Simplifying building custom bots

Core Problem

Building and Customizing chatbots on Haptik’s platform was a convoluted process requiring extensive engineering workarounds and adding 20-25 hrs per project and also hindering brand integration.

/

Target Audience

Platform Primary Users - Implementation team ( consisting of Conversation designers, Developers and QAs)
Chatbot Primary Users - End Users/ Customers of our clients

My Role

I co-owned this project along with the Product manager, conducted user research activities, worked on a proposal plan to convince key stakeholders, led the team through design process, implementation, re-setting scope, tracked metrics.

Key Collaborators

2 product Designers, 1 Senior PM, 1 Engineering Manager, 2 Front End Engineers, 2 Backend Engineers and the leadership team at Haptik

Major Goals/ Success Metrics


  • Consolidate various aspects of bot building and simplify the process bringing down the time spent in customization

  • Reduce the added number of hours per project down to 5 hrs

  • Upgrade the chatbot interface to be able to provide better brand integration

Problem Discovery

My first bot custom building experience was rather tedious and it piqued my curiosity about customizing for clients.

Upon further investigation through multiple conversations with customer success team, implementation team, platform product managers and gathering data on client feedback I discovered many issues.

Insights

  • Implementation team ( Conversation designers, developers, QAs )had to work across multiple sections on Conversation Studio ( our bot building platform) to build and customize a bot

  • Customization was limited, requiring extensive engineering workarounds when clients wanted to customize the bot

  • The implementation ended up spending an extra 20-25 hrs per project ( A project generally requires about 45-60 hrs ) as a result

  • This hindered brand integration significantly

  • In addition drop-off rates were high on some of the customized bots due to occasionally missed navigation flows

Roadblocks

  • This re-design had not been picked up due to complexity of the project.

  • A redesign would involve major changes to the platform and how bots were built.

  • This would potentially interfere with ongoing projects and delay them as well

  • It was high risk with unclear timelines

  • Many parts of the product were owned by different stakeholders which would mean a lot of shake up

  • I had one designer other than me and had yet to build the team.

My Approach

Getting stakeholder buy-in was not going to be easy and hence I had to go about it in an unconventional way.

  • Identified 3 chatbots with high dropoff rates, worked closely with the implementation teams to fix UX flows and navigation - This led to decrease in dropoff rates by 18%, 21% and 25% respectively within 2 weeks.

  • We conducted user research to understand chatbot perceptions, interactions, pain points while conducting a thorough audit of the platform and contextual inquiry with the implementation team on the bot building process

  • I collaborated with the senior PM of the platform and proposed an incremental platform redesign aligned with bandwidth, roadmap restructuring, and emphasized value in reducing engineering bandwidth, shortening bot building time, and enhancing UX.

  • We presented the proposed plan to our key stakeholders - CEO, CPO, CTO, VP of Engineering, Head of Marketing, Director of implementation and convinced them to add this to the roadmap

  • Parallelly collaborated with recruitment to expedite hiring two designers

Design Process

After getting the stakeholder buy-in, I took a participatory design approach involving both product and engineering as necessary in the design creation process. I then led the team through an affinity and IA mapping exercise to form a shared understanding and prioritising features. I then worked closely with the PM to plan our sprints

Phase 1

In this phase we decided to work on the following features,

  • SDK Integration - Consolidate bot customization features and reduce engg effort

  • Chatbot interface upgrade - Upgrade the bot interface to be able to provide better brand integration

  • Integration with Smart Agent Chat - To provide seamless agent transfer to reduce drop offs

  • Content Management - Able to bring together adding content in a seamless manner

  • Saved Conversations - Extend the user’s chat history from 8 hrs to 72 hrs

  • HSL Elements - Carousel, Forms - Able to have different options for Carousel, add icons to forms

  • Add Attachments - End Users to be able to add attachments

Platform Designs

At this stage, I led design brainstorming and explorations, conducted reviews and oversaw implementation, ensuring regular feedback loops, scheduling regular stand-ups to discuss progress

Chatbot Design explorations

Implementation

For Phase 1 - We were able to implement the following features

  • SDK Integration - Consolidated customizing the bot interface

  • Content Management - Able to bring together adding content in a seamless manner

  • HSL Elements - Carousel, Forms - Able to have different options for Carousel, add icons to forms

  • Add Attachments - End Users would now be able to add attachments like document and pictures

Tradeoffs

Some of the tradeoffs we had to do were,

  • Extending saved conversations from 8hrs to 24 hrs as it would increase our cloud costs

  • Agent handover process with Smart Agent Chat put on hold

  • Interface Upgrade would be taken up in the next phase as we had a bandwidth crush


Results

In the 6 months from implementation we had,

  • Reduced each project time by 15 hrs on an average

  • Seen increased adoption of the customizable HSL elements which enabled us to reduce engineering workarounds on 18 projects

  • Led us to creation of easily editable Templates which would reduce the turn around time of similar bots to 20 hrs down from 40 hrs

  • Seen highest adoption of add attachments to e-Commerce bots for returns and refunds

The interface upgrade was not take up due to a change in company strategy to move from bot focused services to other AI based commerce initiatives.

Learnings

My biggest learnings from this project were,

  • Ensuring the best way Design has a seat at the table is by doing incremental changes that show quick results through design to get stakeholders to invest in design

  • Highlighting the importance of research and securing the budget helped to drive a strong narrative of why change was needed

  • Building a strong collaborative partnership with the Product Managers and Implementation team was key to successfully execute the project and overall increase operational efficiency in the Implementation team


  • Due to lack of design documentation, I had no idea about previous attempts at fixing this issue. I authored a design document and implemented the process of mandatory documentation

Professional Product Design leader, I lead with a strategic mindset and a passion for purposeful design, translating vision into experiences that matter.

Pages

Case Studies

My Writing

Contact

deepika.a.nayak@gmail.com

+91 6366619863

+91 9972401550

Professional Product Design leader, I lead with a strategic mindset and a passion for purposeful design, translating vision into experiences that matter.

Pages

Case Studies

My Writing

Contact

deepika.a.nayak@gmail.com

+91 6366619863

+91 9972401550

Professional Product Design leader, I lead with a strategic mindset and a passion for purposeful design, translating vision into experiences that matter.

Pages

Case Studies

My Writing

Contact

deepika.a.nayak@gmail.com

+91 6366619863

+91 9972401550