Tyler Sarkis CREATES

Overview

Problem

The Inventory, Planning, and Replenishment (IPR) department at The Home Depot is categorized into several sectors depending on the flowpath, origin, and retail space of inventory. Something The Home Depot (THD) prides themselves on is home-grown electronic systems and reporting. Utilizing these, many of the domestic flowpaths have clear line of sight into quantities, timing, and proactive alerting before something goes wrong operationally.

An exception to this is the Bulk Distributions in-bound ordering team. There are several layers of complexity to this specific line of business, starting with how the vendors are contracted. Because the vendors in this space operate as consignment vendors, the BDC analysts do not always have visibility into the quantity of inventory en route, ETA's, nor alerting of delays. This causes problems within the entire supply chain, and inability to answer questions such as:

Where is the product?
When will it arrive?
How long will current inventory support business given historical sales and forecasting?
Is there enough coming to support sales?
What portion of inventory on the way is already spoken for?
If there was a delay, what caused it?
If we are running low on inventory, what are the main drivers?
At what point do we stop waiting for the responsible vendor to send inventory and look to place an expedited order from a third party?

Assumptions

Some of these problems could easily be solved by adjusting the vendor expectations and contracts.
- That is very hard to do.

If the vendor owns the transportation process, we should be able to get tracking numbers and gather ETAs and tollgates ourselves.
- Also very hard to gather.

We should have direct communication with the Distribution Centers to understand inventory levels and space constraints. At minimum, we should be able to alert them to not turn product away for quick cross-dock scenarios.
- Again, with a company as large and high-velocity as THD, this is a massive and nearly impossible lift.

The BDC In-bounds team is a group of seven associates. Additional initiatives within the IPR Solutions department sought to understand what amount of time and effort went into non-value add manual tasks that could be easily automated. Through shadowing and interviewing, I was able to gather a pulse on task-time allocations, friction points in established processed, and an understanding of the amount of time that goes into non-value-add tasks.

Top Friction Points

Commentary:
- Lack of practical understanding regarding use case at the analyst level.
- Tedious revision cycles from leadership or disagreement on root causes following research.
Reporting:
- Reactive and limited to visualizing the coming week.
- Delayed by "fire drills" and external task requests.
- Significant data discrepancies and reliance on guesswork.
Vendor Emails:
- Reporting utilization for vendor communications causes computer lockouts.
Lack of DC Space Visibility:
- Slow internal negotiations for space capacities/constraints.
- Recalculation of capacities takes substantial analyst time annually.
Lack of Issue Foresight:
- Inability to anticipate and address problems proactively. through in the photos.

Non-Value-Add Solution Opportunities Targeted

Daily Data Refresh and Manual Reporting:
- Dedicated associate performs daily refresh, leading to large file sizes and slow access.
- File size contributes to computer corruption issues resulting in multiple computer replacements.
- Manual creation of vendor performance email weekly.
Exception Reporting:
- Goal to provide preemptive alerts to address issues before they escalate significantly.
- Key pillar to automating vendor performance emails.
Out-of-Stock (OOS) Pressure Commentary:
- Pressure commentary required when inventory levels drop below certain thresholds.
- An AI and machine learning tool was developed for auto-populating commentary for other branches of IPR. Goal to expose the BDC space to this tool.
Z-Stock Ordering and Timing:
- Outdated software used for ordering. Often, this will lead to resolution too late, resulting in missed sales.
BDC Max Inventory Level Review:
- Manual annual data refresh for capacity calculation.
- Aim to automate data refresh and optimize inventory management.

In researching current state commentary process, I was able to find what the most common root causes are reported in the BDC space.

By pulling the previous 2 months' worth of data, I created a summary page full of discrepancies. These can be linked to human error, lack of communication, and a lack of standardization. There is a strong lack of use case understanding as well -- once the analysts submit commentary on main drivers, they don't know what those comments are used for or who reviews them.

I was able to create the pie chart visualized, creating themes around the mis-aligned root causes. This was used to engage the Business Intelligence (BI) team to begin development of logic that would flag each of these.

While BI was beginning research, I began meeting with BDC's and my leadership to align on direction. These meetings were scheduled on a weekly-basis and quickly got very technical and off track when BI began presenting the logic supporting analysis. It became clear that our business partners were not understanding the BI terminology that was going into the SQL query to define root causes. Additionally, due to the confusion and tension, we found ourselves discussing the same topics week after week and we were not moving forward.

I jumped in and was able to set up a weekly cadence to meet with our BI partners, create a deck that digested and translated BI language to business language, give our BDC partners time for a pre-read, and presented the material in front of both parties to ensure clarification and understanding.

In order to automate the data refresh of the reporting used, we first needed to understand what each of the 186 columns were used to inform. Through excel formula analysis and interviewing analysts, I was able to create a definitions document that informed what the data in each column was used for, and where that data came from. This was sent to the Product and IT teams to begin creating a query that could replicate the manual file automatically.

Upon first inspections, there were 16 columns that were targeted as the "most important" for reporting, and the calculations had much discrepancy between what was populating manually and what was populating from the query. These columns primarily inform minimum need to support sales, what is already on the way, and how much will be needed to support sales for next week.

An analyst was identified to validate data pulled on a weekly basis and report out to the weekly Product/IT/IPR leadership teams. After several weeks of no progress, we realized that additional complexity meant the analyst did not have the availability to complete user testing to move this project forward. I was able to step in as the test user as without user testing, this project would have failed and could not have continued progress in a timely manner.

Current state, after two months of comparing 30k lines of data on a weekly basis, looking for discrepancies - this tool has been rolled out in Beta view to the BDC team. In partnership with BI and BDC leadership, I aided in onboarding the BDC team to an excel interface that further simplified the process by connecting reporting directly to big query. The team simply clicks the refresh button in the reporting file to get the most up-to-date information.

Next steps are to create hardened data sources on the IT-side, then expose this to Product for full systemic support.

The Home Depot - Bulk Distribution Center (BDC) Ordering Efficiency

Overview

Problem

Assumptions

Research and Planning

Top Friction Points

Non-Value-Add Solution Opportunities Targeted

Development

Out-of-Stock Root Cause

BDC Meeting template

ROOT cause heirarchy

Current State

Automated Check File Query