Artificial Intelligence
Self-Assessment for Retailers

Today, artificial intelligence (A.I.) is finding its way into retailer processes to automate and improve key capabilities. But, as in any journey to improve processes, it’s essential to begin with an assessment of your current capabilities. This self-assessment is designed to help you do just that. Answer these questions to find out where your retail enterprise falls on the scale of A.I. readiness.

How A.I. mature are you? Answer 7 simple questions to find out.

^{Section 1}

Supply Chain - Allocation and Replenishment of Inventoryi

The most expensive component of retailing is the cost of inventory, so making sure the right products are in the right place at the right time is critical to profitability. A.I. can help determine how to allocate and replenish products in the most efficient way. A.I. also helps with fulfillment decisions executed by order management and can be used to automate warehouses.

The most expensive component of retailing is the cost of inventory, so making sure the right products are in the right place at the right time is critical to profitability. A.I. can help determine how to allocate and replenish products in the most efficient way. A.I. also helps with fulfillment decisions executed by order management and can be used to automate warehouses.

How are you handling allocation and replenishment of inventory?

0) No capability - A.I. is not used in supply chain.

1) Inconsistent - While a forecast is available to drive inventory decisions, most inventory is allocated or replenished through manual, spreadsheet-based ladders and push tools.

2) Inefficient - A linear replenishment algorithm is utilized but must be manually adjusted frequently based on budgets, supply availability and promotion. Allocations continue to be created in spreadsheets.

3) Competent - Both Allocation (push) and Replenishment (Pull) decisions are algorithm based with <50% of the decisions manually adjusted, but they are separate algorithms and UX that require a category to be forced into one model or other even though some items should float between.

4) Performing - A single A.I.-based network optimization model that accounts for push and pull of inventory based on sales plans, item lifecycle in a store, inventory levels, price/promotion activity with less than 30% of decisions requiring manual adjustments.

5) Differentiating - Machine learning or even deep learning are used and are fully automated, requiring minimal human oversight. The UX is alert vs. spreadsheet oriented.

^{Section 2}

Demand Forecastingi

An accurate demand forecast helps to deploy resources efficiently, but its difficult to guess from one season to the next. Traditionally, this was done by very experienced merchants that leveraged their years in the industry. Then statistics were employed to base a forecast on previous history, often incorporating new trends. Now, machine learning can be used to forecast based on product attributes, which is much more accurate.

An accurate demand forecast helps to deploy resources efficiently, but its difficult to guess from one season to the next. Traditionally, this was done by very experienced merchants that leveraged their years in the industry. Then statistics were employed to base a forecast on previous history, often incorporating new trends. Now, machine learning can be used to forecast based on product attributes, which is much more accurate.

How do generate forecasts?

0) No capability - Forecasts are strictly based on last year.

1) Inconsistent - Forecasts are created by applying an average weekly sales number to a seasonal profile.

2) Inefficient - A time-series algorithm based on a like item copying of history is used for an initial forecast, then a person edits a majority of the forecasts based on experience. Some guessing still occurs. All promotions and clearance moves are entered manually.

3) Competent - Manually entered history exceptions and outliers are used to reduce the number of human edits to the baseline forecast. A causal based approach to promotional and clearance forecasting is available, but all causals must be manually merged together into a single promotional or clearance forecast.

4) Performing - Instead of forecasting demand by SKU or style-color-size, the forecast uses machine learning to forecast demand for product and location attributes so that new items can be forecasted based on demand for its attributes. Same principle for the attributes of a promotion. Forecasts are managed through a confidence factor driven alert process. No manual review is required.

5) Differentiating - Forecasting largely operates unassisted using deep learning. Most forecasting alerts are focused on lack of data (errant data is cleansed by validating against product images etc.) and supply constraints that require the merchant to resolve with the vendor. UX can be a BI solution due to the minimal editing requirement.

^{Section 3}

Customer Intelligencei

Knowing your customer is key to meeting their needs and thus selling more products. When customer attributes are collected, they can be used by machine learning to build profiles and predict customer behavior. The ultimate objective is to improve the experience so customers spend more.

Knowing your customer is key to meeting their needs and thus selling more products. When customer attributes are collected, they can be used by machine learning to build profiles and predict customer behavior. The ultimate objective is to improve the experience so customers spend more.

How are you using customer intelligence?

0) No capability - Any customer data collected is rarely analyzed.

1) Inconsistent - Occasional use of data mining to get customer insights, but always done in offline batches.

2) Inefficient - Data scientists are retained to build customer models using tools. May do simple segmentation and calculate Customer Lifetime Value.

3) Competent - Collecting and analyzing data in real-time for use in personalization of the experience, including things like product recommendations.

4) Performing - No use of segments but instead dealing with individuals based on detailed attributes like demographics, psychographics, and historical purchases.

5) Differentiating - Collecting first, second, and third-party data about customers, and building profiles used to personalized experiences and tailor offers.

^{Section 4}

Marketingi

A.I. can help market more effectively, taking into account predictive models that better align offers to customer wants. Machine learning can segment customers or build individual profiles that are used to tailor offers and increase relevance. Marketing automation tools are key to scaling efforts alongside A.I. algorithms.

A.I. can help market more effectively, taking into account predictive models that better align offers to customer wants. Machine learning can segment customers or build individual profiles that are used to tailor offers and increase relevance. Marketing automation tools are key to scaling efforts alongside A.I. algorithms.

How are you handing marketing?

0) No capability - No use of A.I. for marketing.

1) Inconsistent - Typical marketing blasts without attempts to target. Some attempts at segmentation, but very basic.

2) Inefficient - Some segmentation to better target customers, and some automation for emails and advertising online.

3) Competent - Predictive models select the best offers for customer segments based on collected profiles and deliver offline through emails and coupons.

4) Performing - Real-time marketing to customers with appropriate content based on journeys that help customer complete discover products and complete sales.

5) Differentiating - Individual target marketing so that all offers are relevant to the specific customer. Electronic selling methods are enabled to create tailored offers as the Customer shops rather than being limited to a small pool or predefined offers (instant bundles etc.)

^{Section 5}

Store Operationsi

A.I. helps with talent acquisition and talent management by assessing the skills of employees and deploying those skills effectively. It can also power vision systems that automate checkout, intelligent task scheduling, and detect fraud/shrink.

A.I. helps with talent acquisition and talent management by assessing the skills of employees and deploying those skills effectively. It can also power vision systems that automate checkout, intelligent task scheduling, and detect fraud/shrink.

How are you using A.I. in stores?

0) No capability - No use of A.I. for store operations.

1) Inconsistent - Use of knowledge-based support systems to diagnose and fix issues reported by store employees.

2) Inefficient - Using A.I. to detect unusual situations that could be indicative of fraud or shrink.

3) Competent - Using talent science to hire and better match employees to roles. Using A.I. to better predict store traffic and right-size labor schedules..

4) Performing - Use of robots for cycle counting, customer assistance, or aisle clean-up. Perhaps using natural language processing to answer employee questions or price look-ups. Store and fulfillment labor plans are updated automatically as promotion, assortment, marketing and inventory plans change.

5) Differentiating - Computer vision employed to detect product movements, helping with stock-out detection or automated checkout.

^{Section 6}

Pricing & Promotionsi

Setting prices for products that maintain margin yet are market competitive was often an art, but adding A.I. brings some science into the endeavor. Using sales data, competitive data, and forecasts can help set the regular, sale, and markdown prices for products in a way that maximizes sales.

Setting prices for products that maintain margin yet are market competitive was often an art, but adding A.I. brings some science into the endeavor. Using sales data, competitive data, and forecasts can help set the regular, sale, and markdown prices for products in a way that maximizes sales.

How are you using A.I. for pricing?

0) No capability - No use of A.I. for pricing and promotions.

1) Inconsistent - Some prices, usually markdowns, are derived using A.I., but regular price and promotions are still manual.

2) Inefficient - Use of A.I. to make various price recommendations, but not trusted and therefore overridden by humans often. Using store clusters.

3) Competent - Strong use of A.I. to set regular and markdown prices, and some use for promotions. Clearance goods are allocated to the most profitable clearance locations/fulfillment centers in alignment with markdown decisions.

4) Performing - A.I. influenced promotions to enforce category goals, including taking into account things such as cannibalization and competitive positioning. Set prices uniquely by store.

5) Differentiating - Store and Customer Segment pricing is automated to respond to recent sales/margin results, budget adjustments, inventory levels and competitive positioning with little manual intervention.

^{Section 7}

Assortment Localization & Personalizationi

Ensuring the options (products and levels of service) presented to each Customer whether she is shopping in store or in app are critical to avoiding a lost sale. The use of A.I. capabilities in the Assortment process (from going to market to assorting stores) can help ensure the shelf/rack orthe first page of products shown to each Customer contains the best collection of products that minimize her time to shop to fulfill her needs.

Ensuring the options (products and levels of service) presented to each Customer whether she is shopping in store or in app are critical to avoiding a lost sale. The use of A.I. capabilities in the Assortment process (from going to market to assorting stores) can help ensure the shelf/rack orthe first page of products shown to each Customer contains the best collection of products that minimize her time to shop to fulfill her needs.

How are you using A.I. for localization?

0) No capability - No use of A.I. for Assortment.

1) Inconsistent - Assortment analytics for last year and recent floorsets are available by cluster, but nothing forward looking (does not answer what should we assort?). Assortments are manually constructed by Store Cluster.

2) Inefficient - The assortment is forecasted by copying sales history from a like item (and then manually adjusted), once it is created by the Merchant at the store level assortments are manually constructed by store cluster.

3) Competent - The Assortment is forecasted through an attribute driven approach that avoids the need to select like items and requires few manual edits. Assortments are constructed by Store Cluster.

4) Performing - Each store’s assortment (no longer by store cluster) is optimized (recommended to the Merchant) through the use of an attribute driven approach (which attributes are most important to each store’s customers or each zip/postal/Customer cluster). Competitor’s history is considered in forecasting and localization recommendations.

5) Differentiating - 1. Before going to market, a shopping list of items is prepared for the Merchant based on attribute level trends and seasonal preferences of each store’s customers or each zip/postal/Customer cluster. 2. Store/fulfillment center level assortments are adjusted automatically (respecting buy program constraints) to keep, drop and add items to reflect recent trends, inventory levels and competitive trends.

Get your results

To see your results please fill in the following:

Fields with * are required.

First Name*

!Error message goes here.

Last Name*

!Error message goes here.

Email Address*

!Must be business email.

Company*

!Error message goes here.

Phone Number*

!Please enter your phone number.

Country*

!Please enter your country

Check this box to receive future marketing communications from Infor.

!Error validating email address.

_{Infor values your privacy.}

Your A.I. Readiness Score

Below is how you stack up against competitors who are fully optimized for A.I. retail.

Results Key

Your Score

Summary

What to do

0-5

What are you waiting for? Most of your competitors (not just Amazon) are way ahead of you. But it’s time to start adopting.

Start by making sure you are collecting the right data, like product, customer, and employee attributes. Make sure you’re collecting sales history, out-of-stock, and inventory positions. All this data needs be stored in a data lake if possible, so multiple systems can easily access it.

6-10

A toe in the water: There are time-tested accelerators to move you to the next level quickly.

Focus on any “no capability” or “inconsistent” dimensions first, especially in the areas of forecasting and customer intelligence.

Easy places to get started are Forecasting, Replenishment and Talent Science. Those are great areas to learn about A.I. and will have a material impact to the business.

11-20

Solid foundation for continued investment. Most of AI’s benefit is ahead of you.

Do the things that impact the supply chain first, then optimize the others.

Once you’ve got a solid forecast, this enables allocation, replenishment, and assortments. AI will help make sure the right products are in the right place at the right time.

Don’t stop focusing on collecting quality data – look for more data opportunities where AI might be applied. Push forward with AI in Promotion and Assortment and linking Merchant decisions to Store Ops.

21-35

Exceptional leadership in the AI area. You are at the tip of the spear.

You likely have a solid data lake feeding several AI systems to manage the supply chain and marketing areas. This is quite impressive.

At this point, it’s time to optimize and automate.

Share your score with your Infor representative to find out how Infor Demand Management can improve your A.I. readiness.

Learn more about the next generation of A.I. retail software at http://infor.com/demand-management.

Artificial IntelligenceSelf-Assessment for Retailers

How are you handling allocation and replenishment of inventory?

How do generate forecasts?

How are you using customer intelligence?

How are you handing marketing?

Store Operationsi A.I. helps with talent acquisition and talent management by assessing the skills of employees and deploying those skills effectively. It can also power vision systems that automate checkout, intelligent task scheduling, and detect fraud/shrink.

How are you using A.I. in stores?

How are you using A.I. for pricing?

How are you using A.I. for localization?

Get your results

Your A.I. Readiness Score

Results Key

Please enter your email address:

Artificial Intelligence
Self-Assessment for Retailers

Store Operationsi

A.I. helps with talent acquisition and talent management by assessing the skills of employees and deploying those skills effectively. It can also power vision systems that automate checkout, intelligent task scheduling, and detect fraud/shrink.