Assumption Map

Making the assumption map helped of team figure out what was actually viable versus what was a stretch. Here are the four main things we realized:

1. We realized our coolest feature, the auto-detecting reminders, is dependent on users being okay with sharing their personal schedules. If people feel like the app is being too intrusive  or they just don’t keep a digital calendar, the whole automation part of the app basically dies. This is why we put “Calendar Syncing” in the riskiest quadrant; it’s our biggest hurdle.
2. We love the idea of using humorous quotes to make the app feel like a friend, but there were other factors we had to consider. There’s a fine line between a supportive, funny ping and a reminder that’s just annoying when you’re already stressed about a midterm. We need to make sure the “vibes” actually help people eat rather than just making them want to mute their notifications.
3. A big insight was that a reminder alone doesn’t put food in someone’s hand. If a student is stuck in a 4-hour study session at Green Library and gets a “go eat” notification, they’re still going to skip the meal if they don’t have food on them. This is what pushed the Snack Box from a “maybe later” idea to a core MVP feature. We have to solve the access problem, not just the memory problem.
4. We initially thought any healthy snack would work, but the map made us realize that if the food is “mid,” nobody is going to pay for a subscription. For this to actually be a business, the snacks have to be something people actually look forward to eating, not just emergency fuel.

Designing Assumption Tests:

1. 1) Hypothesis:

   We believe that people will let us see into their calendars

   2) Test: To verify that we will ask them to lets us into their calendars.

   3) Metric: And measure whether or not we are in

   4) Criteria We are right if we can access their calendars.

2. 1) Hypothesis:

   We believe that Gen-Z style motivational quotes will increase users’ motivation to follow the food-reminder intervention.

   2) Test: To verify this, we will provide participants with a list of quotes, including ones we predict will be highly motivating (Gen-Z style) and ones we predict will be less motivating (control quotes). Before collecting data, our team will assign a predicted motivation rating (1–5) to each quote.

   3) Metric: And measure participants’ perceived motivation for each quote on a 1–5 scale and compare their average ratings to our predicted ratings using Mean Squared Error (MSE).

   4) Criteria: We are right if the MSE between our predicted ratings and participants’ average ratings is ≤ 0.25 (meaning we are within about half a point on average), and Gen-Z style quotes receive higher average motivation ratings than control quotes.

3. Step 1: Hypothesis
   1. We believe that busy students check their phones frequently enough during typical meal times that push notifications can effectively serve as eating reminders, even during high-pressure/busy periods and schedules.
   2. Step 2: Test
      1. To verify that, we will send time-sensitive text messages to participants at strategic meal times (breakfast, lunch, and dinner windows) over a 3-day period. These messages will contain a simple prompt requiring a response (e.g. reply with a “👍” when you see this!”). We’ll send messages at varied times within each meal window to account for different schedules (e.g. early breakfast (8:30am), mid-morning (10am), lunch (12:30pm), afternoon (3pm), and dinner (6:30pm and 8:00pm)). Participants will be informed that this is a phone engagement test but will not be told the specific 10 minute response window we’re measuring to avoid artificially inflating their phone-checking behavior.
   3. Step 3: Metric
      1. We will measure the percentage of messages that receive a response within 10 minutes of delivery, segmented by time of day and participant schedule context (e.g. if they’re in class, studying. etc.). We’ll also track the average response time and whether certain meal windows show systematically lower engagement. This would show us when alternative intervention strategies might be needed.
   4. Step 4: Criteria
      1. We’re right if at least 70% of messages receive responses within 10 minutes across all meal times. This threshold accounts for realistic cases where students may not have access to their phones in classes, meetings, or deep studying sessions where phone access is limited. If breakfast times show lower response rates (below 50%) this would validate our survey findings that breakfast is the most challenging meal to support and may require different intervention timing (e.g. the night before). If overall response rates fall below 60%, it would suggest that push notifications alone are insufficient and we should explore other complementary reminder channels (e.g. location-based triggers, smartwatch vibrations, etc.)