Voice Skill for Placing a Pick-up Order for Coffee

Coffee has become a morning tradition for many people, it’s part of just about everyone’s routine. My assumption here is that most people consume coffee in the morning while heading to work. Walking-ordering-waiting to drink coffee takes at least 20 minutes out from their work time.

Goal

What if a voice could place an order for coffee so that one could save time on ordering and waiting at the shop.

Target Users

Coffee lovers who use Starbucks app to place a pick-up order ahead of time.

My role

Survey, Design proposal, existing app feature analysis, Feature conceptualization, Persona, Dialogs with error and recovery, Usability testing, and Conversation flow

Team

Individual Project

Duration

May - June 2019 (4 week)

Research Method

I created a survey using Google forms: In the survey, I asked multiple choice and written response questions to identify the behavior of users in order to design a voice skill for coffee lovers.

Findings and Insight

Assumption: Do users want to save time by placing an order for their coffee ahead of time?

• If users are using the Starbucks app to order coffee ahead of time, then they might as well use voice skill to place an order from their car or home ahead of time.

Have you ever used the Starbucks app for placing a pickup order for a cup of coffee?

Findings: 55% of people use Starbucks app to place a pickup order ahead of time.

Assumption: Do most of the users prefer to drink the same type of coffee every time?

• This would help me to add a favorite feature in the skill, so users can repeat the same order every day.

  Do you prefer to drink same type of coffee every-time?

Findings: 73% of the users said that they prefer to drink the same coffee every day. This also concluded that the skill should account for direct orders from Starbucks menu. Like “ Place an order for Mocha from Starbucks” rather than going through the menu.

Assumption: Do most of users prefer to drink coffee in the morning while heading to work?

• If most of the users drink coffee in the morning, then they are more likely to use this skill while driving to work.

What time of the day do you prefer to drink coffee?

Do you usually pick a cup of coffee while heading to work?

Findings : 90% of the user drinks coffee in the morning, and 40% said that they pick-a-cup of coffee while heading to work.

Design Proposal

An interactive voice assistant to help coffee lovers place a pickup order from Starbucks. The most important aspect of the feature would be to utilize the time while the user is driving and place an order for their favorite cup of coffee using a voice assistant. The user could save his order in favorites for the next time.

Feature

1. Ordering coffee from Starbucks

2. Customizing the order

3. Saving orders in favorites

4. In-skill payment by saved credit card

5. Reminding the time left to pick up the order

6. Postponing the order

Persona

Use Case 1

Anna is running late for a meeting scheduled at 8:00 AM. She wants to place an order for Caffe Americano with 2 shots before heading to her office for the long day meeting.

Happy Path Dialog

Usability testing and Learnings

I used the “Happy path dialogs” to test the concept and learned a lot from the guerrilla testing conducted with my class mates.

Learnings

Please refer the numbers in between both the dialogs

1 and 5 - Never ask a question in the middle of sentence and always end a sentence with a question.

2 and 3 - Ask clear and concise questions.

3 - Always offer distinctive sounding options “ Hot or cold coffee”

4 - Always provide 3 or fewer options

6- Never put the burden on the user

Happy Path Dialog with Error and Recovery

Interaction

Here an an example of script to show how system understands the pattern and value to give an output as logic.

Use Case 2

Anna wants to postpone her order by 10 mins as she got stuck in traffic and her google maps is showing 10 mins delay to reach her office.

What I learned

Never ask a question in the middle of sentence and always end a sentence with a question.

In the usability testing sessions, I observed participants were trying to answer after the first question: for eg, “what size would you like to order for Caffe Americano? 8oz, 12oz, 16oz or 20oz” but at this point the voice speaker is still talking, and is not listening to what the user is saying. So by writing dialogue this way, there's a high likelihood that the interaction will be unsuccessful, either because the smart speaker hasn't listened or the user doesn't know how or when to answer. This can be frustrating for the user.

Ask clear and concise questions.

A good voice experience should aspire to feel like a natural conversation, Asking clear and concise questions like “ What would you like, Hot or cold coffee? can prevent the conversation from becoming lengthier than it needs to be.

Always provide 3 or fewer options

More than three options are difficult for users to remember. This can mean users are busy thinking about what they've forgotten when they should be making a decision. Decision-making involves weighing up the available options, and that's very difficult to do if you can't remember all of them.

For example, The seven coffee options are split into smaller chucks of 3 options at a a time “We have got seven hot coffee options. 1.Caffe Americano 2.Brewed coffee or 3. Cappuccinos . Choose one or ask for more'“. Users are able to process the information more easily, weighing up three options and deciding if they like what they've heard or if they want to hear more.

Never put the burden on the user

When a voice experience blames a user for the problem, it can go beyond frustration and become upsetting.

Asking the user to remember their order will burden them to save the name in their memory for future use.For example, “To place an order next time, you could just say order my favorite Caffe Americano”. rather a good practice would be too just say “To place an order next time, just say "order my favorite coffee at Starbucks"

Add an Extra Layer of Security

During the test, users asked how would the payment take place after placing an order. I replied that the system will use the existing saved payment method. Users were worried what if someone else tries to place an order from my voice assistant.

For such scenario there should be an extra layer of voice security where the system could recognize the voice of the user.