Information Gathering

Design task

The following project was a short design task assigned as part of an interview process. It was completed over the course of one day.

THE BRIEF

"The AI Supplement Company" aka "tasc", is a (fictional) company that uses Artificial Intelligence (AI) to create custom supplement formulations based on things like your lifestyle, health issues and basic information such as age and gender. To create custom supplements for their users, they need to gather that information in an easy way.

In summary: Marketing site → gather their information → result → checkout

TASC has already built the marketing site, and they currently sell supplements without customisation (so a purchase flow exists). They now want to design and build the "gathering information" part of the flow.

They want to design, build and launch this in 8 weeks.

OTHER KEY INFORMATION:

They need to gather about ~40 different answers/inputs from a user in the "gathering information part" for the AI to be effective - so it makes sense to split these up by section.

The key pieces of information can be split into sections, which are "Basic information" (age, gender), "Lifestyle" (exercise, diet), "Health" (any medical conditions or medication they're on).

In the lifestyle section, we need to gather 4 key pieces of information:
1. How much they exercise per week
2. How they rate their diet, e.g. from "very unhealthy" to "very healthy"
3. How much alcohol they drink per week
4. Whether they currently smoke
- If yes, how much, e.g. from "once a week" to "several packs per day".
- If no, did they used to smoke?
- If yes, how much

Competitor analysis

I had a quick look at other health and vitamin companies using AI flows to gather information about customers. How were they asking people questions and presenting answers? What are they doing well, what could be done better

Accessibility and best practises for our user group

Indicator of percentage completion and section indicator
Given the number of questions giving users an idea of how far they are through the survey is an important consideration. At the same time, not having this detract from the question being asked is important.
I also wanted to include an indicator of what section of the form users were on, both using text, and indicated with design, A possible suggestion would be using colour to differentiate the 3 sections.

Selectors, slider and input buttons
Given the user group, I wanted to keep the selectors as uniform as possible all the way through the intake.

In structuring questions, I considered the maximum number of selectors per questions and tried to split answers to minimise options, but maximise information.

For questions about diet, alcohol and exercise, whilst sliders were used very effectively by Thriva (pictured below), I felt that these wouldn’t be easy for all users on mobile. I also wanted to avoid using tooltips as they add another step, and given that Alva already seems to avoid using them. (My concern was also that tooltips on mobile would not work well for the upper age-range the demographic)

Save an finish later
Initially I considered if we would want a “save and complete later button” given the length of the questionnaire. However, as it is essentially an intake form before purchase I didn’t feel this would make sense for now - and could be assessed based on the time it took to a user to complete the flow and with user testing.

Structuring questions

Given the number of questions being asked in this survey, I understood the importance of keeping the questionnaire as short as possible but needing to consider the granularity of answers.

The diet section, for example, could go into much greater depth, and the alcohol section could be shorter, but in doing so may get less accurate answers

Diet: Discrete vs indiscrete selector for diet

Initially I felt the diet question would work well with a slider with variables (very unhealthy, neither healthy or unhealthy, healthy), allowing for people to correct for bad eating / self-judgement, given fluid nature of eating habits.

The pros are that a slider gives a user more control over their answer, but anchoring the slider causes bias when the user answers. I felt the benefits of using the same format of answer selection all the way through, coupled with the benefit of a clearer discreet answer outweighed the slider in this case.

Consider subjectivity: Diets are very subjective, and food intakes play a big part in vitamin intake – so this section would need to be fleshed out for more detail to get a clearer picture of user’s diets to more accurately suggest vitamins.

Alcohol: Putting sensitive questions deeper into the survey-flow

I split alcohol consumption question out to make it simpler - removing the need to explain alcohol units and used conversational language and flow to remove any sense of judgement that a user might experience.

Even though units are commonly used in online surveys and forms, I chose not to use them as people are not as familiar with them as we would like. Having to work out their own units adds cognitive load (And I had made the decision to not use tooltips from a design-perspective).

Consider under-reporting: It’s widely known that people underreport alcohol consumption, so if this is important for the flow it may be worth asking other questions to get around this, or to just assume underreporting and a flat % across the board.

Exercise: Including physical activity
For this age group in particular I’d want to know if things like physical activity, vs exercise only, could be included. Given more time I’d want to spend more time researching this and dedicating time to the language and inputs around these questions.

Smoking: Removing qualitative language
Originally had categorisation of smokers as occasional / light / moderate / heavy with the numbers. I removed this to improve grammar, and as smoking is seen as a vice qualitative descriptors may add judgement / alter answers even with quantitative qualifiers.