Insights

Designing future retail, modernizing Item Master

Sreejith Periyadath, UST Data Services

In the article, “It’s Time To Retire Item Master: Here’s What Should Replace It”, Peter Charness proposes UST’s approach to modernizing Item Master into a Product Information Resource Repository (PIRR). The first step in the 5-part journey is designing a complete PIRR that captures the various attributes that make the item data rich enough to support the complex internal systems, growing consumer demands, and futuristic analytics requirements. This article proposes a practical approach in designing a full-fledged PIRR (Item Master).

Item Master, as Peter puts it, is the nerve center for thousands of applications. One of the aspirational objectives for retailers is to systematically integrate the retail functions such that the same rich information is continuously shared across all teams and all systems to enable automated decisions. This requires deep thinking about the correlation between various retail functions and how Item Master data attributes impact and enhance the business functions, both internally (IT systems) and externally (consumers). Leveraging Item Master data across the retail chain provides an immense possibility to massively accelerate fundamental new capabilities (in particular AI/ML) across various business functions (Figure 1).

Figure 1: Leveraging Item Master data across the retail value chain

However, retailers find it hard to obtain these benefits because of the lack of structured thinking in identifying the possible use cases so that they can invest appropriately for enablement at the system and technology level. Most of the time, retailers invest significantly in technology to enhance the capability to hold and enrich master data (second step as outlined in Peter’s blog), with a mindset of fixing past issues without figuring out how to leverage the enriched data. It is necessary to reverse the process to start with the vision of how exactly to leverage master data while venturing into the technology implementation. Retailers make considerable investments in effectively collecting and exposing the right set of item attributes and values and leveraging that information to create business value.

We will explore a proven methodology to identify relevant use cases that can lead to defining a complete PIRR. If done right, the possibilities are endless at making the retail value chain smarter (Figure 2).

Figure 2: Possibilities of leveraging item attributes across retail functions

If so, what is the challenge?

Identifying the attributes and values to be captured for every product line and the specific item is a tough challenge. Retailers struggle to comprehend the breadth and depth of item information, that is, all the possible item attributes and various meaningful values that each of the attributes can hold, while at the same time focusing on the relevant and important business drivers. An even bigger struggle is identifying, capturing, and classifying the appropriate or necessary attributes and values out of the plethora of possibilities. Organizations usually take the approach of digging deeper into each of their business functions and related systems to identify how to integrate item information into their processes and decision-making (Figure 3). This is a limited approach and a daunting task. This approach also falls short in reaping the full benefit of the possibilities offered by richer item information.

Figure 3: Mapping item information across business functions

A structured and innovative approach

What we need is a structured and innovative way to identify the relevant item attributes and values and integrate them into retail functions to derive value.

STEP 1: IDENTIFY A COMPREHENSIVE LIST OF POSSIBILITIES

Progress starts by identifying a comprehensive list of possibilities – referred to here as “use cases” – of how item information can be utilized. This is an imaginative and creative brainstorming process to be held amongst business thinkers and leaders. It is irrelevant at this point as to how much of the product information can be traced down or whether it is available at all, and how it can be integrated into the business processes or information systems. The only focus at this point is to gather as many use cases as possible.

How do we do this? One way is by correlating how item information can influence the sales drivers (Figure 4). Sales driver refers to the factors that can drive up sales. Sales drivers can be broadly classified into;

1. Customer or personal profile

2. People or social demographics

3. Environment or weather or external factors

4. Events of various kinds

Under each sales driver, there can be many sub-categories or sub-drivers.

STEP 2: IDENTIFY THE ATTRIBUTES FOR EACH PRODUCT CATEGORY

Item attribute refers to general product characteristics. Item attributes can be broadly classified into color, style, texture, dimension, material, utility, variety, ingredients, shelf life, etc., in addition to the brand and price. Not all are applicable to every product. Also, the values for the same attribute can be different for different products or indeed vary for the same product in different locations. One store’s “basics” can be another store’s fashion.

STEP 3: MAP THE USE CASES

The idea is to be creative and figure out use cases connecting attributes and values to sales drivers (Figure 4). Visually mapping the attributes to sales drivers will trigger various possibilities (use cases) that are otherwise hard to imagine.

Note that all the various values for the attributes need not be known at the time of deriving use cases. However, having all the attributes identified will increase the likelihood of uncovering all possible use cases.

Figure 4: Use cases connect item attributes and values to sales drivers

STEP 4: REFINE AND EXPAND THE USE CASES

Refine the identified use cases with specific attributes and values and how exactly they correlate with the sales drivers. More importantly, this helps identify new attributes and values that need to be captured in the Item Master. This process feeds back into the previous step to iteratively identify more use cases and expand on the identified ones. A few back-and-forth iterations will bring up a finite set of use cases, item attributes, and values.

This process (figure 5) solves many challenges and answers many questions that the retailers struggle with.

1. What is the right set of attributes for each product category or product?

2. What are the various values for each attribute that are relevant for business?

3. What is the purpose behind collecting the rich set of item data? How are we going to leverage this information to drive business value?

Figure 5: Use cases connect item attributes and values to sales drivers

Applying new methods to provide a systematic and efficient process

Our current research and implementation efforts are to leverage AI to systemize this process of automatically identifying attribute values for new products while cleaning up Item Master data in support of the many use cases and correlations outlined here. We see immense possibilities with AI-based image recognition to populate product attributes and ML algorithms to clean up existing Item Master attributes. The structured process outlined above gives the algorithms the learning context and process steps to derive rich and correct attributes, realizing the use cases practical and achievable.

In conclusion…

Retail organizations are already late in taking a serious look at the possibilities offered by rich product attributes in realizing business growth and expanding the customer base. A structured process supported by advanced analytical methods is the only way to make it happen. Connect with UST for a comprehensive roadmap, implementation capabilities, and subject matter expertise.