Phased re-designs

Question

I have worked on several projects(line of business web applications) in my career that involved a complete UI overhaul. Most of these projects were enterprise web applications that took a substantial amount of time to re-design. Some of them were done with a complete swap and some done in phases.

I was surprised by the reactions of the customers. If I had bet on the outcome I would have lost my pants. The wholesale swaps caught a ton of grief from the customer. The design really wasn't the issue, it was that it changed period. The phased changes went smooth as silk but looked strange for a while. A very undesirable UI for the interim.

Is it better to do complete re-designs in phases or a complete swap with the new design?

If dealing with a phased approach, is it worth the trouble to maintain both designs for a while, allowing the user to choose which want to use?

Even if possible, is it effective, or will the same problems surface when the old version is no longer available?

Answer 1

Think you need to look at it from the other view. Both approaches are correct if handled/planned. What's important on re-designs i work with is managing the users. Start a fun but informative section on the home page discussing the plans. Poke fun at the old site, stick in some polls, involve them in the change.

Answer 2

It really depends on the situation. One anecdote might help:

I design an app that included an HTML Email editor. The original design was error prone and had alot of problems. There was no way to "fix it" so we scrapped the design and rebuilt from scratch...much better this time.

However, the customers freaked out. "You moved my cheese!!!" I had to spend about 5-10 minutes per customer on the phone showing them the difference, calming them down and making them happy. Complaints about this feature disappeared.

So in the end, I did a full swap of something and it was extremeley painful. Then they calmed down and it was all OK.

Facebook redesigned, everyone screamed, then they calmed down. Complete swap is OK if you are prepared for the short term insanity.

Answer 3

I like the metaphor - a big web application is like an airport. This statement has some consequences:

1. People usually don't visit it just for fun - they need something. So you should help them reach their goals and provide great experience.
2. You can't just close an airport to build a new one. If you need to improve something you should try to create as little inconvenience as possible.

It's also true for web applications (at least for big ones). Everything could be improved evolutionary. In fact you can create an absolutely new interface by changing different parts of the old one. But of course it's good to have a complete picture of what should be done. If you redesign different parts without understanding of the whole project, achieving good results would be much harder.

Answer 4

+1 for the incremental approach, realigning instead of redesigning. Google, Yahoo, Amazon do only small changes at a time: you don't see big changes on their websites, though they're constantly changing, tweaking something.

You can find more info on UX Myth: You need to redesign your website periodically.

Answer 5

I would say phased is better than complete swap. But you have to manage it well.

A complete swap is indeed too rough for people. Joel Spolsky once said that rewrite from scratch is something you should never do. Although that related to writing code, I think it applies to users as well. You'll loose experts when they have to get used to something completely different. Phasing can help a lot. Also by giving you more time to involve your users in the process, because they can understand it more then with a complete change.

I think the best way to manage it would be beta versions. In which the actual thing continues to run and active customers can help to make the change in a second place. This way you also communicate clearly to users who otherwise might complain just about the change.

(There is one caveat: when the app/site/thing is hardly used. Then I think from a design and moving-forward view, it is better to do a complete swap. But the usage and project shouldn't be too big then.)