Starbutter - Star Horoscope

Developing a virtual astrologist for millennials

Screen Shot 2019-01-25 at 2.52.11 AM.png


During Spring 2018, I interned at Starbutter is a AI Chatbot startup in Berkeley. I worked with the CEO Arun Rao and Head of Product Samaneh Pourjalali to redesign and manage their recently launched horoscope bot “StarHoroscope”

Through out the semester I began with user research / competitive analysis, then moved to ideation and lo fi prototyping, to hi fidelity prototyping, and began collaborating with engineers to ship new features to the live bot.

The Problem

At the time I joined the project, StarHorscope had just been launched and it’s main functions were presenting a user a reading that was scraped from an astrology site. The dialogue between the astrologist Zora was limited.

Arun and Samaneh brought me on the team to create a more cohesive user flow and dialogue, and design new features to attract a millennial audience.

The ultimate goal of the redesign was to increase user activation and acquisition before larger companies created similar bots.

initial chatbot introduction

initial chatbot introduction


Areas of Opportunity

I love astrology and horoscopes, so the product itself was right up my alley. When first playing with the bot I saw two main areas that StarHoroscope could differentiate itself.

First was the use of birth charts, which are charts that report the position of the celestial bodies at the time of your birth.

Secondly was the personality of Zora being a chatbot our product could go the extra step and create a more personable relationship with the user increasing both retention and engagement.

Competitive Analysis

I began by research how people are currently sharing, consuming, and interacting with horoscopes and astrology. I wanted to know what competitors were doing well and what they were missing. I also wanted to make sure that users expectations were met in terms of the type and detail of content on the bot. I found that the two main mediums for horoscopes were apps and social media.

Astrology on Apps

Looking through the app Costar and Horoscope daily were the most popular and represented different user bases as well.

Screen Shot 2019-01-25 at 3.15.03 AM.png


  • Popular among serious astrology enthusiasts

  • Provides natal chart information & daily reading

  • Allows users to connect with friends and compare compatibility on natal charts

  • Builds legitimacy by using NASA data on planetary positions      

Horoscope Daily App

  • Most popular horoscope app on IOS

  • Simple app architecture

  • User doesn't build account but selects their sign to receive reading and compatibility

  • Does not integrate natal chart information

Astrology on Social Media

Twitter and Facebook represented the largest astrology communities. Twitter, found most engagement through memes related to people’s signs. While on Facebook there were several thousand groups relating to astrological sign and advice.

User Research

Interview: I interviewed 6 people for 30 minutes to gain perspective on millennial’s attitude towards horoscopes as well as

Contextual Inquiry: I asked 10 people to play with the bot and walk though what they were doing and verbalize any frustrations they had with the bot.


After both user and competitive landscape research I synthesized the following insights regarding user behavior:

  • People crave validation through astrology

  • Every sign has bad qualities, but highlighting good qualities results in higher engagement and loyalty

  • Readings that are more motivational rather than informative are more popular

  • Memes have become the main medium for astrology to be consumed on social media apps

  • Memes making tagging friends or sharing with friends easy

Pain Points


the readings didn't seem personalized

Lack of share-ability -

other than taking a screenshot your reading there was no way to share content

Lack of memory-

The bot would ask for a user's sign every time a feature was initiated


Synthesizing interviews, survey and pain points I created 2 very distinct user personas. I used them to inform my ideation and prototypes. 

Persona 1- Horoscope expert

Persona 1- Horoscope expert

Persona 2 - Horoscope amateur

Persona 2 - Horoscope amateur



I took to the drawing board and came up with some ideas to not only make StarHoroscope comparable to other astrology apps, but allow it to its medium as chatbot make it different

New Features:

  • Profile memory

  • New on-boarding flow

  • Compatibility

  • Personalization

  • Words of affirmation

  • Celebrity matching

  • Astro meme

  • General readings

  • Natal chart

Zora’s Personality:

  • Millennial,

  • Mysterious Elder

  • Maternal Kooky Aunt

I sent out a survey in order to validate which of my ideas caught potential users attention- 26 people aged 18-27 filled out the survey. The Maternal Kooky aunt personality and compatibility were the most wanted features.

Screen Shot 2018-08-07 at 12.08.04 AM.png

Initial Sketches

Below are some sketches of what an onboarding flow would look like and how new features would be presented to users.

Mid-fi Prototype

taking the information from both insights and ideation, I used botsociety to mock out what an updated StarHoroscope would look like.

Prototype Features

a flow chart that describes the new on boarding flow for users

a flow chart that describes the new on boarding flow for users

Screen Shot 2019-01-25 at 4.02.39 AM.png

Onboarding screens - due to the lack of memory previously I wanted to address this pain point by creating a flow new users go through that sends their information to our server to be stored.

Asks for Birthday, location of birth, time of birth, and name.


New features -

  • Personalized reading - open conversation with Zora - in response to the pain point of vagueness

  • Compatibility- provides the user with a breakdown - as this was the most request feature

  • Daily Reading- daily reading based on their sun sign - core functionality of app

  • Motivational Reading- Text from Zodiac mind is served on a picture - based on research insight

User Testing

Using the prototype I sent the bot to 10 users and collected their feedback to iterate on and improve in the next version.


  • The chatbot is friendly- the use of "my dear was endearing"

  • The emojis!!

  • The idea of getting a personalized reading was very exciting for users

  • The avatar was inviting  

  • The menu option made navigation easy as features were more centralized

  • Compatibility was broken up into sections.


  • The motivational readings are pretty general

  • The on-boarding can be confusing

  • Clarity on the personalized reading portion

  • Some text boxes seemed long

  • Some of the Buttons were hard to read in the main menu.

Hi-fi Prototype

The layout of the bot remained the same, but I cleaned up the onboarding screens to be more intuitive. I refined the screens a bit more with better graphics.

Added user insight of memes into a new section.

Hi-fi screens

Hi-fi screens

Meme feature

Meme feature

Shipped Version

The live version of the chat bot is HERE.

Many of the suggestions have already been put into place though interns over this summer are working on rolling out the personalized reading feature which serves readings or advice based on previous interactions and information a user has provided the bot.

Shipped version of the bot

Shipped version of the bot


Take aways

This was the first time I singularly led a design revamp, so I was able to learn quickly how to stay organize break down sprints into scrums and keep stakeholders updated on a consistent basis

This was also the first time that I was able to see my design recommendations be implemented so quickly, after speaking with engineers and getting approval from the CEO my prototypes became a reality!