I’m the father of an adorable baby girl. This is my wife’s and my first kid and as normal we are overwhelmed A LOT by the things we need to do keep her happy and healthy (healthy is the really difficult part.)
She is almost 2 now, but I remember a time when we had to keep track of how often she pooped, the color of it, how much she drank, whether it was formula or breast milk, how often she slept and for how long. It is a lot to remember and we didn’t have a log book or app so we ended up just telling each other and generally remembering it.
I Should Design an App for That
That got me thinking about a voice app for Alexa or Google Home where you don’t need to remember to pull your phone out and log in a little calendar. Why not do what we did, talk, but to one of these assistants? Conversation is much more natural in this instance and only makes sense that you can be able to whisper to Google at 2 am that your baby just slept for 3 hours and drank 6 oz of formula.
The more I thought about using these devices for tracking baby schedules the more I thought about the times we could have used a quick reference book on common problems with a newborn. Instead of calling someone, why not just ask Alexa? Why not have the app be able to look up at home solutions to common issues that come up with newborns and stressed out parents?
With these main use cases in mind, I started looking for new parents who thought this might be a good, or a terrible, idea. I started a poll on Facebook and talked to some of my friends who had just had a baby.
I found that most of the parents I interviewed wanted to know if the information provided in the app was backed by anything more than just a random internet search. I needed to find a way to reassure the users of my app that the information came from a trusted source. This was my first big challenge of this experience.
Reassure the Users
To do this I decided to cover the on-boarding experience of the app. Here, I can present the user with information regarding where this app gets its data, and what it uses with the info the user gives it.
A lot of this reassurance could happen in the opening statement of the system persona. I initially wanted it to be personable like talking to a comforting grandma
Hi I’m Deborah! Let’s get to know each other. What is your name?
Then after some testing found that to be too forward and it didn’t give the user any info on where the info was sourced.
I moved on to version 2 of the intro which gave more information that the system was backed by, but was a little too uncanny valley for my testing users.
Hi I’m Deborah! I have helped thousands of new parents with their kids and am backed with 20 years of experience in Pediatrics.
Finally, I realized the system persona should represent a hospital here in the Utah Valley area. That way it can leverage the experience of the hospital and retain its one-on-one style.
Hi I’m Deborah with Intermountain Healthcare! We have helped thousands of new parents with their kids each year and are backed with decades of experience. Let’s get to know each other. What is your name?
It’s a bit longer but provides all the essential information and ends with a simple narrow-focused question.
With the intro out of the way I needed to get info from the user about themselves and their newborn, but I didn’t want it to feel like an interrogation. I added lots of positive affirmations and reasoning behind the data I was asking for.
How cute! Do you want to tell me her weight and height? I can use these measurements to better recommend healthy options for your little girl.
I also wanted to give the user an out if they felt uncomfortable giving a metal cylinder in their home lots of personal information.
What if the User Falls Asleep?
I remember countless late nights with a baby that just wanted to eat and I kept falling asleep while feeding her. I had to think if the user falls asleep while talking to the system how should it handle this?
First off, I didn’t want the user to miss the information that they requested because they nodded off briefly, but I also didn’t want to annoy a sleep-deprived parent. Balancing this interaction was more interesting to design than I expected.
After the user heard what he needed to hear, how much and when to feed his daughter, he walks away to go and warm it up. This leads to our first No Input error. The system persona, trying to be helpful, restates the previous prompt in a clearer manner.
Olivia should drink 5 oz of formula or breast milk
Then after the second input prompt the system assumes that the user has walked away for fallen back asleep and closes the app with the assurance that she is still here to help.
If you need me again just ask Alexa to open Intermountain Healthcare and I’ll be there!
When I was designing this I didn’t know how often the system should re-prompt the user with the same info, but after testing it on a new parent they emphatically said that more than one re-prompt, on a sleep-deprived brain, they would have thrown the speaker away. So to keep the speaker out of the garbage the system only re-prompts once then kindly closes itself.
Choose the Right Words
My third hurdle in this design process was choosing the right words to say at the right time. I didn’t want to be cheery and flamboyant when talking about sickness or casual when recommending formula amounts.
When I first designed the feeding script it looked like this:
New parents don’t want estimates.
At least can mean anything from 5 to 70 oz of milk and when your baby is only 8 pounds every ounce needs to be counted. I removed the “at least” from the script and was pleased with how it sounded. With this little change, the system sounded much more like the kind but professional nurse of a system persona I was going for.
Then, when the system has a conversation about a sick baby I didn’t want it to say “Good night” or “I hope she gets better” in closing. The night is not going to be good if you have a sick 2-month-old laying right next to you, and “I hope she gets better” sounded creepy coming from a speaker no matter how kind the system persona was. I finally settled on “good-bye.” It is formal without being distancing and with a positive intonation it can still be a friendly phrase.
Improvements to My Design Process
This has been a challenging and rewarding design process. I enjoyed talking to people about their parenting habits and seeing how accepted virtual assistances are becoming.
I think my largest take away from this experience is that even though I’m a potential user for this app I still can’t think of all the problems or ways to use it. Talking with others helped me design something that could be much more widely accepted, rather than something that myself and a few of my friends would use.
Lastly, I have really enjoyed conversation design and have written about my first experience in this field here. I plan on writing at least two more stories about my voice design journey so get excited!