Portfolio

AI Summary

“Just in one day we can have one or hundreds of conversations and yes, I need to review each of them” for a conversational tool, the increase of chats it’s a good new, but not for everyone like the Sales Rep. and Admins that must review each chat to understand what are happening in the conversations and if the bot it’s answering the way it supposed to be, and the answer was in front of our eyes: AI

Company:

DRIFT

Role:

Product Designer, UX Researcher

Date:

Feb 2024

Tools:

Figma, GreatQuestion, Miro

Why does a new feature had such a negative feedback?

On the second half of 2023, Analytics team decided to migrate from a untenable but interactive section to review the conversations, to a sustainable but simple data table. This change immediately receive a negative impact from the users and the amount of complains and tickets in Customer Suppot increase critically.

R E S E A R C H

We might empathize with our customers, understand their workday and how did they use and expected from our platform. After multiple rounds of discovery calls and reading all the feedback collected, we figure out their main pain points and other interaction problems that we might be facing by updating this section.

“I just simply don’t get it. Why on earth did you sunset it? It was so cool, I was able to easily review the chats just scanning the page and now I’ve to go one by one, spending a lot of time on it”

“Other tools in the bot’s market allow me to do faster my job, not slower. The amount of time I spend doing my job it’s really important to me, it also impacts on my comission”

P A I N · P O I N T S

I N T E R A C T I O N

The usability of the page requires a lot of clicks to review the chats.

A R C H I T E C T U R E

Not enough or easily to find information of what is happening during conversations.

A R C H I T E C T U R E

Drill-down to other reports like Playbooks and Opportunities

Design challenges

How might we recover the acceptance of our customers to the new report?

How might we create a cohesive experience for our users?

How might we prioritize the usability of the product?

Design challenges

How might we recover the acceptance of our customers to the new report?

How might we create a cohesive experience for our users?

How might we prioritize the usability of the product?

·

My biggest challenge during this project, was to make our users change their minds about the negative perception of our new Chats data page and naturally encourage them to use it.

To achieve this, we must understand exactly which information our users were looking for and the priority of each, so the solution can be designed based on it. Also as a request from the Product VP, we must start including in the improvements a feature that uses AI (Chat GPT).

·

My biggest challenge during this project, was to make our users change their minds about the negative perception of our new Chats data page and naturally encourage them to use it.

Design challenges

How might we recover the acceptance of our customers to the new report?

How might we create a cohesive experience for our users?

How might we prioritize the usability of the product?

·

My biggest challenge during this project, was to make our users change their minds about the negative perception of our new Chats data page and naturally encourage them to use it.

Design challenges

How might we recover the acceptance of our customers to the new report?

How might we create a cohesive experience for our users?

How might we prioritize the usability of the product?

·

My biggest challenge during this project, was to make our users change their minds about the negative perception of our new Chats data page and naturally encourage them to use it.

V I S I O N · A L I G M E N T

V I S I O N
A L I G M E N T

U S A B I L I T Y

The usability of the page was the most critical pain point to solve and the one that requires more effort from the engineering team, since some microinteractions were not contemplated (at least at this point) in the DDS.

A possible solution for this might not be only moving a button, since the problem of closing the drawer and click on the next still there. Changing to clickable rows was not allowed by our system.

K E Y · I N F O R M A T I O N

Analyzing the feedback that we receive about sunsetting Convo Analysis page, we divided it in 3 features to have in mind for the Chats data improvements:

01 · QUICK FILTERS
Instead of following the current pattern to find out the information, they can quickly select them.

02 · "SNEAK PEAK"
Showing an small amount of the conversation make them understand what it’s happening in the chat.

03 · OVERALL INFORMATION
C.A. page didn’t have it either, but it’s something that we also receive as feedback.

Analyzing the feedback that we receive about sunsetting Convo Analysis page, we divided it in 3 features to have in mind for the Chats data improvements:

I N F O R M A T I O N · A R C H I T E C T U R E

I N F O R M A T I O N

A R C H I T E C T U R E

Following the new flow that we set up in the ROI project, Chats data are now more important than ever.

The goal is to connect them to the activity timeline, so our users will be not only able to understand if an opportunity it’s “influenced” or “sourced”, they can also dig into the actual conversation and see what happens there and why we’re claiming it.

D E S I G N & P R O T O T Y P E

D E S I G N &
P R O T O T Y P E

01 CARDS

Instead of continue having a data table, my proposal was to go back to a more interactive way to review chat by chat by just scanning the page easily reading the summaries and the topic tags.

01 CARDS

Instead of continue having a data table, my proposal was to go back to a more interactive way to review chat by chat by just scanning the page easily reading the summaries and the topic tags.

01 CARDS

Instead of continue having a data table, my proposal was to go back to a more interactive way to review chat by chat by just scanning the page easily reading the summaries and the topic tags.

01 CARDS

Instead of continue having a data table, my proposal was to go back to a more interactive way to review chat by chat by just scanning the page easily reading the summaries and the topic tags.

02 TOOLTIP

An easier way to show up the summaries and that it can be meaningful for our users it’s to have access to it before opening the chat.

02 TOOLTIP

An easier way to show up the summaries and that it can be meaningful for our users it’s to have access to it before opening the chat.

02 TOOLTIP

An easier way to show up the summaries and that it can be meaningful for our users it’s to have access to it before opening the chat.

02 TOOLTIP

An easier way to show up the summaries and that it can be meaningful for our users it’s to have access to it before opening the chat.

03 DRAWER

The quickest way to start introducing the AI Summary it’s by keeping the data table and the drawer but adding some small improvements that upgrade the usability.

03 DRAWER

The quickest way to start introducing the AI Summary it’s by keeping the data table and the drawer but adding some small improvements that upgrade the usability.

03 DRAWER

The quickest way to start introducing the AI Summary it’s by keeping the data table and the drawer but adding some small improvements that upgrade the usability.

03 DRAWER

The quickest way to start introducing the AI Summary it’s by keeping the data table and the drawer but adding some small improvements that upgrade the usability.

E A R L Y · T E S T I N G

As a result of the early testing in this project, 11 of 15 users preferred Version 01, 3 of 15 Version 02 and only one the Version 03. However, during some session with the team, the preferred version was the number 03, since it requires less maintenance and it’s easily to implement.

I moderate two sessions with key stakeholders to find a resolution for this controversy.

In the first session we had an informal discussion where the creative and engineering team share the reasons why should be one version or the other. For the second session, I synthesize the different insights and with both perspectives in the table, we vote.

F I N A L · D E S I G N

K P I · C A R D S

I decided to highlight 3 key data related to the chats: First the sentiment, a new feature that also uses AI to detect if the conversation went with a positive mood, negative or neutral.

S N E A K · P E A K

We used to show up the first part of the conversation in C.A., but I changed to the last part, so our Sales Rep. can understand why the conversation ended.

R A T E · T H E · B O T

To introduce a way to give us an evaluation of the Chatbot performance, we’re also using the “sneak peak” space to it.

N E X T · S T E P S

After the MVP version, the next steps for this project it’s to also include a full version of the chat integrated and AI filters that also include a new way to easily select the right filters to find the information that we’re looking for, just using a sentence.

This project was programmed to start the development on March 3rd, 2024, but was postponed until further notice.

All projects

Transform problems
into solutions.

Let's talk!