers initially purchased products at a specific Retailer, didn’t purchase in the candy category and then went a subsequent shopping trip and purchased candy later the same day. These results suggest households may not have been satisfied with the product assortment or prices of products for the candy category at the first Retailer.

Walk-out rates by category are inherently lower than walk-out rates by Retailer since we are focusing on subsequent trips resulting in purchases within the candy category (i.e., shopper had to purchase candy on their subsequent trip rather than any purchase). Walk-out rates do vary by category, which could be a reflection of how frequently shoppers purchase items for a category. Candy, Drinks, and Bread category purchases represent 3.0%, 4.7%, and 1.7% of all purchases in our dataset, respectively. However, lower variation in walk-out rates across Retailers for a specific category may suggest little variation in product assortment or price for that category. For example, if there is little variation in product assortment, package size, or price for the bread category across all Retailers then there is less of an incentive to make a second shopping trip to purchase packaged bread at a different Retailer. Lastly, there is variation in walk-out rates across Retailers within a category which suggests some Retailers are better at capturing the appropriate product assortment or providing more appealing pricing and promotions. Columns 3 and 4 provide results for the Drinks and Bread categories, respectively. The Drinks category appears to have more variation in walk-out rates across Retailer which could reflect their product assortment or prices.

Table 3 presents walk-out rates for specific brands. The brands shown reflect frequently purchases products in their respective categories mentioned in Table 2. Similar to the category walk-out rates constructed in Table 2, brand walk-out rates are inherently lower than walk-out rates by category and Retailer since we are focusing on subsequent trips resulting in purchases of a specific brand (i.e., a shopper had to buy Brand 1 on their subsequent trip rather than any product or category).

Walk-out rates do vary across brands, which may reflect how frequently shoppers purchase products within a category. Brand 1, 2, and 3 purchases represent 0.3%, 1.0%, and 0.2% of all purchases in the dataset, respectively. Additionally, these findings may also represent how well the brand is being placed in its product assortment or priced and promoted. Tables 1 through 3 provide valuable insights at the Retailer, category, and brand level, as variation across Retailers clarify how well Retailers and CPGs capture purchases through their product assortment and pricing strategies.

Conclusion

In this article we analyze shoppers’ trip patterns to construct walk-out rates and understand how shopper’s are choosing to make multiple shopping trips. At the Retailer level, the variation in walk-out rates suggests some retailers are more effective at capturing products and package sizes that households are more likely to purchase. Although certain Retailers, such as convenience and dollar stores, inherently have smaller store footprints and ultimately fewer products and package sizes, they can still optimize their product assortment to decrease their walk-out rates. We also observed that certain categories have higher variation in walk-out rates across retailers, which could be due to differences in product assortment and pricing and promotion strategies. At the brand-level, these same factors could also influence the variation in walk-out rates across retailers.

There are many possible explanations for these differences in walk-out rates. Future analysis could explore various different mechanisms such as directly comparing product assortments and prices/promotions between retailers. In addition, consumer-level analysis could provide insight on which households shop at certain stores for particular items or which households are inherently multiple-trip shoppers or one-store shoppers. In future work, we hope to incorporate consumer demographics and Retailer characteristics to better understand what influences walk-out rates and the potential actions Retailers and CPGs could take to lower their walk-out rates.

If you’re interested in learning more about how walk-out rates are created and the insights we can uncover, please feel free to reach out to us at info@tickr.com. We hope you found this article helpful and look forward to hearing from you!

Citations

[1] Numerator (2024). Numerator OmniPanel Data. Numerator https://www.numerator.com/omnipanels/