1 of 16

Space Models:

Optimizing store space allocation

at Target

2 of 16

Problem Statement

3 of 16

Retail Concepts

Luggage

Electronics

Sporting goods

Apparel

Cleaning supplies

Pet care

Food and Beverages

Checkouts

Toys

Store Layout

Shelf Space

4 of 16

Opportunity

  • Annual store remodels

  • Transitions and new stores

  • Optimal space usage within stores

5 of 16

Space Allocation Overview

3-pronged approach to space recommendations

Strategy

Multi-variate regression analysis that outputs optimal space allocation to maximize sales $ and gross margin $

“Build the Box” alignment that provides space constraints (min/ideal/max) and prioritization by division

External insights from outside our walls to make holistic space recommendations

Store-Specific Space Recommendations

Modeling

External Insights

5

6 of 16

Space Modeling Framework

Regression Analysis

Attempts to isolate historical relationship of Sales/Margin to Space by controlling for other variables.

Plot Sales +

Gross Margin

Regression curves (store-specific by division)

Optimize

Finds optimal points on the curves to optimize total store sales and gross margin, forcing trade-offs across divisions.

Output

Recommends space allocation for each division.

Space Model is a multi-variate regression analysis that outputs optimal space allocation to maximize sales $ and gross margin $.

7 of 16

Multi-variate regression models

A statistical technique that uses several explanatory variables to predict the outcome of a response variable.

Algorithms and Metrics Used

 

8 of 16

Model Selection

Algorithms and Metrics Used

 

 

Simple Linear Regression

X

Y

9 of 16

Given a set of items, each with a weight and a value, determine which items to include in the collection so that the total weight is less than or equal to a given limit and the total value is as large as possible.

Algorithms and Metrics Used

0/1 Knapsack Problem

Subject to

Space Constraints

Maximize

 

Tablet

V: $200

W: 1 lbs.

Camera

V: $400

W: 2 lbs.

Phone

V: $300

W: 0.43 lbs.

Laptop

V: $1300

W: 3 lbs.

Knapsack

W: 5 lbs.

10 of 16

Key Metrics for Optimizing Space: Productivity and Elasticity

Space Productivity

A rate that expresses the value ($ sales or GM) generated by a unit of space

The names and numbers provided here are for illustration purposes only.

    • Productivity is static
    • Example: Apparel at $95 Per Store per Foot per Week (PSPFPW), on average, is more productive than Electronics at $50 PSPFPW

11 of 16

Key Metrics for Optimizing Space: Productivity and Elasticity

  • More $ are generated when moving from point A to point B than from point B to point C
      • A-B is relatively elastic (steeper curve)
      • B-C is relatively inelastic (flatter curve)
  • Productivity at A, B, or C is $ divided by footage

Space Elasticity

Describes the relative relationship between space change and the change in performance consequently.

The names and numbers provided here are for illustration purposes only.

    • Space elasticity is dynamic
    • Example: Elasticity measure would tell us which among Apparel and Electronics is more elastic.

12 of 16

Space Model Output for a Store

  • Regression Plot

Division 

Current Space

Model Recommended Space

Personal Care

116

184

Beauty

300

410

Paper and Plastics

100

124

Luggage

56

54

Pets

82

65

Shoes 

105

55

The names and numbers provided here are for illustration purposes only, and do not represent actual figures.

The names and numbers provided here are for illustration purposes only.

13 of 16

Process and Architecture Workflow

The user requests for a model to be built.

This request is stored in a database and then triggers an Oozie workflow.

The Oozie workflow triggers the model build code

Model build code is executed, and the output is stored in the database, which is then rendered to the UI.

UI

14 of 16

Space Model Impact

Division

Last

Year Priority

Current Year Priority

Significant Prioritization Shifts

Division A

4

1

+3

Division B

3

2

Division C

5

3

Division D

10

4

+6

Division E

1

5

-4

Division F

6

6

Division G

2

7

-5

Division H

9

8

Division I

8

9

Division J

7

10

-3

Ideal to Max Space

At or Below Ideal

Nearer to Mins

The allocation of space for different divisions in the current year is determined based on the outputs of the Space Model, strategy outcomes, and external insights.

The process aims to strike a balance between the needs of different divisions.

Divisions within the green box will see a space increase and divisions in red will see a decrease in space.

15 of 16

Conclusion

Space Models

    • ease the decision-making process for merchandise specialists

    • help allocate space to merchandise groups

    • recommend expanding and shrinking space

16 of 16

Thank you