Guinness UDV has rationalised the use of online analytical
processing tools developed piecemeal around the business to build a
common data model.
Stories of companies embracing the e-business model are getting old
hat now and many can point to a long, if somewhat chequered,
e-history. But the story does not end at the point where the
company embraces the Internet as a viable business medium and
develops an e-strategy. There is still work to do before it can
live happily ever after.
Grappling with the effects of e-business and adapting to the
changes it imposes is an ongoing process. One major task is
managing the quantity of data that is generated by Web
transactions. For alcoholic beverage firm Guinness UDV this led to
the implementation of a common data model based on business
intelligence software.
Guinness UDV was formed last year when parent company Diageo merged
its two alcoholic drink interests, Guinness and United Distillers
& Vintners. Both had been using business intelligence for some
time but, according to Guinness UDV's data technologies director
Steven Sharpe, before 2000, "It tended to be tackled on a fairly
market-specific way, with each market doing its own thing."
Compounding this problem was the fact that, following a series of
mergers, Guinness UDV found that it was using a variety of business
intelligence offerings, including 17 different online analytical
processing and reporting tools.
Sharpe says the company realised it needed to set up a "more
standardised, common approach" across all of its operations. "What
we really needed was a data management infrastructure of which the
reporting component was very important but alone would not
succeed," he says.
At the beginning of 2000, the company began work to develop a
common data model, taking an approach that Sharpe says "you might
call datawarehousing". Its initial aims were to allow the
integration of data from multiple sources and to provide an
aggregated view of that data. For this, the group turned to
business intelligence software from Cognos.
As Sharpe explains, Guinness UDV already had a large number of
Cognos licences deployed globally. Cognos' pricing and support
models were appropriate and it had a good approach to managing a
relationship.
"It was the right answer for us at the time and it continues to be
the right answer," says Sharpe. "Cognos supports business change."
Once Guinness UDV had consensus on this approach, it implemented
four pilot projects covering the key areas of its business.
Datawarehouse models were set up for sales, supply chain,
manufacturing and finance - each using Cognos business intelligence
software at the front-end. These pilots were run in parallel and
lasted an average of nine months.
Based on the results of the pilots the company came up with a
single standardised answer. Sharpe says that solution is now being
deployed in all the company's large and medium markets and the
group is considering whether it is appropriate for its smaller
markets.
Sharpe claims there were no major teething problems but admits
there were issues that needed to be resolved - the key one being
data quality. He says the success of business intelligence
solutions often relies on the quality of the data so any issues
need to be addressed fully and methodically.
"You can have the best technical implementation of business
intelligence, but unless you have good quality data you won't have
a successful solution," he says.
"Ideally, you need to change the structure and culture of an
organisation so that data quality is managed at source," says
Sharpe.
But cultures don't change overnight. And although Sharpe says that
using data cleansing tools is not the answer, as they only offer
temporary solutions, he admits that in practice they are needed to
complement this approach. No matter how good your business
processes are, there will always be some failures, he says, and
"data quality is a process failure issue".
In other words, companies should manage data quality at source but
they should also expect problems to occur. These can be resolved
using data quality tools. Reporting tools can then be used to
identify which processes are failing and effectively "close the
loop".
Firms need to realise that "data ownership is not an IT but a
business issue", says Sharpe. However, getting the business side of
the company to take ownership of data is not easy, not least
because "data is not sexy - it is pretty boring stuff, quite
honestly".
Sharpe believes a key reason for the success of Guinness UDV's
content data model and its use of business intelligence software is
the work it did at the beginning to make sure end-users would buy
in to the system. "At the end of the day, you have to sell business
intelligence solutions to your organisation," he says. "Buy-in at
the start is so important."
Guinness did this in a number of ways. It ran a series of global
workshops and identified interested parties - typically business
intelligence and datawarehouse managers - and got them involved in
shaping the model for the business. A 15-strong "global application
team" was established at the beginning of 2001 to oversee the
project. "It maintained a standardised approach to maximise return
on investment; minimise time to deployment and minimise
cost."
But this standardised approach is not set in stone. Sharpe says
that flexibility has been built into the model to allow for
country-specific differences, such as pricing models. The company
also plans to make enhancements to the underlying common data model
to allow it to adopt a more flexible approach to hierarchy
management and incorporate hash key algorithms - which are used for
creating digital signatures.
Although it was originally developed as a logical data model, the
common data model has become part of the global IT strategy and is
now the standard Guinness UDV information management solution.
Sharpe says that demand among its users is "far greater than we'd
anticipated, so we must be doing something right" - a point he
illustrates by quoting an Australian user who likened the
transition to a move "from shithouse to penthouse".
Software in the public house
Problem
Guinness UDV found it was using many different business
intelligence and reporting tools to cope with all the data created
by e-business initiatives and wanted to create a standardised
approach to data management
Solution
The company created a common data model,
incorporating software from business intelligence firm Cognos.