In response to “Enterprise Mashup: Lead, Don’t Follow”
This post is in response to “Enterprise Mashup: Lead, Don’t Follow” by Forrester (April 22,2008)
“If business users aren’t demanding mashups now, they will in the near future. Because mashup tools enable them to quickly create ad hoc Web applications that ‘mash’ together data from multiple sources to create any entirely new perspective on information”
I embrace the idea of mashups because it gives the business users their agility back and it reduces the rigidness that tomes with BPM and Workflow. However, I’m worried about it’s success because it feels like a deja vue.
Mashing together data from multiple sources to create a perspective on information, is not new at all, That is what we used to do in portals. We all know how successful that was. Not.
Pitfalls then were the non-availability of data sources-and user influence. User influence has two sides: one is that they could personable a lot in the portal but they hardly did; the other is that they could not personalize a lot on the data streams and were limited to what IT had prepared for them.
Key to mashups is, that users creates the application or widget without coding to combine multiple data sources. So, are mashups the new personalization options of the portal? To some extend, I can say: Yes. There is more.
“Mashups are at the mercy of their underlying data sources”. The responsibility of IT then becomes to “create mashup-friendly data sources and services.”
A safe bet is that – at the enterprise level – they hardly exist. They never took off when we were all deep down into portals.
We can re-phrase the definition of a mashup as the web 2.0 equivalent of a portal sourced with high quality highly flexible data sources and providing first class personalization. And if we forget about web 2.0, we have de-hyped mashups and can focus on the real challenges that are waiting for you.
Forrester sees many challenges in avoiding a lot of mess when using mashups.
You don’t want them to become the next generation Excel spreadsheets that steers a multi-million dollar business, without the proper business application necessities like support, back-up and training.
So on the scale between mess and good old portals we need to stay somewhere in the middle.
Let’s assume that the tools will be there to not only help you to do just that, but also to allow business users to compose the mashups in an easy way. This by the way is a valid argument no to lead but to follow despite the advice from Forrester.
Yet, I find it very difficult to see all this happening. Again because of the failure of so many portals: the lack of suited data sources.
Content as a Service (CaaS) is not the answer because – in my opinion – it is a technology approach to deliver content to these mashups that deals with the mechanism of ‘how to deliver’ and not with ‘what to deliver’. We must think from the business perspective first.
When you do, you will noticed that their demand is changing rapidly, In width and depth of the used data. And in the number of required sources. That’s when I start wondering… Do they understand the relationship between two data sources? Can they predict what data they need?
An example: within a bank, the account number of the first account also becomes the party number. For standing account #431256 of customer A, the party number also becomes #431256. The savings account #439867 for customer A is linked to #431256. Depending on the source, that same number can refer to 2 different entities. How does the business user make the proper choice?
Another example. One business user just needs the address data for a party. Another needs the total balance of all accounts linked to the same party. If IT prepares the ‘party data service’, what do they need to include? Everything including all possible accumulations? Does that appeal to the business users? Remember: by definition of the mashup IT must deliver before the business user has his instant need…
There is another way of looking at mashups that can help in providing more insight in the business line of thinking. Mashups can be seen as a new sort of front-end to an OLAP cube or Data-warehouse. In stead of fixed reports they now deliver dynamic on-demand output tailored to a specific and instant business need.
Would a data-warehouse, as a single data source, be the holy grail? Maybe. I think it stands a better chance then individual CaaS-alike data services. Maybe mashups need the best of portals combined with the best of data-warehousing.
As said at the beginning, I embrace the idea of mashups. I do however see too many things to overcome first, before this takes of and become successful within businesses. Time will tell.
