spam me
Design for Behavior Change
PROBLEM
Inadvertent phone scrolling on social media causes loss of focus, loss of productivity, and lost time.
​
SOLUTION
A tool that allows users to create custom Instagram bots that nudge users to stop scrolling.
​
I led a team of designers and software engineers to build this solution.
Try Me!
Baseline Study
A week diary study with 8 participants to understand people's scrolling behavior.
This study consisted of a pre and post-interview and a survey for participants to complete whenever they scrolled. We asked them to share usage metrics, how they felt afterward, and what led them to start/stop scrolling.
​
Analysis
We used a variety of techniques such as affinity mapping, journey maps, and connection circles to understand the causes and results of scrolling in our participants' lives.
​
Journey Map: Feelings using phone during transition times
​
​
Key takeaways
Connection circle: Causes of scrolling
We learned that people tend to scroll for either a mindful break, or out of habit, and they tend to either want to stop usage and replace it with something, or just be more intentional with their usage. Many participants experienced negative feelings after usage, and that time spent scrolling varied wildly (from 5 minutes to two hours!) if not timeboxed.
​
Key factors contributing to increased scrolling:
-
Scrolling as procrastination
-
Apprehension about upcoming activity or task
-
Empty transition time that isn't long enough for significant activity
​
Research Methods
Through a series of extended user studies, we iterated and refined our solution to reduce aimless scrolling
Intervention study
A 5-day study where participants' Instagram scroll was interrupted by posts that encouraged alternate behavior
​
​
We created an Instagram account @stopscrollingstanford to simulate what a bot to interrupt scroll might look like and had our 6 study participants follow the account. Over a 5-day period, we posted memes to the account (around 1-2 per hour) either urging the viewer to stop scrolling or suggesting other activities they could do instead and had participants track their social media usage.
​
Example Postings & Profile
Scrolling Data Recorded by Instagram
Although participants' scrolling time was slightly reduced during the study it was inconclusive if the intervention was the cause.
Based on the qualitative data gathered in free response survey and pre/post interviews we found that the bot was not as influential as we had hoped — partly because participants rarely saw the bot’s suggestions in the right time or place to act on them, and partly because participants often saw the posts during break times when they had no desire to stop scrolling. If we were to run the test again, we might try to make the content more specific to our followers’ interests and better aligned with their schedules.
​
Designing the service
Based on the results of the studies and user interviews we designed and prototyped our vision for SpamMe
Low fidelity concept mock-ups
We started with low-fidelity hand sketches and wireframes to align on the core functionality quickly.
Wireflow: Editing bot preferences directly in Instagram
Wireframes: Key pages of service
Brand Design
We collaborated to develop a visual brand to embody our concept using a branding exercise, mood board and style tile.
Brand Descriptors
Positioning - Casually cool
​
Promise - Make you laugh while helping you change your habits
​
Personas - Fun and hip
​
Product - tech for the average person
​
Personality - Casual funny
​
Pricing - Free but asks for donations to support project
​
Mood Board
Style Tile
Prototype Usability Testing
We performed usability tests with 4 users to improve the service interface. The two most critical issues and the corresponding corrective action are outlined below.
Issue 1: Product offering was not clear from the home page.
Fix 1: Include a “learn more” section below the signup button with a description of the product.
Issue 2: Not enough context for why in the wizard.
Fix 2: Include a tutorial pop-up window that details what the user will be expected to input in each step and how the information supports the bot.
