I am a user experience designer with experience designing, developing and leading the creation of software products. I am seeking an opportunity to work with a team that is passionate about building great technology products.

What are conversational interfaces good for?

(Original post on Medium)

Many businesses today are looking to build human or bot chat interfaces as new channels to engage with their customers. A key question to ask here is what are chat interfaces really good for? I spent some thinking about this question. In this post, I want to talk about 3 areas where I think conversational interfaces can really be truly better than existing solutions.

Chat as a way to help users choose

I recently had a great experience chatting with an agent at the Adidas web store through their embedded Intercom messenger. I wanted a specific kind of shoe which is usually a bit hard to find so I thought I might ask an agent for some help. I was looking for a pair of casual shoes that I could wear to work with jeans and tees. Not as casual as sneakers but still comfortable enough to wear all day at and after work. I gave a rough description of this to the agent and within a couple of minutes he gave me 3 great suggestions, one of which I actually wanted to buy.

A few years ago, at school we did a project for Facets multimedia — a non profit movie rental store in Chicago that deals primarily in independent and foreign films. Almost every customer we spoke to, spoke about getting great recommendations from the folks at the store and about relying on the board of recommendations that the staff put together each week to find the next movie to watch. I contrast this with the number of hours of my time I have spent looking for something to watch on Netflix despite the millions they have spent on their recommendations engine.

Humans are really good at understanding rough descriptions, connecting the dots and making great recommendations. Algorithms, on the other hand are terrible at this. These can be especially useful in fields that require knowledge and judgement like buying wine online, buying clothes that go well together or picking a movie to watch etc.

Chat as a way to help users get stuff done faster

I tried the Shopspring bot on Facebook messenger recently and there is one thing I came away feeling quite sure of — any scenario which requires or benefits from the user being able to browse through various product options is a terrible use case for a stand alone conversational UI. The shop spring experience on the messenger feels very restrictive and I feel like I haven’t seen enough choices to make a purchase decision.

In the demo of Viv at Tech Crunch disrupt, Dag Kittlaus showed some scenarios that are great candidates for a conversational interface. Sending money to a friend, booking a hotel that you have previously stayed at, picking flowers to send someone. He goes on to show how quick and effortless these workflows are through Viv.

One of the key components here is that the platform, Viv in this case, stores your payment information, addresses, contacts etc. Every purchase then could be as simple as finding the product, clicking ‘Buy’.

Scenarios where the user knows exactly what he or she wants or has very few choices to make before making a purchase decision can be executed an order of magnitude better through conversational UIs than the experience any app or website delivers today.

Chat as a way to make the user experience more personal

I recently reached out to United Airlines on messenger and asked them about my frequent flier number. Since I was on Facebook, the agent could trust my identity and addressed me by name. He only needed to ask me my date of birth to give me my frequent flier number. The whole interaction took less than 5 minutes.

A couple of weeks later, I messaged United again ‘Hey! Can you tell me the points balance?’. As you can see below, from the context of our previous chat history, the agent was able to answer my question right away without any further back and forth.

Interacting with businesses like this feels a lot like interacting with a person. They know who you are and remember what you have previously spoken about. Furthermore, going back to a website and finding there a history of your interactions with the business or a new message that an agent left for you after your last interaction makes the experience of visiting the website feel more unique and personal.

These are early days in the evolution of chat as a channel and we are seeing a lot of experimentation in both human and bot powered chat experiences. As an experience designer, it is exciting to experiment with and learn about what this new medium should and should not be used for. What experiences can be automated through bot chat experiences and what scenarios benefit from human agents.

I look forward to engaging in and contributing to this conversation.

What might a watch interface be good for?

As a little design exercise, my wife and I (Yeah, I’m married now. It has been a long time since my last blog post) spent some time thinking about the question “What are scenarios in which a user’s computing needs can be served best on a watch interface?”
Here are a few types of scenarios that we came up with:

Times when you take out your phone to quickly check a small piece of information. Apple showed cases of checking the time, weather and stocks. You can see that extending to an app to check your bank balance or a utility bill. These could be served by apps or Apple’s ‘glances’ depending on how important the information is to the user.

Times when you are regularly going back to check some information like tracking the status of an order or a flight or the score of a sporting event. These could potentially be served by temporary widgets which live on the watch face or a swipe away while the information is needed.

Times when you take out your phone to take a quick action like identifying a song on Shazam, calling an Uber, flashing your boarding pass etc. In these cases, usage could shift almost entirely to the watch from the phone.

Dynamic and super-contextual information. Like Google Now on steroids. So when you are using public transport and you reach a bus stop, the watch can have easily accessible, information about when the next bus is expected. Citymapper has a mock for a similar use case. When you are walking off a flight, it can tell you which belt your baggage is expected at.

After thinking about a lot of these use cases, I think the watch should derive context based on what you are doing in the real world or on your other devices and make easily accessible, information that might be useful to you. This might mean that you wouldn’t need to invoke an app on the watch to see that information, it should just become available to you. You could possibly just give permission to an app to use the watch as a second screen like giving an app permission to send you push notifications.

An example of this could be say you are using public transport and using Citymapper on your phone to navigate. The watch interface showing you next bus timings should automatically become available on the watch a swipe away from the watch face without you necessarily invoking the Citymapper app on the watch.

In any case, I think these are exciting times as the design and development community experiments with various ways that this new device can bring value to users’ lives.

How the failure to understand design contributed to Nokia’s downfall

A friend sent me a link to this very insightful blog post by Adam Greenfield, the former head of design direction for service and user experience design at Nokia. The post is titled (for some reason) ‘Nokia: Culture will out’  http://bit.ly/gfNIxV

I think it does a great job of articulating some of the problems that exist in many organizations today, including my own, where design is not given its due.

Quoting some particularly insightful points here, in the order that they occur in the post:

“This was that you could no longer think of mobile phones as communication devices. You had to conceive of them as interface objects through which users would experience content and command functionality that ultimately lived on the network.”

“Of course you still want to produce your offering for the lowest achievable cost — but that cost is bound up in intangible, nondeterministic dimensions of design, in ways that are only partially- at-best quantifiable.”

“It’s just not particularly wise to allow engineers to make decisions about things like product and service nomenclature, interface typography and the graphic design of icons: they’re, I daresay, not even neurocognitively equipped to do so.”

“My point is merely that, at Nokia, engineering has been allowed to displace what is properly the company’s design prerogative almost entirely.”

“It’s not that the NFC-based, phone-to-object interaction didn’t work. Of course it did: it had been engineered perfectly. But what it hadn’t been was designed.”

“A designer committed to the user and the quality of that user’s experience gets this in a way only the rarest engineer seems to.”

And this one I thought was a particularly brilliant insight for me, maybe because I am an engineer by education and a designer by passion:
“Designers are also, by training and predilection, inclined to design for the usual, where engineers are taught a kind of rigor that compels them to account for, and overweight, low-probability events.”

“I have to conclude that it’s this inability to even perceive the clear makings of an unacceptably bad user experience, let alone address them as profound obstacles to success in the marketplace, that leads to situations like this.”

UX Aha moment- McDonalds Order Information System

Have you noticed how efficient the staff at McDonalds is? Have you noticed how the person who takes your order is different from the person who puts it together on your tray? There are sometimes more than one person working on putting your order together. I was awestruck when I saw this. So I asked the guy who took my order, how do you guys do this?

So he pointed to a screen behind him. The folks who were putting together my order were continuously referring to it.

Here’s how it works… for the sake of simplicity, let’s call the guy who takes my order the order-taker and the folks who put it together the assemblers.

Order initiation and completion: As soon as order-taker finishes taking my order, the order comes up on the assemblers’ screen. Once it’s served to me the order-taker clicks a button on the screen and it disappears.
Order Representation: The order is represented in a rectangle on the assemblers’ interface which is otherwise a black blank screen. The screen can display, if memory serves me right, up to 12 such order-rectangles. This contains individual colour-coded line items for my burger, fries, coke etc. The non-veg items are represented in red, the veg in green, fries in yellow and drinks (well at least my coke) in blue.
Timer: On the top of the rectangle there is a timer which indicates how long it has been since my order was placed. Once it crosses a certain threshold, the timer starts blinking.
I understood all of this in the 140 seconds it took for my order to get ready.  This was an aha moment for me in terms of software UI design. Simple and effective!

Though now that I think about it, there are somethings I am curious about…
How do they represent Small, Medium or Large for fries and coke?
How do they know which items are already on my tray and which are yet to be served?
How do they know who is assembling a particular order?
What happens when the number of orders exceed 12?
What happens if I change my mind about an item after I place the order?

Maybe I’ll test them a bit next time 😉

Also, my guess is that this is a small part of a larger information system. They probably have some information displayed in the cooking area too.

Who cares about Usability?

This post has been in the making for many months. When I started to think about getting into usability and UxD, I started to think… Who really cares about good usability? I mean people learn to live with bad interfaces as long as they can achieve their goal. Think about IRCTC (the Indian Railways online ticketing website) or ICICI Bank’s online banking and trading portals. They have poor interfaces, lots and lots of poorly organized data on each page, and yet people use them because they can achieve their objectives. As another example, I love Cleartrip but when they started charging extra money for booking, I started booking on the airline websites, which are much worse in their usability. Given these, I thought, which business in their right mind, would spend a lot of money on usability? (which is essential, so I can get a lot of money doing it!) Or think about it this way, how much does good usability really impact business?

After many months and many conversations here is my list of some places where I think usability really matters.

1.      Highly competitive spaces
Think about e-Commerce. There are hundreds of online shopping websites out there. If the user can’t quickly find what he wants on your website, he’ll just find another one. If the product cannot be found on your website, as far as the user is concerned, it does not exist on your website.

2.      Places where usability has safety implications
Aircraft cockpits, car cabins, controls in say the office of a nuclear plant. In these places, if the user hits the wrong switch at the critical time or doesn’t get the right feedback or fails to recognize an alert, it could lead to a loss of many lives.

3.      Interfaces where speed is essential
The example I think about here is call centres. While an agent is on the phone with a customer, he/ she not only needs to have all the information that the customer might need, he/ she also needs to know where exactly to find it and fast! Consider a typical call centre, lots of customers call every hour with lots of different questions. The agents are under pressure to complete each call as fast as possible because the more calls they process the more money the call centre makes (directly or indirectly depending on the business model). Add to that, these places have very high attrition rates, 30-40% attrition is common, so the company can’t spend too much training their agents. Thus the agent needs to be able to find the data he/she needs in minimum time with minimum training.

4.      Self-help websites
Take the example of online banking or portals where a telecom operator puts up information like the current bill amount, schemes available etc. These help not just the customer do his work faster but also help the businesses save millions in customer support. The alternate to an online money transfer for example would be customers coming to the bank, which means the bank needs more agents and thus higher wage bills. The alternate to checking you bill, validity, offers etc online would be to call up customer support, which is a cost to the company. Thus self help websites need to allow the average customer find the information/ execute the tasks easily and intuitively.

5.      Not Facebook!
Well earlier I used to think that social networking is such a competitive space, you have to have a great user experience for the user to chose your product over another. I have come to realize, however, that people have already invested so much in Facebook (/ your favourite social networking site) with their contacts and conversations and photographs and groups that unless Facebook screws up really bad(Which is what happened/ is happening to Orkut in India), people are still going to use it despite Idiotic (!!) additions like Live Feed and News Feed!!

So that’s my list so far. What do you think? Where else does usability really matter? And where else does it not matter? Let me know…drop me a line. Ok gotta go.. Facebook calling 🙂

Brilliant example of Anticipating the User’s Need


The arrow points to this fantastic feature on my iPod Touch. While listening
to Podcasts or lectures on my iPod I often take a break and then resume
later. So to gather the context again I need to go back in the lecture a
bit. This feature does exactly that… hit this button.. oh wait …can we
still call it a button? Maybe there should be new names for these controls
on touch interfaces. Anyway that’s not the point…so hit this control and
you go 30 seconds back in the podcast/ lecture. It’s so much simpler
compared to grabbing the slider and pulling it back by the same duration. I
love it! I have started to use it pretty often now.

Though I think they haven’t done a very good job of visually designing it’s
representation. I, for one, couldn’t figure out what the control meant till
I hit it accidentally one day. Do you think it conveys ‘hit this to go back
30 seconds’ well enough?

Anyway, what I find amazing is that, firstly they anticipated the need to go
back a bit on a podcast and secondly, the 30 seconds duration, which I have
found is a good enough duration to get the context again. What kind of user
observation would they have done for this? And how? I mean did they follow
users around? Did they screen record usage? Is giving the device to a bunch
of employees good enough to give you such an insight? How would you
anticipate this need?