Getting the IoT into Tesco: Internet of things UX for the mass market - IoT 14
1. Getting the IoT into Tesco
or: IoT user experience for the mass market
Tesco Toton by Roger
Claire Rowland/@clurr
BLN Internet of Things 2014
Wednesday, 16 April 14
2. Hello :)
- Independent UX and product
consultant
- O’Reilly author: “UX design for the
consumer internet of things” (due end
2014)
Previously :
- Service design manager for AlertMe
- Head of research at Fjord
- Smarcos: EU consortium researching
interusability of interconnected
embedded devices
Wednesday, 16 April 14
3. photo by steven de polophoto by lyzadanger
photo by nickpophoto by david ward
Consumers are a very challenging
audience for IoT
Wednesday, 16 April 14
Before I worked in IoT, a lot of my design and research experience was with consumers, and I got interested in how new
technology becomes mass market. In my IoT work, I deliberately focused on research and testing with people who were not
early adopters, and this talk is informed by what I learned from them.
One thing I definitely learned was that the kind of mundane day to day activities to which many consumer products relate are
some of the most socially complex and interesting things to design for, therefore it’s really easy to get things wrong.
In many respects, consumers are the most difficult audience for a complex new technology. Upshot: The IoT is still tech led
and still not quite ready for consumer primetime. There are still many technical challenges but there is also a lot we don’t know
about design.
4. Mass market
products
should
- Solve a real problem
people have (value)
- Offer a good solution
(desirable, usable)
- Come at a cost
(financial, effort) that
feels in proportion to the
value
Wednesday, 16 April 14
What makes compelling mass market product?
Many examples in home automation going back up to 40 years of systems that never managed to convince people en masse
that what they did was worth the money and effort to install.
This is x10 powerhouse for the commodore 64 from 1986, allowed you to control appliances, lights, heating from any phone.
this system is nearly 30 years old and we’re not all using it, but those are exactly the same kind of things about which people
are still getting excited on Kickstarter :)
5. UX for IoT is not
just UI and
industrial design
UI/visual design
screen layout, look and feel
Platform design
discovery, control and coordination for
interconnected devices and services
Service design
customer lifecycle, customer services,
integration with non digital touchpoints
Productisation
audience, proposition, objectives,
functionality of a specific service
Industrial design
physical hardware: capabilities
and form factor
Interaction design
architecture and behaviours per
service, per device
Interusability
interactions spanning multiple
devices with different capabilities
Conceptual model
How should users think about the
system?
Many different layers of
design shape the end user
experience
Wednesday, 16 April 14
Platform design is concerned with discovery, control and coordination experiences for interconnected devices and services. A
complicated one that’s a whole talk in itself... think of it as the UX angle on interoperability.
6. UX for IoT is different...
We don’t (yet) expect Things to behave
like the Internet
The average consumer is going to find
it very strange when objects take time
to respond, or lose instructions.
Wednesday, 16 April 14
In addition to the complexity of the design thinking required, also some inherent challenges putting the internet into things.
We accept latency on the internet. It’s out of our control. We also accept a little bit of unreliability. We know that web pages
may be slow to respond, and that Skype calls fail. We don’t like it, but we accept it.
It’s not normal for this to happen in the physical world. We expect physical things to respond immediately and reliably.
But when you put internet latency/reliability into physical objects, that might not happen. Over the internet, your lights might
take 2 minutes to come on. We’re going to have to be careful how we design this and what promises we make, or it will just
feel broken.
7. 3 key guidelines for
successful consumer IoT
products:
-Solve a tangible problem
-Keep the conceptual model simple
-Make distributed interactions feel
coherent
Wednesday, 16 April 14
There are lots of things we could talk about, like setup experiences, service experience, industrial design, interface design for
embedded devices, platforms.
For the sake of time I’ve picked 3 fundamental guidelines that provide a framework for much of the UX:
8. Solve a tangible problem
What does it do? Why would I want it?
Wednesday, 16 April 14
9. In areas where they don’t have expert knowledge or are short on time
consumers need products, not tools
Product Tool
Wednesday, 16 April 14
The majority market is used to buying products that make a promise to solve a specific problem and come reasonably well
configured to solve it.
10. Nest do productisation really well
Wednesday, 16 April 14
Leaving aside recent problems with user interaction, the Nest protect advertising is very interesting: it doesn’t talk about
connectivity at all, it’s all about how it is a better smoke/CO alarm.
11. Belkin’s mobile app is good, but
a connected socket is a tool that requires
users to solve their own problems
Wednesday, 16 April 14
Belkin WeMo has a pretty good UX, but is essentially an example of a tool: something that requires end users to define and
solve own problems.
It requires an imaginative leap even to think about what you might do with it. Tools can be powerful things, and empowering
more people to use them is a great aim. But they require a lot of attention invested and I’d argue most majority consumers
won’t have space for more than a couple of these things in their lives.
12. This, though, is a great idea
Wednesday, 16 April 14
WeMo Crockpot slow cooker, out this spring
Phone app allows you to adjust temperature, change timings, turn it on and off remotely.
There’s a clear benefit, and the remote control aspect fits perfectly with context of use of a slow cooker: left running in house
whilst you’re out.
14. Conceptual models should be simple
Wednesday, 16 April 14
A conceptual model is the user’s understanding of how something works and what it can do.
15. Connectedness requires users to
think about system models
Wednesday, 16 April 14
Connectedness adds in additional complexity. There are extra bits. Some of the infrastructure devices are not very transparent
in function: what do they DO? A hub is a mysterious box to many people.
There are more things that can lose power or connectivity and more places where code can run. All of those have implications
as to how the system works.
In the case of a lighting system with automated rules that turn lights on and off at different times, in order to predict whether it
will still run if your internet connection goes down, you essentially have to know where the code governing those rules runs. If
it’s on your phone or in the cloud, then it won’t; if it’s in a hub it will.
16. Beware the
surprise package
Taking a successful mass
market product and making it
back into an early adopter
product
Scott Jenson, ‘The Simplicity Shift’
Wednesday, 16 April 14
17. You can try to
explain the system
model...
BERG Cloud bridge: transparent network comms
Wednesday, 16 April 14
18. ...but users
shouldn’t have to
understand
exactly how a
complex system
works in order to
be able to use it
successfully
Wednesday, 16 April 14
You have to explicitly design conceptual models, it’s not just about getting people to understand the way that you have built the
system
Learn what they need and design to fit the way people currently think: existing knowledge, behaviours and beliefs
19. Coherence across
distributed interactions
Interusability: distributed UX
Cross-Platform Service User Experience: A Field Study and an Initial
Framework. Minna Wäljas, Katarina Segerståhl, Kaisa Väänänen-Vainio-
Mattila, Harri Oinas-Kukkonen MobileHCI'10
Wednesday, 16 April 14
20. vs
Composition
Distribute functionality to suit the context of use
Wednesday, 16 April 14
Tado thermostat has no UI, it’s all on the phone. Keeps bill of materials down as UI components are expensive. Making a
good phone UI is relatively cheap. It’s an elegant choice but has limitations: if you don’t have your phone to hand, or it’s not
working, or you’re a guest in the house without access to the phone UI, you can’t adjust the heating.
This is the old AlertMe I worked on: a standard thermostat with phone and web apps, which are easier to use (the one you see
here looks rather plain as it’s an unbranded version). This means that you, and your guests or other residents without
smartphones, can still use it as a conventional thermostat. Both approaches have value in different contexts.
21. Consistency
Create device-appropriate interfaces that feel like a family
Wednesday, 16 April 14
Nest example: rotating bezel on wall thermostat makes clicking noise as you turn it. The touchscreen interface uses up and
down arrows to control temperature (rotational controls are inefficient and inaccurate on touchscreens), but still makes the
same click.
23. 19
2 min delay
21
...ensure data is up to date on all platforms if possible
Constrained devices often suffer
discontinuities
iphone by daniel, cloud by edward boatman, router by joe harrison, thermometer by ashley reinke, radiator and boiler by axeny virtinsky
Wednesday, 16 April 14
A lot of the knottiest design problems I’ve run into in my work are continuity challenges. It sounds obvious: If i interact with
the service on one device, I would expect all other devices reflect that change in state. e.g. if I turn the target heating
temperature up on my wall thermostat, you’d expect the new temperature to be immediately reflected on the smartphone
too.
But sometimes this isn’t technically possible.
Many heating controllers in the UK are battery powered, and sending data to the network uses a lot of power, so they only
connect intermittently. It’s possible to have a delay of up to two minutes before which a command sent from a smartphone
is registered by a heating controller, during which time the two devices appear to be giving different information about the
status of the system.
This breaks fundamental usability principles.
24. Good consumer UX for IoT
is deceptively hard
A final thought
Wednesday, 16 April 14
25. Tesler’s law of the conservation of complexity:
As you make the user
interaction simpler you
make things more
complicated for the
designer or engineer
Larry Tesler, former VP of Apple
Wednesday, 16 April 14