Good thoughts on how the waterfall method doesn't work for design or development. A response from an enterprise UX person:
I'm not a big fan of orthodox Agile methods, with the "code first, ask questions later" mentality. I think there is a benefit to spending some (small, controlled) amount of time up front to identify the tasks the user needs to perform and to translate those tasks using user-centered methods into at least a skeletal UI design.
However, I am a big fan of the general concept of iterative design and development. You are right: this is where Enterprise software typically falls down. But it's not just the UX process that needs to be iterative, the whole dev cycle needs to become much more iterative.
Typically, massive amounts of time are spent up front identifying requirements and doing design work, and the dev cycle gets pushed later and later. As we all know, iterative development and design gets critical pieces of the system working early in the cycle. These pieces of live code should be usability tested (using rapid testing methods for quick feedback cycles) and that information pumped into new designs for the next iterative development cycle. It doesn't have to slow down the process -- it can be done quickly and can greatly enhance the end-product.
Unfortunately, that's not the model followed at most enterprise software companies. I think the lack of consumer-level polish in enterprise software is a result of many factors, and cannot be fixed by focusing on just the UI design process. The enterprise dev cycle is still mostly a waterfall model, with dev cycles that take years, rather than months. There's no way to rapidly iterate on a UI and keep up with consumer trends in an environment like that. Also, the typical enterprise toolset does not lend itself to rapid iteration.
Finally, to really achieve a stellar user experience, the company has to foster a culture of excellence in UI design. It has to come from the top down (see Apple). Too often, engineering or marketing concerns trump UI design, and the end-user suffers. Design and development should complement each other every step of the way.
So, that's my diatribe. We are totally on the same page about where enterprise apps need to go. And I think Oracle is making good progress towards bringing the best of the consumer Web to enterprise apps. What's needed to move them to the next level is not just changes to the UI design process, but a cultural change around the methods and tools used to develop enterprise software.
manalang
· 1 year ago
@dustin amen bro... we need to turn this place upside down and start building apps in an agile fashion.
Meg Bear
· 1 year ago
well I'm certainly not going to defend ERP applications and their (our) usability track record. I will say, however, that we don't just do usability testing on prototypes, we also do it on the final product and attempt to adjust to what we find. Of course, not always, but at least we do try.
The problem, as I see it, was best described to me that "UE is fashion". It changes regularly, most development of complex systems just can't keep up. Before we get it built it's out of date. The real answer is to have better architectural support for regularly changing (and updating) the UI without the cost of re-doing the application. This makes it easier to adjust to changes both from user feedback and from technology innovations. One of the whole points of a SOA is to isolate the business logic to allow for more rapid and frequent iterations of the presentation layer. Not somewhere we can get overnight but I think we really are trying to set ourselves up for better success in the future.
Jake
· 1 year ago
I like a minimal UI approach OOTB (think terminal), with the data surfaced in any number of standardized APIs produced in several languages with equivalent functionality (a la Fire Eagle). That way there's a working UI, but customers are encouraged to go nuts and build the UI they really want; embrace customization, let your app developers flourish.
That's just me though. Not necessarily the best model, but it speaks to the O'Reilly's principle of data as the next Intel Inside.
All Comments
I'm not a big fan of orthodox Agile methods, with the "code first, ask questions later" mentality. I think there is a benefit to spending some (small, controlled) amount of time up front to identify the tasks the user needs to perform and to translate those tasks using user-centered methods into at least a skeletal UI design.
However, I am a big fan of the general concept of iterative design and development. You are right: this is where Enterprise software typically falls down. But it's not just the UX process that needs to be iterative, the whole dev cycle needs to become much more iterative.
Typically, massive amounts of time are spent up front identifying requirements and doing design work, and the dev cycle gets pushed later and later. As we all know, iterative development and design gets critical pieces of the system working early in the cycle. These pieces of live code should be usability tested (using rapid testing methods for quick feedback cycles) and that information pumped into new designs for the next iterative development cycle. It doesn't have to slow down the process -- it can be done quickly and can greatly enhance the end-product.
Unfortunately, that's not the model followed at most enterprise software companies. I think the lack of consumer-level polish in enterprise software is a result of many factors, and cannot be fixed by focusing on just the UI design process. The enterprise dev cycle is still mostly a waterfall model, with dev cycles that take years, rather than months. There's no way to rapidly iterate on a UI and keep up with consumer trends in an environment like that. Also, the typical enterprise toolset does not lend itself to rapid iteration.
Finally, to really achieve a stellar user experience, the company has to foster a culture of excellence in UI design. It has to come from the top down (see Apple). Too often, engineering or marketing concerns trump UI design, and the end-user suffers. Design and development should complement each other every step of the way.
So, that's my diatribe. We are totally on the same page about where enterprise apps need to go. And I think Oracle is making good progress towards bringing the best of the consumer Web to enterprise apps. What's needed to move them to the next level is not just changes to the UI design process, but a cultural change around the methods and tools used to develop enterprise software.
The problem, as I see it, was best described to me that "UE is fashion". It changes regularly, most development of complex systems just can't keep up. Before we get it built it's out of date. The real answer is to have better architectural support for regularly changing (and updating) the UI without the cost of re-doing the application. This makes it easier to adjust to changes both from user feedback and from technology innovations. One of the whole points of a SOA is to isolate the business logic to allow for more rapid and frequent iterations of the presentation layer. Not somewhere we can get overnight but I think we really are trying to set ourselves up for better success in the future.
That's just me though. Not necessarily the best model, but it speaks to the O'Reilly's principle of data as the next Intel Inside.