November 17th, 2010

Myki – a lesson in Usability

My user experience with Myki, the new ticketing system in Melbourne.

— Matt Fellows —

For those who aren’t aware or don’t live in Victoria, Myki is a new ticketing system introduced for public transport across Victoria. I’m not here to talk about the 1.3 billion dollars spent on it or the politics around it – that’s been said and done.

What I want to highlight, is the terrible user-experience I’ve had with Myki, in particular the HCI aspects of it, and why I believe some of the problems were completely avoidable.

My story

I’d like to first start with a short story and describe later exactly what had happened at the system level.

Genesis

After a few weeks of my best mate telling me that I should get a Myki because they were cheaper, I regrettably did. The first few days were a breeze, the Myki came in the mail within a couple of days and I was up and running. Nice. Apart from the annoying placement of the Myki reader at the major train stations (placed knee height under the existing paper card reader) it worked well.

Until one day I tried to get on and it and it said I had no credit, in fact as it turned out, I was able to overdraw my card into the negatives. How could that have happened? Luckily, I still had an old card that I could use so that I wouldn’t miss my train. The next morning prior to leaving I remembered that I needed to top-up the card. Happily, within a few minutes I’d quickly logged on and purchased a 28 day discounted pass. Upon arriving at the station I was surprised to find that the credit hadn’t yet shown up on my card. This time I had no spare card, and was late. When I sat down on the train and opened up my laptop, underneath the big confirmation message there was another one further down indicating purchases made via the internet or call-centre would take up to 24hrs to take affect. What?! Why not tell me before I paid.

The day after, thinking that my card would now be active I was further shocked to find that it still hadn’t taken effect. Late again. That day, further digging around on the Myki website and looking at my account details I had realised that because I was in the negative, I couldn’t just buy a 28 day ‘pass’ – I had to top-up the ‘money’ mode of the card to at least $0 before it could be used! So, knowing that I had to purchase online with at least 24hrs notice, I now had the choice of either buying a paper card, or the daunting task of using the Myki machine for the first time to add the money to my account. I thought I might as well give the Myki machine a crack.

As I knew I would have lines to compete with, I arrived at the station a bit earlier than I normally would have so that I had time to deal with the machine. Well, I underestimated the time I needed. As the train I was supposed to catch whizzed by, I was left stranded and extremely frustrated. I had placed the card onto the machine, which correctly displayed to me my Myki account information, I put it back in my wallet so I wouldn’t forget it. I had chosen to put $20 on to the card but had no cash on me so I selected to use the EFTPOS option. Sliding my card in and out I was frustrated to see the ‘please insert card’ message on the screen. It took me some time to realise, but unlike any other EFTPOS machine I’d ever used, I needed to leave the card in the holder. I then entered my PIN and watched the screen – watched it tell me it couldn’t top-up my card. I assumed that because I’d taken so long with the EFTPOS that it had timed-out or something. Starting over, I quickly repeated the steps that worked and had the hourglass animating in no time. Still, it couldn’t put the money on my card. I don’t know what possessed me, but I had the idea to leave the Myki card in it’s little holder too. Repeating for the nth time, it worked. I felt like the biggest idiot, but I then thought to myself “I’m a reasonably intelligent person, who deals with HCI every day – this is their fault!” Why couldn’t it tell me that I had removed my card and prevented me from moving forward? I was left thinking… well, I won’t repeat it.

Summary of issues

Design for all, optimise for most

It is clear that the designers of the Myki ticket machines have tried to take into account all potential users: Young, old, disabled etc. What is equally obvious, however, is that there is a bias toward the disabled meaning that the vast majority of abled users are subjected to an awkward experience.

What do I mean exactly? The machines are designed for the vertically challenged – take a look at the figure below – our ex-Transport Minister Lynne Kosky is shown using the machines and can be seen standing back about 3 feet so that she can see what is on the screen. It’s around about her hip height. So what? Well, for one you can’t both stand back 3 feet and interact with the screen at the same time, so you end up in a strange dance-like interaction with the machine, constantly bobbing up and down, back and forth.

Myki - Lynne Kosky showing how easy it is to use the Myki ticket machine

Myki - Lynne Kosky showing how easy it is to use the Myki ticket machine

Now, admittedly I’m 6’3 but it literally comes up no higher than my hip height. Here is a photo taken from my eye level from a distance of about 4 metres:

Myki ticket machine - Photo from my eye level

Myki ticket machine - Photo from my eye level

As you can see, if I were to stand directly in front of this machine i would be looking at some old terracotta bricks. In fact, this is my view of the screen standing about 1ft away:

Myki ticket machine - view from regular (~1ft) standing position

Myki ticket machine - view from regular (~1ft) standing position

Apart from hurting my neck, I can barely see a thing. In fact, I need to stand back about 1.5m to actually be able to read the screen clearly:

Myki ticket machine - view from 1.5m standing position

Myki ticket machine - view from 1.5m standing position

As for the Myki readers (below), ironically they are too high for many disabled users (see this article, apparently Australian standards state they should be < 90cm, but the new scanners are 120cm) and these users were able to apply for a free travel pass instead – so now we have a system that is optimised for disabled users but used by even less because they can travel for free anyway?

Usability principles: Design for people

Solution: Implement a screen that can be tilted to a suitable angle for any user or even a sliding screen that can be adjusted for height. I’m sure there are other more creative options.

Fail Early

By far, the most annoying and disappointing experience was the interaction with the ticket machine itself. Apart from it being designed for Dwarfs, the Mental Model built up in all of us from years of ATM, EFTPOS and other POS machines do not apply to the Myki machines. For example, ATM’s suck your card inside the machine, and give it back to you when it’s done. There is no ambiguity – the ATM retains the card as long as the interaction is in progress. In contrast, the hand-held EFTPOS machines commonly seen at fast-food restaurants or supermarkets afford a slide-in, slide-out interaction. We all get it and are familiar with it. Unfortunately, the Myki EFTPOS facility mixes the two: it affords a slide-in, slide-out interaction, even though it expects it to be left in.

As for inserting the Myki card itself, it also expects it to be left on it’s little cradle for the duration of the interaction. Even though the affordance of this requirement is weak, ordinarily it would be OK if the system appropriately informed the user that it needed to remain on the cradle. And when I thought about it, it makes sense – you don’t want to accidentally put your money on someone elses card. But who’s thinking about that when they’re in a hurry to buy a ticket? However, the system let me progress to the point of payment, failed and gave me a crappy, unhelpful message instead.

As a person working in the HCI field, it really disappoints me to see things like this. If it confuses me, it’s likely to have confused others.

Usability principles:

  • Affordance: The machine does not send strong signals on how it should be interacted with
  • Mental Model: The machine does not work as I would expect, because it deviates from the behaviour of common existing machines, such as an ATM
  • Keep the user informed: The machine should at least give me an accurate update as to what is happening (i.e. “You have removed your Myki from the reader, please put it back…”)
  • Fail early\Error Prevention (Forcing function): The system should prevent me from interacting with it so long as my Myki is not in the machine
  • Error messages: The system should give me helpful advice as to how to proceed (i.e. “Sorry, I could not put money on your Myki because your Myki is not present on the reader. Please leave your Myki on the reader and try again…”)

Solution: Prevent the user (forcing function) from progressing if the Myki card is removed from the cradle. If they do, tell them why they can’t progress and how to move forward (constructive error messages). As for the debit card, I imagine that hardware isn’t cheap to replace so probably the best that can be asked for is a big fat sign.

Visibility of system status

One of the reasons I was able to go into a negative balance on my card was because I was given poor feedback at the time of swiping my card on and off. For one, the display only lasts about 2 seconds (Thanks to Brandon for pointing out that the display will actually hold the reading until the card is removed from the reader) and secondly the reader display (unlike the old one) is very susceptible to vandalism and glare. So, no wonder I went in to the negatives – I probably missed negative (-) symbol in front of my current balance. So, I do my best to remember when my card expires or look online and top up before it does so I’m not stranded. If I try and touch my card again, it correctly detects that I’m swiping too quickly and doesn’t charge me (correct) but doesn’t re-display my balance and expiry. This is sub-optimal because I now need to remember. The old system was good, because it gave you instant feedback in close proximity (in time) to the action performed.

Here are 2 examples of machines in my area in the morning:

Myki reader - reflection makes it difficult to read

Myki reader - reflection makes it difficult to read

Myki reader - slightly vandalised but with rising sun

Myki reader - slightly vandalised but with rising sun

Solution: Unfortunately graffiti is part of the decor of most train stations, so the best you can ask for are screens that are less susceptible to it, or better protected. Allow people to touch their cards to query the current balance.

Solution: User Testing

I’m willing to bet lots of money that the makers of the Myki system did not do any of the following:

1. Test their proposed system with real users, in real situations

Testing any of these machines in an isolated, controlled environment is OK if you are simply testing functionality. However, these machines should be designed to work when people are in a hurry, and stressed. Buying a ticket with 5 or 6 people waiting behind you is a stressful thing; you’re conscious that whilst your busy fiddling with your debit card, that 5 people are impatiently waiting behind you thinking they’re going to miss they’re train. You’re distracted. This means that the system needs to be simple, intuitive and needs to behave as expected first time. The user is unlikely to efficiently resolve roadblocks, and will generally just give up quickly and resort to bothering the lady behind the counter to help them (at the time of writing they can’t do anything with a Myki yet, unlike the old system – i.e. you have to keep trying, or buy and old ticket).

I’m confident that user testing would of helped to reduce this confusion.

2. Test their system outside of a controlled environment

Assuming that basic user acceptance testing was actually performed, it is clear, at least to me, that had the creators of Myki actually put a test reader outside in the sun and glare at peak times of the day – where the sun is setting and rising – that their test subjects would have identified that they couldn’t read the screen in the approximately 2 seconds given to do so.

Furthermore, the same applies to the ticket machine which, from the limited number of stations that I’ve visited, seem to be placed where there is a lot of people and limited space. Oh yes, and aren’t used by 5 ft tall Neanderthals.

Final word

The project did get something right, and I want to acknowledge it here: The station that I currently get on never had a ticket validator near the entrance closest to the car park (it was inside the building 100m away) meaning that if you already had a ticket and constantly are running late for a train you would have 2 options: run to the validator and get on the train, or simply get on the train and face the ticket inspectors down the line.

Thanks for putting a Myki sensor in the obvious place.

Further reading on usability principles discussed

Other issues, just because

“Touch on, touch off.”

Commuters are exposed to excessent and annoying voice-over reminders and advertising, reminding us to ‘touch on, touch off’. Needless to say that it was going to cause usability issues. I have no doubt the project team thought that “users’ would get used to it”. Yes, maybe we will. But what about those people who use public transport infrequently, new commuters, interstate\international visitors etc. The fact that we need to be reminded every 10 seconds tells me that the creators of the system realised that it was going to be an issue (whether through user-testing [unlikely] or obvious reflection) and decided rather than to fix the root cause they should train the customer instead.

User’s of the system are expected to not only validate their ticket when getting on public transport, but also to validate it again whilst getting off the mode of transport. Apart from the fact that the Myki sensors are at best awkward to use (the card needs to be placed flat up against the surface – so much for the proximity reader) this has the undesirable effect, for example, on trams where there is both congestion when getting on and even worse when getting off.

Why the touch off? The reason is so that the system can charge you correctly (they’re doing us a favour). Various Zones in Melbourne are charged at different rates so not swiping off could mean that you get charged a provisional fare (clarified below to a 2hr Zone 1 on trams. It is stated to be a 2 Hour Zone 1-2 fare on trains but  goes on to also say you won’t be charged more than $9.92 for not touching off). For this reason, I’ve seen distressed passengers fighting through congestion to swipe off so that they aren’t overcharged and miss their stop.

The congestion applies equally for trains and buses too, however thankfully the sensors for trains are on the platform itself.

Solution: So, what to do? Well, the original promise of proximity sensors will get us some of the way there. Other cities (for example London) have moved to a flat fare.

  • http://www.myki.com.au Brandon

    There are a few inaccuracies in this. Firstly, the topup machines and readers are the same used in many places around the world with similar smartcard ticketing systems. They are made by ACS systems. Secondly, balance doesnt just remain on the yellow reader screens for 2 seconds, it remains on there for as long as the card is held to the reader. Thirdly, touch off is not required on trams unless you are only travelling in Zone 2 and want to get the cheaper Zone 2 fare. The default tram fare is 2hr Zone 1 $2.94 full fare. You also don’t need to touch off if you’re using myki pass in the zone for which it’s valid. That said, I’ve been using myki for just on a year now and have never been unable to touch off. As Metcard is phased out, myki readers will replace the existing Metcard validators and top up machines will be installed on trams. You’ll also be able to top up on the bus and at many retail locations including 7-11s.
    Also worth noting that while myki is an RFID card, it’s not the same as a security pass like many of us use to gain entry to work. Those cards only have to be read, whereas with myki there is a readwrite data transfer taking place, therefore it will take slightly longer, hence the message is “touch” on, not “tag” or “swipe”.
    Thre reader hardware actually allows for myki cards to be read from a distance of up to 10cm, however this was reduced to avoid accidental touch on or off.

    While you do make some valid points, usability is generally determined by those least likely to take up new technology, the fact that there are now many elderly people using myki with no trouble is testament to its usability.

    That this comment stupidly requires me to enter a website before continuing certainly detracts from its usability!

    • Matthew Fellows

      Thanks for the clarifications Brandon, especially around the behaviours of the reader itself and ‘touching’ off.

      What’s confusing to me, is that if the default fare is a 2hr Zone 1 how it will be policed for those in Zone 2? It is most certainly a reasonable default, but probably there is enough doubt in most peoples minds and as stated – the ‘touch on, touch off’ drone – to want to touch-off just in case. Better to just have a flat rate and avoid ambiguity and accidental fair evasion i.e. ‘I didn’t know I had to touch off…’.

      It is fair to say that Myki has done a good job and I will update the article to reflect that. Certainly my stance appears very negative which is a bit misleading given that I have completely ditched the old system and now use Myki daily. The real (intended) emphasis of the post has to do with the specifics of the ticketing machine and admittedly I got a bit excited :)

      …and quite right about the URL, looks like a bug to me!

  • http://www.myki.com.au Brandon

    No problem Matthew.

    If those people travelling only in Zone 2 don’t touch off, they’ll be charged the default fare, which is, as I say, 2hr Zone 1. If they want the cheaper fare they need to touch off. So there really is no way that they can fare evade, except by not touching on at all.

    You might want to have a look at the myki thread on Whirlpool, there are a lot of people in the public transport industry who post on there and you can gain a lot of information from them and from other myki users

  • http://www.riddellclare.com.au greg

    Terrific article
    You will love this

    My local vline station is Riddells Creek – its one of those semi rural places, folk commute to Melb cbd via vline
    Our myki terminal is facing north, directly north, kid you not.

    But they have a rainsun shade over it. That works as long as the rain is near vertical, ans can never work with the sun angle.

    They could have faced it east west or south with no user access issues, able bodied or wheel chair bound.

    Utterly inept but I will wager the project manager had a PMI qualification and has met all goals.

  • krzystoff

    Brandon, you may be factually correct about the usage of Myki around the world, this system was first used in 1996 in the US and is now used in most US cities and many developing nations around the world.

    16 years on, many major cities have switched to more advanced, secure, simpler systems. that said, many myki-level-equivalent AFC systems remain in use in that have few or none of the usability issues that Victorians face with this implementation of the same, well-matured technology. it was touted by the former State government that KAMCO designed and purpose-built system for myki, but the evidence has continued to demonstrate that have shoe-horned cheap components into a poorly designed and untested framework.

    like Metcard and the Scratch tickets, Melburnians must face a future daily battle with this system and all its warts and buboes, until the next one replaces it in a decade for 10 or 20 times the cost of myki.

  • Madeline

    I agree with Krzystoff’s comment that ‘cheap components’ have been ‘shoehorned into a poorly designed and untested framework’.
    I think this is particularly clear in the recharge machines. Where a dated atm design has been dumped next to the Myki recharge screen. The two designs do not compliment or correlate with eachother and without clear feedback, you are left looking back and forth between the tiny inset atm screen and the Myki touch screen.
    It is true that these systems have already been implemented internationally, but with huge amounts of research and testing.
    It appears as though Melbourne said- “i want one” and bought it off the shelf without consideration of actual human interaction with the design.

    The worst part is that the blame is passed onto the users for their incomprehension of how to use the system. There is no ownership of the the fault in the design.

    My argument, is that we are the users- and good user interaction should be intuitive to us. In being so counterintuitive the design might have met the brief of the client, but it has certainly not the met that of the users.

  • Madeline

    I think Myki could incredibly improve their system, just by looking at your observations from a users perspective

  • Cameron

    I think much of the idea with myki is that the designers / govt know that people will make mistakes be it forgetting to put a credit through, top up, tag on or tag off, have enough credit. It’s like the professional charity workers in the street. They don’t want your lose change, they want your credit card numbers. The government want to charge / fine people for not being able to use a system which clearly has many flaws. In WA, Londaon and Brisbane, you can still buy one way tickets if you dont want, cant use etc a Myki / Oyster Card. Not here.

  • http://uptakesolutions.com.au Brenton Johnson

    This is great! thanks!

  • http://www.purplesquirrels.com.au/ Ben Foster

    Great article – this pretty much sums up everything I have experienced with Myki, what’s worrying is that it’s now 4 years down the track and all these issues are still present!