Open D Express

Play

Open D Express is a web-based chatbot that offers delivery services.

Open D Express

WEB-BASED CHATBOT

Currently, Open D Express is part of OpenDialog's Academy and OpenDialog's user manual.

Even though the chatbot allows users to schedule and reschedule the delivery of their packages when these are in transit, among other things, the chatbot was built as a demo to illustrate OpenDialog's software capabilities to potential users and clients.

Tools used

Projectdetails

Open D Express is a web-based delivery chatbot that I developed end to end while working for OpenDialog.

Test

Implement

Prototype

Ideate

Define

Empathize

UX design thinkg

Methodology

UNDERSTAND

Test

Implement

Prototype

Ideate

Define

Empathize

UX design thinkg

Methodology

some work samples

I just realized I won't be home when my package gets delivered, so I need to reschedule the delivery soon

I ordered something online and I want to make sure it gets delivered when I'm home

User profiles

In order to understand the needs of users who may interact with a delivery chatbot like Open D Express, I started by drafting some user profiles. These profiles drew from quotes from actual users and allowed me to learn more about the context where this chatbot may be used as well as users' goals, needs and wants. User profiles are relevant in that they provide a good basis to start designing a product.

EXPLORE

Test

Implement

Prototype

Ideate

Define

Empathize

UX design thinkg

Methodology

I brainstormed a list of adjectives to define the chatbot's speech style

I wrote sample phrases to experiment with the tone of the chatbot

I used the Big Five to craft the chatbot's personality

Chatbot's personality and tone

It is important for chatbots and voice assistants to build trust with users. To achieve this, the chatbot's personality and tone are key. I started by doing some research to compare the personality and tone used by other delivery chatbots already available in the market. This allowed me to have a clearer idea on how I wanted Open D Express to interact with users.

Sample conversation flowchart

The resulting flowchart was a diagram that depicted the so-called "happy path" and other possible alternative paths in the conversation. I also took into account API integrations (e.g. access to a database with tracking numbers for orders).

Flowchart

Drawing from the use case and the user profiles, I created a flowchart to represent the high-level structure of the chatbot.

MATERIALIZE

Test

Implement

Prototype

Ideate

Define

Empathize

UX design thinkg

Methodology

Sample utterances

STEP 4

Main components

Turns

Transitions

STEP 2

STEP 3

STEP 1

Chatbot building

I built the chatbot using OpenDialog's software. I used the flowchart I had previously created as a basis and I mapped the different flowchart's actions into various conversation components.

see some examples

SHORT MESSAGES

Users sometimes skip or ignore information when the message is too long. To facilitate legibility, I broke long messages into shorter ones. Usually, I included only one sentence per message.

CONSISTENCY

To avoid confusing users, I was consistent with the terms used in the messages. For example, I always used the term "schedule" instead of "arrange", "program", or any other synonym.

ACTIVE VOICE

I avoided using the passive voice in the messages whenever possible, since it can be vague and ambiguous.

CONVERSATIONAL TAGS

I added messages with tags, such as "ok" or "great" throughout the conversation. These show that the chatbot understands and that it is being cooperative.

Drawing from the chatbot's personality and tone that I had defined, I wrote the messages for Open D Express. In this process, I followed conversation design and UX writing best practices.

Message writing

Play

Play

Play

Wherever possible, I gave users the option of responding either via a button click or via text. In both cases, the chatbot would be able to handle their responses and continue the conversation.

MULTI-RESPONSE

If none of the times provided for delivery works for the user, they can provide their desired delivery time and an agent would contact them to confirm their choice. This is possible through an API integration that stores users' emails and desired delivery times in a database.

AGENT

Chatbot optimization

I added an FAQ conversation component to the chatbot, in case users had questions. To do this, I created a Knowledgebase on Dialogflow and integrated it with the chatbot.

KNOWLEDGEBASE

When entering important information, a user's input may contain typos. To prevent this from happening, I added confirmation messages wherever necessary.

When given the "schedule" and the "reschedule" options, a user may choose the wrong one. For this reason, I added a way to redirect users to the right conversational path based on the tracking number provided.

A user may not know where to find their tracking number when asked about it. For this reason, I added a short description explaining where they have to look for it.

To prevent users from having an unpleasant experience while using the chatbot, I anticipated some common problems they may have while interacting with Open D Express and found creative solutions for them.

REDIRECTING USERS

CONFIRMING USER'S INPUT

PROVIDING EXTRA INFORMATION

Anticipation of issues

TECH INNOVATION

COMPANIES

DEMO USERS

feedback to improve the chatbot will come from three main sources

As technology evolves and OpenDialog's software offers new features, these will be added to the demo in order to showcase what OpenDialog can do.

Companies potentially interested in using OpenDialog for their business needs will also have access to the demo and will provide feedback that could be used to highlight certain chatbot capabilities.

As people use the demo to learn about OpenDialog (e.g., through the Academy) , I will get feedback from these individual users to improve the chatbot. This information will be collected through surveys.

Metrics for success

Since the Open D Express chatbot was developed as a demo, metrics to measure its success cannot come from actual users that want to track their package deliveries, but rather from people who will use the chatbot as a demo to learn about OpenDialog's capabilities.

Thank you!