Now, I'm not one to keep quiet about a new customer self service device that I think is lacking, and today is no exception.
Recently, South West Trains implemented almost overnight a replacement to their station self-service ticket machines. The devices are touch screen operated and only accept card payments from cards that have a smart chip and a PIN.
The machines went in to Waterloo station a few weeks back, and since then I've seen the queues growing... 
Today though, I had the opportunity to use one for the first time, and it's pretty bad!
But before I dive into what's wrong with it. Let's get straight to the point about how this lack of usability affects passengers and SouthWest Trains:
1 - Queue time is longer for everyone - causing station flow problems and customer aggravation - maybe even late trains with passengers being later on to the platform
2 - Staffing costs are higher - people need more help, and this costs wages - and so, ticket prices
3 - The machines are putting passengers in a bad mood - a passenger in a bad mood is more likely to abuse or assault staff or cause problems for other passengers
4 - The machine could actually cause an injury for someone with a joint or muscle weakness through making them contort themselves to see the card reader
5 - Robberies and theft will increase - while passengers are distracted by the machine, their bags, mobile phones and pockets are vulnerable
So...
- operational costs will increase
- crime will increase
- customer confidence will be knocked
- on time performance of trains will suffer
- the cost to remedy the defects in the system are going to be large whereas they could have been sorted out months before go-live
(now we get detailed...)
Why is it so bad?
The machine's main interface is a touch-screen, set for me at about waist height - I checked, and it's spot on for the optimal height based on anthropometrc data and accessibility requirements, so no bad there.
There is however, a secondary interface, a card reader and PIN entry keypad. This is set to the right of the touchscreen with the keypad at a sensible angle. Mistake one though, is that there is an LCD display on the card device, that is set flat upright, so to read it, I have to bend down. With a queue of people pressing behind me, this made me get a little intimate with the man behind me.
So, the device ergonomics aren't great, but it's not the worst I've seen. What about the user interface?
The basic premise is that, based on where the machine is, there is a list of tickets for appropriate destinatons.
The list itself puts each ticket type, price and description in a box bordered with a two-three pixel line. The background colour between these box areas and the rest of the screen is the same. Seeing as there is no other interface device though, the user doesn't have much choice but to touch the screen, despite the fact that the touch targets don't look much like buttons, and don't have any tactile look or feel.
There are two columns of tickets, and I didn't even look at the second column, probably because by that time I didn't care,... I'd found my ticket. Or rather, I'd had it pointed out to me!
The machines are staffed on a 1:1 ratio by staff helping people use them (yes, that's one person per machine). So, when I first got there, I was asked "what ticket do you need?" - I declined the help and said that I would like to try to find it myself. However, after a few seconds of looking up and down the screen, I couldn't find it. It turned out to be the 4th of about 8 options in the first column.
The challenge here is that the interface needs to be dynamically data-driven. i.e. dependent on location, the tickets available in the back-end system, etc. Therefore, inclusion of graphic elements that helped pick out the key ticket types a bit better, was presumably vetoed.
However, because every box looks the same, I have to read every box to find my ticket. All of the text inside the box is the same weight and size, so it's difficult to pick out things like destination or ticket type (travelcard, etc.) when scanning. Hence, I had to read through carefully, whilst under pressure from the queue behind me, and I failed... "ok, go on then, show me..."
For a passenger who buys the same ticket every day, then it's going to turn out just fine once they've learned where it is in the list. However, who buys the same ticket every day? Surely you buy a weekly ticket don't you? In which case, after a couple of weeks, you'll probably have memorised where your weekly ticket is in the list. But otherwise, you're going to be visiting very intermittently, and therefore not get the chance to learn the application, or you're going to be buying a variety of tickets over a period of months.
But how is the list ordered, and will it change?
Presumably, the list is ordered based on popularity of ticket; and I guess unless they start building new stations, this isn't going to change that much. But what if usability studies show the list ordering isn't working? Do they change it to enhance the intuativeness of the app, or do they keep it the same to encourage learning? But no! It's not based on popularity, it's alphabetical... which means it stops after the 16 or so items it can display on one screen! SO, what about Shepperton?? It's a popular destination, but it's not on the home screen.
So what?
Well, if the screen was designed better in the first place, allowing passengers to pick out their tickets quicker based on zero knowledge of the app, then future enhancements would be just that... enhancements!
The initial screen needs to support scanning, and people are looking either for a destination, a series of zones for a travelcard, or a duraion, for a season ticket.
Tickets should be grouped according to these, and a device (graphic or typographic) that allows these to be picked out quickly when scanning, should be employed.
It should also support more easy browsing of destinations and ticket types, rather than forcing people to type in and search for their destination.
So, after I've chosen a ticket?
Well, I waited for a while to work out what to do next.... I was staring at a screen that told me my ticket was £12.40 - I guess I need to pay then... put card in machine? No, that doesn't work...
Oh.. flashing yellow thing here, says 'touch to confirm' - ok, let's do that then.
What now? Screen appears to be the same, but no flashing yellow button. THe last time I tried to put my card in I ended up looking stupid, so I'm going to wait this time until it tells me.
But it's not! People behind me are getting impatient!
Finally, I see, in 8pt type, a message at the bottom in red, that says; "pay by card only" - it's not actually telling me to pay now, but I assume this is what it means... so let's try it. Now I notice the Visa symbols to the right,.. but the change was so subtle I didn't notice them appear.
What I discover later, is that the secondary LCD display on the card reader is actually asking me to insert my card... but of course, as we know, I can't see this. I wondered whether it was just me, as a 6ft2in tall bloke who this applied to, but as you'll see from the pics, I caught quite a few people having to stoop to see the keypad, let alone the display.
Do you see it?
Do you see it now?
So, in short, it didn't serve me very well; and from a bit of guerilla ethnography in Waterloo this morning, it's not serving other passengers very well either.
So,.. clever clogs, how would you do it then? I hear you cry?
Well, I don't know. But here's how I'd approach it;
Who are the users?
As far as I'm concerned (based on primary research and observation at stations!), there are three main types of user for this system:
Season-ticket buyer - whether weekly, or monthly, this person will buy a ticket for a particular station, or a series of travelcard zones. They're consequently only going to purchase one to three times a month. And you generally only ever buy one, for yourself.
Regular traveller - like me, doesn't always take the train, so it's not worth buying a season ticket. I use the machine regularly though, to buy no more than three different ticket types (travelcard to get to the office, return to Gatwick to go see Virgin Atlantic, or some other random location for a client visit). People may be travelling in pairs or groups, so multiple ticket purchase may be required.
Ad-hoc traveller - doesn't travel very often. May be a tourist whose first language isn't English, or someone who simply isn't confident about the public transport experience because they don't do it that often. Group tavel is even more likely, so multiple ticket purchase even more important.
So, it's a real mix of usage - which of course makes it tough to serve each type of user from the same interface.
But an interface that supported season ticket holders with one or two touches that they learn quickly, whilst the common ticket types are suppported on first touch, with the second interaction allowing selection of multiple tickets would be more appropriate.
The ability to browse destinations quickly rather than have to type in the destination. A self-narrowing list that automatically searched when the first letter was pressed would be an improvement, or maybe a scrollable display of popular destinations coupled with this search capability on the same screen. I noticed that a lot of the delays at Waterloo were caused by people having to type in the name of their destination, before then submitting a search only to find that they got no results, or the wrong results..
Then what?...
Oh, come on... I've only just started thinking about it... imagine if you put a proper user-centred design process into play, and some properly qualified interaction designers? I'm not going to try to engineer any more new solutions now.. instead, I'll make the point of this blog;
Usability makes a big difference, and it's way too important to be thought of as an add on. People find it hard to use systems, and making it easy will save you a load of cash - even if it's just about making them feel a little happier or more confident when they walk away.
Just assuming that you've designed it right isn't good enough either - you've got to put proper user-centred design into practice to ensure that you've got it right.
I spent just half an hour casually observing users and found a whole host of flaws. Imagine if the team developing this had done this exercise 3 months before they finished the product? Imagine how much better it woud be, and how much cheaper it would have been to deploy and how much cost would be saved by not having to make too many remedial changes after go-live?
I'm very happy to hear from anyone who was involved in the development of this application, as I'm fascinated to find out what the challenges actually were...