The primary objective of voice-first technology is to connect a user with specific data, information or services. Progressive disclosure — gradually revealing more information at just the right time — is an effective approach for guiding the user on the voice journey, providing information at just the right time.
This sequential interaction helps users easily manage potentially complex queries and processes.
Progressive disclosure also informs users about which capabilities a voice application holds, exactly when users are most in need of that information. This mechanism determines when a user is receptive toward guidance through careful tracking of voice application usage.
Too Much Guidance?
The irony of voice applications is that they are there to aid people in accomplishing their goals quickly, but the apps have to perform this job without providing users with too much guidance to avoid overwhelming them.
I often think of using voice apps as similar to opening an unknown box of goods that you forgot you ordered. You know there will be something useful in there, but you don’t know what, and you have to dig through the layers before you find out.
One solution to the “mystery box” problem is to tell users each time exactly what they can do, mimicking a menu on a website or mobile app. While this is an easy solution, it is truly terrible in practice. Unless you want to also add: “Listen closely, as our menu options may have changed.” This method threatens to overload the user.
New users especially need to see value immediately, and don’t have time to listen to the equivalent of a user manual. (How often do you browse through a user manual?)
Progressive Disclosure: Teaching the User How to Communicate with the Voice App
The first place most people add progressive disclosure is in situations where a user is running into trouble. Your goal is to guide the user with an increasing level of specificity.
“I don’t quite understand. Can you ask me again?”
“I’m still having a difficult time. Why don’t you try asking me for the name of a movie?”
“Hey, I’m sorry, I’m not sure how to take care of that. Try asking me this: ‘What time is Toy Story 4 playing near me?’”
Notice when responses progressively become more specific, and are progressively teaching users what the app expects of them. The increased specificity is useful for multiple reasons. First, the app often will not know why there was a misunderstanding. Responding with “Sorry, I can’t do that” when the user didn’t say anything is a mistake.
Secondly, if the goal of voice apps is to help people achieve their goals more efficiently, responding with the same message every single time is counterproductive.
Because voice apps aren’t yet smart enough to learn like humans, you need to help the humans learn how the voice app communicates. (Yes, you’ll do everything you can to make the app understand any incoming requests, but you will invariably miss something.)
By helping the humans better communicate with the voice app -- in the voice app’s own language -- users get what they want without getting frustrated and moving on.Encourage Discovery
Progressive disclosure is also useful for encouraging discovery. Because the voice app doesn’t ship with a manual, users won’t always know what to do, and you don’t want to tell them before they see value in using the app. Instead, provide guidance as users interact with the app.
Again, you’ll want to keep track of what users are doing, although in contrast to dealing with errors, you’ll instead track what users do across sessions. Keep track of which intents they are hitting and what entities they’re asking for.
Once the user becomes a “regular” (often in the first three to five sessions), you can begin to provide some tips. Look at your running list of requests for the user, and see which actions the user hasn’t seen. Then have your app choose one that is related to a user’s new interaction. “Toy Story 4 is next playing at 8:30 p.m. By the way, you can also ask to buy tickets. Want to try it?”
By providing this guidance after giving users what they are asking for, you’re able to tie together related actions. Also, you already have a strong idea that the added information is relevant to the user.
Waiting until after users have returned signals that they already value what you offer suggests that they will be receptive to engaging more with it.
It’s About Context
Note that you should not simply track how many times a user entered any error state. You are probably already providing contextual assistance when people ask for help, so it’s not different when they need help unexpectedly.
Keep track of how many times in a row people run into trouble, and in which context. An error in one context doesn’t necessarily mean the user is having the same problem in another context. Use that information to be sure that any response is specific to where the user is in that moment.
For example, if a user has encountered a problem on the “transfer money intent” twice, don’t take that to mean that the first error on the “hear balance intent” indicates the user is hopelessly lost. (You can also use this data later to improve the app.)
Tracking information about users sounds scary, and can be scary if that data is used for nefarious means. But still, learning about people is how we develop relationships and how we become useful to others. It’s how we know to bring excess peaches from our backyard tree to Karen, who loves the fruit. And it’s how we know how to provide useful interactions to the people using our voice apps.
Following these progressive disclosure guidelines will help you deliver a meaningful customer experience, which is ultimately what you and consumers want.