Nimisha Malreddy
Product Designer & Researcher based in New York City🗽🍎. Currently a Product Design Intern @ The Met Museum



About
Resume 
Art & Design 


Bot Design for The Ordinary. Bot Design for The Ordinary. 



Role
Conversation Designer
Product Designer

Type
Academic
Contribution
Research
Training data
Sample scripting
Prototyping Duration
2 weeks

Team
Self + 1 Student





Decoding the brand











Bot’s Anatomy


Grasping the essence of the brand's personality was essential in order to accurately reflect and incorporate those characteristics into the bot's persona. This ensures the bot communicates in a way that is consistent with the brand's identity, values, and tone, thereby creating a cohesive and authentic user experience.



What can the chatbot do?

01.

Provide product recommendations based on 
skin concern.

02.

Skincare regimen recommendation based on the ‘Prep, Treat, Seal’ method.

03.

Educate users about 
skincare ingredients along 
with it’s benefits.



Personification
The bot would provides helpful skincare tips and recommendations in a straightforward and friendly manner without overtly human characteristics.
It would have a basic visual representation, in the form of an icon (Ordinary logo), to accompany its messages.





Character traits
Knowledgeable
Informative
 Friendly
 Professional
Non-Judgemental
Inclusive






Flow Diagram



Flow diagram for conversation flow




Error handling
In scenario 01 and 03, the bot can only read one concern at a time. So what happens if the user mentions 2 concerns or ingredients at the same time?

Error handling for scenario 01
Error handling for scenario 02







Sample Scripts


While crafting the scripts for the chatbot, it was important to keep in mind 3 main things:

01 Personality of the bot
The personality of the chatbot should reflect in the conversation to bridge the expectation of the bot to the pre-existing image on the brand.



02 Turn-taking cues
The chatbot should be able to guide the conversation and provide cues to the user on what to do next. This results in a successful interaction.



03 Responding to user’s needs
The chatbot should be able to understand 
the user and provide answers. If there is no suitable response, it should guide the user 
on what to do next.






Testing with Users


What users liked and their feedback:








Prototyping using Voiceflow


01 Provide product recommendations based on Darkcircles and Puffiness as skin concerns.





02 Skincare regimen recommendation for a beginner based on the ‘Prep, Treat, Seal’ method.





03 Educate user about skincare ingredient ‘Peptides’ along with it’s benefits.








Learnings


- Ambiguity on bot’s limitations: Some testers were a little confused about the limitations of the prototype and were trying to ask questions about other skin concerns.
- Although the bot couldn’t answer those, it was still helpful to gain data regarding which concerns users were most likely to ask about - “acne” was a popular one.
- Add to cart feature: Testers felt that ‘Add to Cart’ would have felt like an intuitive feature to be added in the chatbot. 






Appendix


Figjam file 
Sample script 
Training data 
Voiceflow prototype