VUXcellence: CBeebies (and Indiana Jones?)

In the last VUXcellence post, we talked about how to track whether or not your user is having a bad experience, and a way to know when to slip them some extra help. Our context there was around the user's aggregate experience and the frustration they've been building up over the course of many interactions. One thing we didn't talk about, though, is the sort of techniques available us to turn around an individual error case or misfired intent. There's one skill in particular that has taken what I think is a wonderful approach to solving this problem...

---

---

I recently had the pleasure of hearing Lisa Vigar and Paul Jackson talk at the Dashbot Superbot conference about some of the design considerations of the BBC's skill for kids, known as the CBeebies skill. They have the additional design challenge of designing for kids, which ups the complexity considerably. They produced a short video talking about some of the "hardmode" problems they were facing:

There's one thing in particular that I loved about this approach - they knew they were going to have kids going off the rails and taking circuitous or nonsensical paths through their skill, so they embedded fun, custom content inside their error flow. For my part, this is something I've always tried to do with CompliBot and InsultiBot - no matter what path you've taken through the skill, you should always be getting something that we consider to be "content", in the persona of the bot you're talking to, but CBeebies takes the idea of "error path as an opportunity" and knocks it out of the park.

And really, it's something that I wish we'd see more of. Obviously you'd prefer your users never end up in your error flow to begin with, but it's definitely going to happen, so you may as well take the opportunity to try to surprise and delight them. Dogs of Amazon is a great example of trying to turn a bad experience (404) into a good one (puppers). This idea of providing good, interesting error handling is one that came up several times during the Alexa Live event, and in particular I'd recommend checking out two minutes or so of the following clip with John Gillilan if you have a moment:

The one design example I always come back to, though, is from a trip to Disney Land I took many years ago. We had waited in line for the Indiana Jones ride, and about a minute in, the entire ride stopped, and the lights came on, with us sitting in the middle of a nondescript room between scenes. Obviously not the way the ride is supposed to go, but the designers of that ride had prepared for the eventuality, and shortly thereafter we heard noise coming from the speakers of our "cars". At first, it sounded like the traditional "engine won't start" screech of an old car, but then out of nowhere came the voice of John Rhys-Davies, in character as Sallah, giving us a storyline-adjacent reason that we were sitting and waiting. The vast majority of riders would never encounter this failure mode, but still the designers had taken the time to have a famous (and presumably highly-paid) voice actor record extra audio just in case.

Just like that the ride started up again and ran to completion, and rather than coming off frustrated that I had waited in line for a fractured experience, I actually came out of it happier than I'd ever been when I'd ridden it before, having experienced this unique plot branch that seemed specially tuned to just those in my car. This is what CBeebies's "lets have a surprise" feature felt like to me, all these years later.

---

---

So the question to add to your design checklist, then, is what can you do to take a user's accidents, miscues, and fumbling (which they definitely are going to have - VUX is nascent and users say weird things all the time), and turn that into a positive experience? Or to put it more succinctly, how can you "turn that frown upside down"?

CBeebies has raised the bar, and I'd really love to hear what else other folks are doing along these lines.