In-Store Service Automation

Using Google Home

The Home Depot

Summer Intern Project


The goal of this project was to automate the process of setting up a service offered by The Home Depot with the help of a Google Home device for customers in stores.
My role as the UX designer was to design and modernize the voice experience the
user would have with the device.
To comply with my non-disclosure agreement, I cannot show the crucial parts of the project. However, I can outline the higher level objectives and, more importantly, my role as the UXer on the product.
+    To automate an in-store service and create awareness about the service with the help
      of a Google Home device.
+    To create a production ready prototype of the product during the 12-week period.
+    Integrate with an existing voice agent (currently used for a different service).
At The Home Depot, it is highly encouraged to follow hypothesis driven research. During the discovery phase, generating a hypothesis helps guide the research while also keeping the team aligned on the project objectives.
Google Home
Generates more product
Saves time for store associates
The Home Depot Design Process
Problem Discovery
Research, Synthesize
Solution Framing
Ideate, Prototype
Solution Iteration
Build, Measure, Learn
We used the Enterprise UX team's Discovery and Framing process, which is based on the double diamond approach. Due to the nature of our project, we focused more on the Alignment phase during the first few weeks, and then worked on creating rapid prototypes for usability testing during the Solution Iteration phase, which falls in line with the company's push for agile methods. 
Value Proposition
To fully understand the value proposition of our product, we have to take a step back and look at the smart speaker market. This market isn't just growing, it's exploding.
Google Home sales were up 483% in Q1 2018, and it is reported that 1 in 5 smart speaker owners use the product for e-commerce.

Source: Edison Research Smart Audio Report | Canalys estimates, Smart Speaker Analysis, May 2018

Gartner, Edison Research 2017 | Strategy Analytics, February 2018

What does this mean for our product?
As more people get familiar with the technology and its applications, we believe that user engagement with the device would increase with time, even in unfamiliar environments like Home Depot stores. As the number of people utilizing the platform for the service goes up, it directly translates to increased revenue for the company. 
Daily stand-up meeting
Pair programming
EUX Design Critique
Local usability testing
In-store testing
Concept validation
To help us we participate in activities such as daily stand-up meetings, pair programming, and alignment meetings. 
We conducted a  Design Critique  midway through our timeline where we presented our work to various Enterprise UXers and got feedback on the product. This proved to be really useful as we got pointers from associates experienced in the UX field who had no knowledge about our space, and thus eliminated bias. 
The Balanced Team
Austin H
Product Manager
Varun Nambiar
Enterprise UX
Vita Tran
Software Dev
Ananya Gupta
Software Dev
The Home Depot follows a Balanced Team philosophy, where every team is comprised of at least one product manager, a UX designer and engineers.
Dialog Design
Assurance about
Qualification call
After I set this up, one of our
 associates  will give you a call you by
the end of the day to confirm the
appointment. What product are you
looking to install or replace?
end of the day
install or replace
Alright! I'll need some basic information from you, like your
name, phone number and zip
code to proceed. Is that okay?
name  phone number         zip
Customer Experience Team
Home Services
EUX Design Critique
Customer Appointment Center
Engineering Team
Legal Team
The conversation was designed in a manner to  minimize cognitive load. At every point of the dialog, the awareness related items were stated earlier followed by the action statements. The aim was to keep the interaction concise, and to the point while keeping it sounding natural (which was achieved by using Google's Wavenet voices, which is an advanced generative model for raw audio). 
We worked with various teams across the company, to make sure the language we used played well with all departments. 
To account for the unhappy path, instead of having generic Interactive Voice Responses (IVR) for error cases, we created  unique fallback options  at every point of the conversation to be able to keep the user within our flow to get them to the desired output. 
Here's an example of how we designed around acquiring
a user's phone number in multiple ways. 
Could I get your
phone number?
 Case 1 - Happy Path 
Sure, it's
Could I get your
phone number?
 Case 2 
Go ahead, I'm listening!
Could I get your
phone number?
 Case 3 
You could give your phone number by speaking the 10 digits clearly!
Could I get your
phone number?
 Case 4 
Unrecognized phrase
Research Methods
The majority of design decisions were made through primary research methods conducted in the form of observations and usability studies. 
Usability tests were conducted in three stages. 
Phase 1
Table Reading
For the initial prototypes we used the table reading method, where we wrote scripts and had users read it out and give feedback on the flow and phrasing.
The ease of this process allowed us to quickly test various ideas in a short amount of time.
Phase 2
Wizard of Oz Testing
To add a layer of authenticity,
we played Text-To-Speech audio clips via a computer to simulate a Google Home device. 
This allowed for users to give feedback not only on the flow, but also the interaction experience with the device. 
Phase 3
DialogFlow Testing
For our final set of prototypes, we built our dialog in DialogFlow, Google's design tool for building voice interfaces. 
Users were tested on the device directly, creating the truest form of interaction between the user and our product. 
DialogFlow is Google's design tool for natural language processing. We used it design the conversation flow at the front end as well as have it communicate with our code hosted on Webhook on the back end .
It allowed us to build prototypes quickly which gave us the opportunity to test with users early in the design process.
Awareness Generation
Awareness generation required checking off two boxes:
+    How do we let the customer know about the service that we offer without having a store associate present?
+    How do we make the customer start interacting with the Google Home?
 Aisle Violator 
With service?
With service?
We designed two signs, a 60" tall 'sidecap sign' that serves to draw attention from a distance, as well as a smaller 'aisle violator' sign that gives just the right amount of information to get the customer to engage with the device. 
Multiple iterations of both signs were created and tested. 
We built this project from scratch and by the end of week 12, we achieved our goal of building a production ready prototype.
We were also given the opportunity to present our product to the VP of Home Services, as well as the VP of Service Operations at
The Home Depot.
The product is planned to be piloted in-store in Q4 2018. 

© 2021 Varun Nambiar