Chatbot scripts and content design — Hulu

 
 

The goal

In 2018, Hulu’s Viewer Experience team identified low-effort assisted support contact drivers that were forcing customer representatives to spend a lot of time on solving them, instead of focusing on more complex and high-impact cases.

Our tools and strategy teams identified that a chatbot could help by offering another self-service option to viewers. Our team selected Salesforce’s Einstein bot to integrate into our help center. However, someone needed to write the bot dialogs!

What I did

  • Identified 30 top assisted support contact drivers that could be solved via self-service content

  • Mapped user journeys for top contact drivers

  • Wrote 30 dialogs at launch (mid-2018)

  • Created a mini playbook for conversational design best practices to be used by our team

  • Defined and implemented a script creation workflow between the HuluBot product team and my team

  • Created a content gap analysis process based on comparing what users were asking the bot vs. what the bot was answering, helping define what the next scripts would have to solve for

  • Handed off conversational design duties to one of my direct reports who had an affinity and interest in UX writing and conversational design

  • Integrated a new channel into our content strategy planning

Results

  • Decreased assisted support rate for top contact drivers on average by 7% at launch (Q3 2018).
    By Q3 2019, it helped decrease the contact rate by 18%

  • Helped strengthen Hulu's brand voice

  • Added a new channel and responsibility to our team, increasing our influence as writing and content strategy experts (we were the only team at Hulu creating content for an AI product)

 

Home location “Away from home” resolution script

Hulubot 1.jpg
Hulubot 2.jpg
Hulubot 3.jpg
 
Hulubot 4.jpg
Hulubot 5.jpg
Hulubot 6.jpg
 
Hulubot 7.jpg
Hulubot 8.jpg
Hulubot 9.jpg
 
Hulubot 10.jpg
Hulubot 11.jpg
 

This was one of our top contact drivers for assisted support in 2018. HuluBot helped decrease support calls for this issue by 16% at launch (Q3 2018). By Q3 2019, assisted support for this issue decreased by 27%.

 

People I worked with

  • Product managers

  • Analysts

  • Tools team

  • UX designers

  • Assisted support customer representatives