What happens when you bring the UX person in last.

How many of you have been to a conference that has a UX/UI Person on stage discussing the mystic art of software development and design? In that said session they at some point raise the slide that outlines you should engage a UX person early and think about UI/UX from the start.

How many of you then go back to your respective cubicles, nodding in agreement but then immediately go into a new project ignoring the said suggestion?

Don’t lie, I see you looking back in a nervous manner and shouting out reasons like “Well, we didn’t have the budget” or “My boss wouldn’t …” etc.

Meet Mr Wolf


Just like in Pulp Fiction, a guy like me is called in after the crime has been committed. I’m the guy you bring in after you accidently killed someone and its my job to navigate the mess in order to get you back to your life without prison time. If I succeed, you don’t’ spend the rest of your life in jail if I fail, well, learn how to fight using prison rules.

When I come into a team in this situation, the thing that I notice the most is they are looking for guidance around a plan, in that it’s a case of me analyzing the situation, asking a series of specific questions relating to the said scene and then giving them a task list to execute on – whilst being clear to stick to my rules or well, good luck in jail.

The problem with this approach at times is that you have usually one or two people in the room who ask for your help but at the same time are giving your orders on how to clean the mess up quicker – each time they do this they in turn increase their chances of prison time.

It’s a hard balance to participate in from my perspective as I have to figure out a way to firstly give a design and experience to the software’s targeted end users in a way that isn’t just a screen after screen of tree controls and datagrids whilst at the same time having a low impact to the codebase and lastly but more importantly doing it within a very tight timeframe/budget?

Its hard work and you know what, its going to cost you so don’t whine about it.

Had you called me from the start, it would have been a completely different outcome and yes, you’ve heard this thousands of times at whatever conference you last attended – engage a UX person early and let them direct the screens overall compensations – design first engineer second.

I personally have been pulled into over 30+ projects in the last year that have this exact situation unfolding before me, in that it’s the last two sprints of a project, I’m playing a massive game of UX Tetris with WPF/Silverlight or Wp7 and I’m constantly being harassed on time/budget questions.

It sux but that’s the reality of the role I play in this business, the guy who can code and design at the same time. Its why I charge the amounts I do and sure the price attracts attention but in truth If you follow my rules and approach you will come out with a finished result. If you interject along the way with the way you think it should be done, fine, I’ll do it your way but if it fails – given the inexperience so far, it will – then to be fair, you were warned.

My way or the wrong way.

The way I approach situations where I’m brought in at the last hour is via the following routine.

  • The Primitives. In every application you have what I call the primitives, in that these are the buttons, modal windows, textbox's, scrollbars, checkboxes etc. .. the stuff you get out of the box for free with .NET. My first attack posture is to start building out a resource dictionary library for you to bootstrap your UI against. In that for example TextBox and Button controls I start putting into what I call the UI-Shirt-Sizes, Large, Medium and Small. If your form in question requires the user enter 15chars min/max, who cares, the end user is open to the idea that this textbox is a small one that magically doesn’t let me type more than those pre-defined characters.

    If your software has a large sentence like “Find a users profile”  labels on buttons - guess what I’m going to do, re-label that as “Find..” keep it simple less extraneous cognitive load and more assume the user has used software before they picked up yours.

  • The Layout. Chances are you’ve probably put together a UI that I can only describe as a DataGrid orgy followed by copious amounts of Modal windows and screens that probably looks like the dashboard of a Qantas Jet in terms of fields/inputs etc. Just for giggles, I’m also likely to find a TreeGrid control because of some random hierarchy based navigational weird mutation of a need (you know who you are, there's no shame in admitting that)

    I’m going to simplify this down to the point where the data flows in a fashion that makes sense to the outcome of the screens purpose. I’m also going to look for ways to make use of a party trick called “progressive disclosure” as you do want the user to feel like they stand a chance at success should they use your software don’t you?

    This is what I call the hostage negotiation in that chances are there is an entity in the room that is locked on the way it works at the moment and its my job to find a way to get you to release parts of the UI so I can find a happy resolution to the situation. I’m going to ask you to give me a little control over how the UI comes together and in return I’ll turn the lights back on followed by some pizza. We need to build trust and you got to work with me on this one, I can make good on some promises if you do!

  • The Validations. I have seen some crazy ways that developers have approached the simplistic concept of alerting the user that they did something wrong. What I have noticed the most is its kind of OnChange vs OnSubmit mode of approach. The reality is validation isn’t that, as you have the “Hey before we show you this form, here’s where you need to focus”, “Hey I just noticed you filled out that field wrong, can you fix it”, “Hey I am about to send this data off and noticed the form isn’t really done yet?” and lastly “hey I know at the time you sent this the form felt like it was good, but the server just called me and told me its wrong, so can you go fix” .. point is, IDataErrorInfo implementation is only going to work so far.

    I focus on this area is this is where at times bugs tend to get brought up and it can be a case of where the most effort can be spent trying to undo user fail. Its important that one approaches this in a way that makes sense to the end user and you also find ways to decode the error in a meaningful way – not one that aims to reduce the user to a dribbling mess of “I don’t speak computer geek?”

    Validation styling and alert states are crucial.

  • The BackgroundWorker. Its not about just fixing the UI look and user experience fail points its also about shift the work into areas that make the application feel snappy. In WPF the UI Thread is an absolute pain in the butt when you at times talk to WCF – in that I have seen a lot of apps that keep the entire workload under the one thread only. In Silverlight this can be a fairly low risk situation given Async works ok, but in WPF it means your application grinds to a halt until the service layer comes back from the dead. It also isn’t just a threading issue its also a latency issue as well.

    Latency is a buzz killer in making the user feel like the application is responsive, it creates this effect in which the user punishes themselves and attempts to pay their debt by trying again and again etc if left unchecked. Its situations like this I look for ways in making the user aware that they did a good job but at the same time finding ways to NOT remind them of time – as time is the enemy given each millisecond you are banking hate debt with the user?

    This is where I look for ways to use some slight of hand techniques to convince the user there isn’t a problem and everything is fast / efficient in the software. I also may lie to the user if I can eg Please wait while Security authenticates you” – damn those Security Nazi’s I agree, it sux but what can you do – its actually an effective way to pacify users as you all collectively shake your fist at IT Department for always riding you about security – when I reality I’m waiting for blah service to wake up from its slumber?

    Point is, find a common villain to throw under a bus or find a way to keep peoples attention away from their watch (eg: Now herding llamas for the great stampede …< MAXIS do this in their games, it works)

I’ll leave there as this is turning into a tomb of gospel around how I approach my job, but the point is that I do have a process and there is a method to my madness. I’ve been in a lot of fire drills with WPF/Silverlight and WP7 and I’ve now settled on some patterns that have produced results  around nice UI/UX and customers happy.

The reality is this though, you could of saved yourself minimum double through to quadruple the amount of money it cost by bringing me in early instead of late. I can’t say it enough, engage early and upfront you will save, you may be skeptical htats fine, but either way a person like me gets paid – its just a matter of how much?

Related Posts:

  • Arend

    Great post, as a data and treegrid lover this a gold nugget.