The iOS Weather was designed to answer the general daily forecast inquiries from all iPhone users, but sometimes one solution doesn't fit all. Our survey results on 108 iPhone users indicated that only 51% of them are using the stock application, which highlighted the need of a more advanced and intelligent solution to support the usage of broad user groups.
Time and Duration
I started with the assumption that people have different priorities for the information presented in a weather app.
To validate my assumption, I posted a survey on Amazon Mechanical Turk and collected 108 responses from U.S. based iPhone users. This survey aimed at collecting information in the following categories:
The most frequent scenarios for him/her to use the weather app.
Personal priorities for all information presented in the weather app.
Attitudes toward providing weather information in other stock applications.
Attitudes toward customizing information hierarchy in the weather app.
The results indicated that:
There are several common scenarios for participants to interact with their weather apps.
57% of participants check on weather reports after waking up in the morning, and nearly half of the participants want to know about the weather before outdoor activities. Specific information about the rain forecast becomes valuable when the rain starts, and the forecasts for the next several days can be handy when planning on trips.
Participants are comfortable with the idea of integrating weather information into other stock applications.
87% of participants favored the idea of presenting daily forecasts as part of the wake-up alarm. This number told me that, even for people who do not have the habit to check on the weather in the morning, this feature is still appealing to them.
A "one for all" layout design is not the only way to go.
An adaptable user interface that gradually learns from the user's behavior to improve its layout was preferred by more participants. One-quarter of the participants were willing to choose the information they value for the first time they open the app, but only 7% would like to adjust the UI multiple times.
Other than the insights above, participants were paying extra attention to forecast categories like "chance of rain," "extreme weather alert," "apparent temperature," and "drastic temperature change."
While quantitative data gave us insights about preferences and attitudes, qualitative data from user interviews highlighted the pain points. Ideal interviewees were iOS Weather users aged between 18 - 40 who were either frequent travelers or outdoor workers (include outdoor sports players). We expected them to have higher dependence on weather forecasts, also were more experienced with the app itself.
An affinity diagram of user interview insights on iOS Weather.
This affinity diagram supported our previous survey results, also addressed some additional issues including sense-making of the data provided by the weather app and the need to associate forecast information with certain scenarios.
Define the Users and Problems
The preliminary research validates my assumption that Weather app users have different priorities for various information categories presented. The advanced users, include outdoor workers and frequent travelers who's needs go beyond checking the current temperature and tomorrow's forecasts, are not satisfied with the current app design.
From here, I decided to choose these advanced users as our targeted user group, and then created a persona Caroline Smith, a business traveler and tennis player, to represent their needs and frustrations.
Persona: Caroline Smith
Caroline relies on weather forecasts to make life and work decisions, including business trip plannings, tennis trainings, and daily wears. She seeks information related to allergen, UV index, and hazardous weather conditions, which was not fulfilled by the current iOS weather, so Caroline is looking for an alternative solution that is more intelligent.
Weather forecast, beyond a weather app.
During my days at OnePlus, building an interconnection between the stock Andriod apps was already a hot topic. With the support with advanced AI chips, sensors, and all the information an individual can store in his/her smartphone, an intelligent operating system should be able to understand repetitive user scenarios, thus provides appropriate assistance with respect to users' privacy.
iOS 13 wake-up alarm design
However, when we think about the exact scenario...
It is 6:10 in the morning, I wake up by an alarm clock unpleasantly. Knowing that I should get up and prepare myself for a new day, I start to search my phone and mute this alarm with my blurry eyes. At this time, I probably have no intention to read a block of text to remember today's high and low, nor do I have the sense to interpret the wind speed and directions. I will have to open the Weather app when I'm brushing my teeth and go over everything again. The idea of combining iOS bedtime with an alarm clock is nice, but can we execute it better?
The iteration of the existing wake-up alarm.
This iteration only keeps the essential information that a user cares about and is relevant to him/her at the moment:
Current temperature and "feels like" temperature;
Forecast of the first rain/snow of the day (if exist);
Forecast of the rain/snow condition of the first outdoor activity in iOS Calendar.
With this new layout, I expect users to quickly notice a small amount of information in the short timing gap between finding their phones and dismissing the alarm. We knew from the interview that this information was sufficient to support decision-making including clothing and rain gears.
But, in an edge scenario people dismiss their alarms with eyes closed.
This case does exist according to our user studies. Under this situational inconvenience, Siri can be a good substitution to a traditional page view.
iOS Bedtime onboarding and settings, where users can choose "Siri Broadcast" instead of the lock-screen view.
While the iterated onboarding process introduces the "weather report" feature, users can now change the way of this report in the "options" menu. "Page View" is selected by default, and if a user prefers Siri to read the forecast, he/she can select "Siri Broadcast" and will only see a traditional alarm with no additional information.
Use Siri to broadcast the weather forecast after dismissing the alarm.
While 41% of participants check on weather reports before leaving their homes, the iOS Calendar and Reminders, with their time-sensitive and event-specific nature, are also the ideal carriers of weather information.
The reminders with event-specific weather forecasts.
For frequent travelers and outdoor workers, unexpected rain or snow can cause issues. Because the data source "The Weather Channel" provides hourly forecasts for the next 48 hours, it becomes possible to show event-specific weather and temperature data in the Reminders. In this case, no forecast will be shown for "Flight to Seattle" until 11:00 AM, June 29th.
The Weather app plays a dominating role in delivering forecast information, but it is natural that users can have inadequate abilities to process some of the information.
An example from the interviews where we discovered sense-making issues for certain categories such as wind, visibility, and UV index. When presented "Wind: w 7 mph" and "UV index: 3," participants found it hard to transfer them into actual life decisions associated with outdoor activities or sunscreens.
They expressed a need for more affordable expressions of the data, but at the same time, wanted to maintain an iOS styled, clean UI design.
Highlighting unusual weather conditions.
Highlighting unusual conditions: by comparing real-time forecast data to the historical metadata of the city/area, the algorithm addresses unusual weather conditions or the ones that are simply uncomfortable for people. Information includes:
Uncomfortably high/low humidity with the current temperature;
Uncomfortably high/low barometric pressure with the current altitude;
High wind based on the wind speed metadata, or 25% increase compared to the same time yesterday;
"Feels like" temperature that is 2°C/4°F higher or lower than the air temperature;
Visibility below 5 kilometers or 3.1 miles;
UV index above 5.
Highlighting these unusual conditions increase users' awareness at first glance, they can also tap on the tags to quickly navigate to the designated area for detailed data.
Integrated weather hazards warning.
Extreme weather condition is another frequently requested but missing feature in Weather. Considering its impact on users' everyday lives and personal safety, I presented the warning tag along with other tags with colors based on the warning type.
Color-coding the eight types of hazardous weather condition.
Eight types of hazards are covered in this design, and the colors are following the guidance of the National Weather Service and iOS 13 design guideline. In mainland China, the China Meteorological Administration uses blue, yellow, orange, and red to represent the severities of hazards. In this case, warnings no longer have their unique colors but adopt one of the four colors to represent the detailed severity of a warning.
Test and Iterate
After building the prototypes, I recruited three usability experts to conduct heuristic evaluations using the Gerhardt-Powals' cognitive engineering principles.
To find a set of guidelines that fits the design directions, I went through Jakob Nielsen's heuristics, Gerhardt-Powals' cognitive engineering principles, and Microsoft's guidelines for Human-AI interaction, and eventually chose Gerhardt Powals' principles which primary address the issues in how to present data in a way that users can easily interpret and make use of.
I also created a form that provided "Good," "Satisfied," "Unsatisfied," and "Bad" for each of the ten principles. During the evaluation, if the evaluator indicated our prototype as "Good" or "Bad" for a principle, he/she would be asked to briefly comment on the choice.
Evaluators pointed out several issues where the design did not match the guidance.
Noteworthy comments for Weather redesign:
Users may want to personalize and see other weather information beyond the promoted ones, but the current design only promotes the weather information that needs users' special attention.
Hazardous weather warnings are on the same visual hierarchy with other non-emergency information listed below. Users cannot quickly interpret them as alerts.
A comparison among three different front page design approaches.
Noteworthy comments for integrating Weather into Wake-up alarm:
The "Dismiss" button does not function as advertised: when the users click on it, they are expecting to see the home screen, instead, they see a briefing for the weather condition.
"Link to Calendar" in the settings has a confusing wording, the explanation doesn't seem to explain the option well. An alternative way to address it is "show weather for my first activity in the morning."
It seems that the objective is only to show the first activity in the calendar and the associated weather forecast, which is too little information for an entire briefing page. Also, what if it is sunny over the first activity?
The feature is good, but since it is part of the "Bedtime" function, only a few people could notice it, let alone understanding its functionality. More promotions of the feature are necessary.
A comparison between two wake-up alarm settings design approaches.
Noteworthy comments for integrating Weather into To-do list:
Users can be confused about the format difference where some of the reports are hourly temperature/weather and the rest are forecasts of the day especially when the icons are alike.
The question from Siri is confusing where Siri asks whether the users want to be notified about the weather but doesn't provide the information as an answer. For the users, they may understand that Siri is about to tell them the information rather than adding an icon to the card.
Within the Calendar, if users turn on weather information switch, it only displays the weather report for the whole day rather than event-specific information. But in the To-do list, a forecast of the exact hour of the event is provided, which appears to be inconsistent.
Giving the option to be reminded of the weather is not very necessary, it is going to make the users more confused about additional functions/items in the app.
A comparison between two weather-in-schedule design approaches.
This case study is a great opportunity for me to think about "the design of everyday things." People have on average 80 or more applications on their smartphones, and their lives highly depend on the information people receive from these applications. As designers, can we think outside individual applications and let them collaborate with each other? Or even more, create products outside the scope of a single hardware, harness the power of surrounding sensors to promote a comprehensive experience, and eventually arrive at the idea of "ubiquitous computing." And at the same time, maintain respects to users' privacy.