Operational business intelligence promises a faster
route to the complete picture than traditional BI, allowing quicker
and more effective decisions at all levels of the organisation.
Danny Bradbury investigates
Imagine this scenario: you are running the IT department for a
large retailer, and a truck turns up at your warehouse with a load
of teddy bears. The warehouse manager has nowhere to put them, and
does not have the people to deal with them. He has to make a
choice: send them back to the supplier, or route them to another
warehouse - and if he does reroute them, to which warehouse?
Traditionally, such decisions might be made after a lot of phone
calls or, worse still, off the cuff without any thought - just send
them back to the supplier, perhaps incurring a penalty. Instead,
wouldn't it be useful to combine historical data with transactional
data to make a more informed decision? Pulling in data about your
history with this supplier, stock levels in different warehouses
around the country, and regional sales information about particular
product types could help your warehouse manager make a better
decision, but only if it can be presented simply.
Making operational decisions based on complicated data gleaned
from multiple sources is what real-time business intelligence is
supposed to be about. Business intelligence supplier Information
Builders has coined the term "operational business intelligence"
(OBI) to describe it. The difference between traditional business
intelligence and OBI is that whereas the former focuses on
historical data in a datawarehouse to perform long-term analytics,
the latter draws data from transactional systems to better inform
short-term decisions, sometimes on a minute-by-minute basis.
The OBI concept has been around for a while, surfacing in
different companies at different times. But the definition is
nebulous, says Craig McCollum, vice-president for business
solutions strategy at Microsoft, who worked for various business
intelligence suppliers for at least 15 years before joining
Microsoft.
Some people concentrate on delivering real-time information to
operational workers, while others think that daily or weekly
business intelligence reports, delivered to line managers to help
monitor performance, qualifies under the term.
The best way to place OBI is at one end of a business
intelligence spectrum. At the more traditional end sits analytical
BI, where small numbers of managers make strategic decisions based
on analysing historical data, crunched by the datawarehouse.
More regularly updated BI might be used to crunch weekly sales
figures by region and keep your team on track, or may be used to
create a weekly scorecard for your call centre.
Real-time or near-real-time BI sits at the far end of the
spectrum, and delivers data to a larger number of people, such as
call centre operatives and their supervisors. The closer you get to
that end of the spectrum, the more it qualifies as OBI.
Eric Rogge, vice-president and research director for business
intelligence and analytics at Ventana Research, is a strong
advocate of the idea. "OLAP was originally there to support
analysts and that isn't what operations is about," says Rogge. "The
primary aspect of that is more about alerting and prioritisation,
and less about how many angels dance on the head of a pin or some
other arcane analytics."
Not all analysts are as supportive. Real-time OBI does not
involve traditional BI tools, according to consultant Nigel Pendse.
Companies running systems that require such tools - such as railway
control systems and financial trading systems - already have this
real-time operational reporting hard-wired in. "They are being very
patronising about the people who have built these operational
systems," he says, adding that those applications that don't
already have such facilities built in are not going to need them in
the future.
Anyway, says Charlie Davies, group marketing manager of IT
services company Touchstone, companies have been able to generate
real-time reports for years with products like Crystal Reports, so
what's new?
Integrating the data into the line of business applications
behind the scenes is one key part of the OBI argument. The idea is
that the data should appear in the context of the job that the
operative is doing, making the data actionable, so the employee can
make a business decision there and then. "I want this to be the
type of application that pops up on the warehouse guy's computer
where he has the plastic cover on the keyboard because his hands
are so dirty. That's the kind of thing that people are looking
for," says Rogge.
Although application suppliers in some vertical markets may have
that hard-wired into their systems, such bespoke development is
expensive and difficult. BI suppliers want to muscle in on that
business, working with partners to create new applications that
come with live feeds from diverse sources.
Few people, if any, are doing real-time BI delivered into line
of business applications yet. Nevertheless, people are closer along
the spectrum, with some providing report-based information on an
hourly basis, designed for operational supervisors to act upon.
Stephen Segel, finance director for recruitment firm Blue Group
International, says that second-by-second information is not
relevant at all in his business, but hourly reports are of use in
terms of financial control. "We reflect back to operations what
timesheets we have received from the organisation. That goes to
every single consultant in this business (about 100 people). They
can see what timesheets their candidates have received, so they can
phone up immediately to ensure that they get their timesheets
through," he says.
As companies move further along the spectrum from analytical to
operational business intelligence, they will encounter business
performance management, which ties closely into operational
reporting. Applix, for example, sells the TM1 scorecard reporting
system, providing companies with snapshots of how their business is
doing.
Mark Bodger, business information controller at Littlewoods,
uses it to produce a weekly data update for operational managers,
and mixes product information, warehousing stock data and sales
figures that help them to plan their resource allocation. "Let's
say we have two warehouse locations, one for fashion and one for
electronics," he says. "They need to know what the demand will be
week by week in unitary terms for those two warehouses. Then they
could work out their manpower requirements."
Littlewoods is reducing the frequency of data reporting in
certain areas to move it further along the business intelligence
spectrum. "There are developments going on in the call centres at
the moment to do that," he says. "We want data like the number of
calls per minute per customer service adviser. That is a detailed,
quarter-of-an-hourly snapshot of data. It is operational planning
at its finest." That type of application might enable a call centre
supervisor to spot problems with an operative who was not properly
trained, or perhaps with a group of operatives who were taking
longer to complete calls because they could not find answers to a
particular query.
The question is, how do companies move towards real-time BI
without affecting their operational systems? If a transactional
system such as a customer order database is under a heavy load,
hitting it with extra queries to provide call-centre operators with
decision support could slow it down.
Information Builders, eager to sell its integration and
messaging tool iWay, suggests you do it by building business tools
into your message bus. Getting an integration tool to monitor
different back-end systems as they message each other lets you
listen passively to what is happening in your applications without
having to go in and directly query data.
"It sits, listens, detects and, depending on what it hears, can
do anything - decision analysis or simple alerts," says Dennis
Allen, vice-president EMEA, for the company. "It is a more focused
BI approach in that it does not just run reports." So it could
trigger an alert for the call-centre supervisor if, for example,
certain conditions were met concurrently in different applications,
such as the customer database, ordering system, product inventory
and call-centre routing system.
Listening on the message bus also enables you to constantly
update a datawarehouse using a trickle-feed mechanism, rather than
performing a batch update and rebuilding at the end of every day,
contends Cliff Longman, chief technical officer of datawarehousing
company Kalido. This can be useful if you want to blend current
transactional data with historical information to enable people to
make better decisions.
Companies like Information Builders will have a hard time
getting companies to the point where they can integrate data from
multiple sources in real-time and slice and dice the data for
decision-support purposes. Nevertheless, companies continue to
squeeze more value out of their data by pulling it from live
systems into simplified reporting tools or line-of-business
applications on an hourly, daily or weekly basis, to help larger
numbers of staff further down the company hierarchy.
Case study: De Montfort University cracks student data
problem
In the academic world, where students are your customers, IT
managers face a complex reporting problem. At De Montfort
University, John Shelton has to keep 23,500 students happy every
day.
Shelton, who is head of the university's business systems group,
has to populate the Managed Learning Environment (MLE), a student
portal that connects to multiple systems to serve data, including
grades and class information. Students can log into the system and
see grades when they become available, with each student's profile
on the system generally updated on a 24-hour basis. But Shelton
also has to prepare complex assessment reports for each student in
June and October which are submitted to the review board. These
reports are used to evaluate each student's future options.
"They could be entering marks until the 11th hour of the day
before the board sits," says Shelton, describing the pressure the
IT department was under to deliver accurate, up-to-date data. Now
he uses Information Builders' iWay software, its messaging and
integration tool that extracts data from applications and crunches
the numbers at the back-end before delivering them to the reporting
tool. "We use the extraction, transaction and loading (ETL) tool to
pull out information from the live student records system into a
datamart, which is then used by the MLE."
Using the messaging and integration software lets Shelton serve
both the review board and the students by extracting a subset of
information from the university's operational systems without
having to hit the live database with huge queries all the time.
This helps to maintain transactional performance while keeping
reporting data up to date.