Putting data to work
It is not new technology, but business intelligence platforms are now understood well enough by customers to drive real business change. David Braue catches up with some BI trend-setters.
Contents  Introduction From upstart to cornerstone Paradise by the dashboard Towards the future 10 ways to better BI Case study: NT Police put on the map | ||||
The problem was that by the time the project was complete, it was meeting old business requirements and was based on an old system that no longer reflected HBF's rapidly evolving business. It also failed to get critical user buy-in.
On top of this, managers lacked unity in regards to the system's purpose, and change was so difficult to engender that the whole idea ended up stagnating.
"Warehousing is in constant change," says Mark Pereira, manager for business information services with HBF. "If you're still providing the same models a year later, you're not doing it right."
The company's second attempt at introducing BI was much more successful, with access to the company's 1.5TB data warehouse expanded through the introduction of SAS Institute Enterprise Warehouse. Report writers offer a variety of pre-fabricated business reports and pre-built OLAP (Online Analytical Processing) cubes for low-end users, while 35 "power users" get direct access to the system to run ad hoc reports based on changing business need.
Just as important as what was set up the second time around, was how it was set up. The IT division's reputation had taken a battering when the first data warehouse project turned out to be a fizzer, so for the second installation, the IT team stepped back and let the business managers take over. A management steering committee was formed to get business managers to lead the project, and business and technical analysts were tasked with supporting their needs rather than driving the project on its own as they had done in the past.
That approach increased the project's credibility significantly, says Pereira. "The business has a preconceived idea that IT is expensive and slow in doing things," he explains. "We're now a business unit reporting to the general manager outside of IT. By us not being tied up with their process, we don't get bogged down with the core legacy system and just focus on where the business is going."
"By building the trust what we were doing was helping [managers] build their business a lot quicker. [Buy-in] snowballed from there. We don't go big bang, but go down to what is the biggest business itch at the moment -- then we make sure we're scratching that itch."
Now that the BI system is working properly, that focus has turned to areas such as data mining, which has helped HBF better understand common attributes of the many claims it handles. By analysing historical information, HBF is able to quickly pick up on anomalies -- such as a rapid increase in claims from a single provider -- and identify the cause.
The company has also been able to better segment its more than 870,000 members, whose activities and characteristics are now analysed to assess and minimise each member's risk of churning. An initial customer relationship management (CRM) project built around the legacy data warehouse failed, but the company is gearing up to give CRM another go. "BI has given HBF the ability to segment and understand our membership base," says Pereira. "It has changed the way we do benefits."
Contents Introduction From upstart to cornerstone Paradise by the dashboard Towards the future 10 ways to better BI Case study: NT Police put on the map | ||||
Business intelligence was largely a pipe dream that grew out of the conviction that there must be a better way to use all the data businesses were collecting. Several years ago the continuing use of client/server applications, as well as the struggle to clean the data being loaded into data warehouses, limits on computing power available to process flexible queries, and the challenge of presenting BI in a user-friendly way, all spelled trouble for the technology. BI wins were more likely to come from relatively limited applications, with specific departments typically using BI for their own processes.
These days improvements in technology have pushed client-side issues into the background: the now ubiquitous Web interface is easy for most users, while elimination of cumbersome client-side apps has aided the integration of BI views into portals and other desktop configurations. Supported by the overall trend towards consolidating data on a single storage area network (SAN), it is safe to say that today's BI environments are more consistent, easier to use, and more flexible than ever.
More importantly, as HBF found, many business managers that have seen BI in action finally appreciate its potential for business improvement, and in many cases have dropped their antagonistic attitude towards IT's territorial incursions. This has helped companies expand BI from a tool for reporting on the past, into a tool for shaping the future -- which probably explains why, after all these years, BI still weighes in as CIOs' number-two technology priority (behind security), as stated in a recent global survey of 1300 CIOs by research firm Gartner EXP.
Much of the support, Gartner argues, is coming from the intense focus on cost reduction that is putting pressure on all CIOs. While basic quarterly reports provide a high-level view of how a company has been performing, using BI to get better visibility of an organisation's day-to-day activities -- as reflected in its core transactional database -- is now seen as a highly effective way of identifying opportunities for cost cutting.
At New Zealand's Port of Napier, BI has proved invaluable in planning the most cost-effective way to manage continuing growth. Years ago the port implemented Crystal Reports to produce a range of pre-defined regular business reports that were available for internal performance monitoring. More recently the organisation introduced BusinessObjects XI applications to extend visibility to shipping lines, the New Zealand Customs Service, quarantine officers, freight forwarders, exporters, and the port's many other stakeholders.
Giving external access to information about the containers in the port, or those expected to move through the port in coming days, lets stakeholders plan their operations much more efficiently. Customs, for example, can designate which incoming containers it wants to inspect before they've been unloaded.
This access also allows Port authorities to organise stevedoring workers to be ready at the docks to receive incoming shipments marked urgent. Overall, the port staff spend a lot less time on the phone answering queries coming in from customers, who can now get answers themselves.
BI even helps ensure timely delivery of perishable goods: alarms can be set if a container has been left on the dock too long and its contents risk rising above an unacceptable temperature. All container movements are recorded in an audit trail that can prove invaluable down the track. The system "is guarding the reputation of the exporters of New Zealand," says Steve Johansen, CIO, Port of Napier. "We are able to give people information that they hadn't even been able to think about getting before."
Real-time operational knowledge, as compared with the canned reporting of the past, has streamlined the movement of containers through the port. In the process, the system has helped port planners defer capital investments such as construction of new wharves (NZ$1 million per hectare) and additional tugboats (valued at NZ$10 million each).
"Over the past five years we have doubled in size, but [we] have been able to do that without taking on enormous amounts of staff," says Johansen. "By concentrating on technology, we've been able to really look at the numbers and wring every last bit out of what we have now."
Supporting the decision-making process of so many stakeholders forced the Port of Napier to spend considerable time ensuring its data was clean and that its real-time information feeds were delivering accurate results. The effort appears to have worked. "Internally we have people asking where the data is coming from, but I've never had an Internet-using customer query the data," Johansen says.
Contents Introduction From upstart to cornerstone Paradise by the dashboard Towards the future 10 ways to better BI Case study: NT Police put on the map | ||||
Business intelligence vendors are now revisiting customer applications to help them make more use out of investments. In most cases this means converting BI from a reporting into an operational tool.
Hastings Deering, a AU$1 billion distributor of heavy earth-moving equipment based in Queensland, has used its Cognos BI platform to extend its demand forecasting from 12 to 18, and then 24 months.
Such forward views are essential in an industry where it can be 18 months from customer order to delivery of the custom-built multimillion-dollar machines. BI also lets the company's more than 100 BI users track their progress against industry trends, and gives managers views of overall business key performance indicators (KPIs) that were unavailable in the company's previous, informationally fragmented environment.
The company's data model has undergone numerous changes, ranging from integration of outside data sources (such as market indicators, and information about missed tenders) to a logical division that split its 70 million-cell SQL Server database into three separate models.
"We slice and dice that data extensively," says group IT manager John Birch. "Now that we've gone into forecasting, my thought is that we can move it further and deeper into the organisation as well. Financial measures like cash cycle, progress and day sales are good, and we also look at more esoteric ones like injuries, [and] how many staff appraisals have been done."
Port of Napier is using BI dashboards -- which give an easy-to-read view of KPIs -- to maintain a continuous view of the progress of ongoing projects. And RailCorp, the statutory authority responsible for train services throughout NSW, recently announced it would use Cognos BI tools to monitor corporate KPIs in real time in a project that forms the second phase of RailCorp's CPM One strategy.
The general lack of momentum for BI made corporate performance management (CPM), difficult for many organisations to implement in the past. This led to many companies pursuing balanced scorecard systems that attempted to combine various BI reports into an overall picture of the business.
However, separate balanced scorecard tools proved difficult to implement because of the sheer difficulty of tying BI data back to operational features. For this reason, the continuing maturation of BI tools, and managers' growing enthusiasm about its possibilities, have now made it more than possible for companies to use BI as a real-time monitor of business performance.
"Most organisations realise that there's a huge amount of value in the data they've got," says Jeremy Ranking, solutions manager for business intelligence with SAS Australia. "What we've found is that very few organisations are actually able to get that information to the right people in a usable format at the appropriate time. We're trying to encourage people to go beyond what they're currently doing with BI to understand outcomes and how it's going to affect business."
The difference in this approach is the availability of real-time information, which allows managers to strike while the iron is still hot, so to speak. For this purpose dashboards can be immensely valuable, highlighting KPIs to which managers need to respond. You can still offer links to let them drill down into the data, but a well-designed BI system should always let managers know at a glance whether the operational issues that matter to them are within range.
Real-time information does not have to be limited to internal data. A commodities trader, for example, might find value in testing the historical correlation between changes in commodity prices and the volume of incoming customer calls. This information could then be used to add or decommission virtual call centre operators as those commodities change price throughout a day. In this scenario predicted demand might also be delivered to managers as a dial that would climb to the right as, for example, iron ore prices drop and a glut of customers calls to organise futures contracts.
Contents Introduction From upstart to cornerstone Paradise by the dashboard Towards the future 10 ways to better BI Case study: NT Police put on the map | ||||
Tailoring dashboard-based views for many different types of users will provide relevant information, and nothing more, to the people that need it. In the long term, this customisation should finally turn BI from a tool of the analytical elite into something that is used to varying degrees by nearly everybody that is in the business.
To meet this goal, vendors are rushing to the concept of dashboards and the ease of use that it enables. To raise its profile against Cognos, Business Objects, Hyperion, SAS Institute, and other standalone BI vendors. For example, mid-market ERP supplier Epicor is bulking up its own offering to include dashboards that Charles Hamilton, ANZ regional director with Epicor Software, believes will deliver BI for the masses. "We're trying to get BI used by the entire business rather than just being used by key head office managers and a few analysts and accountants," Hamilton says.
Since all enterprise data that the BI tools need is already present in the system, utilising built-in analytics may be preferable for many customers to integrating a standalone BI tool -- particularly in the resource-constrained mid market.
Such customers may also benefit from new approaches to data warehousing and business intelligence. Upstart Netezza, for example, is marketing a BI "appliance", built on the open-source Postgres SQL database, that speeds up data queries by using a massively parallel computing architecture. Netezza's 1TB entry-level model runs on 28 RISC CPUs and scales up to a 27TB system with 672 CPUs, all of which work in concert to churn through massive volumes of data.
One US customer, says to Asia-Pacific director of marketing Mike Kearney, was able to cut its regular BI data processing runs from 23 hours to 50 minutes simply by adding such an appliance. Such speed improvements are extremely valuable in transforming BI from a retrospective tool into a real-time performance management tool.
Other firms have tried to add new value to BI by introducing new data sets in a meaningful way. Australian firm Integeo has bolstered conventional BI with Map Intelligence, a tool that automatically extrapolates spatial information such as addresses to present BI reports with a geographical context (see case study). Geographically based views can be invaluable in spotting clusters of other indicators that might otherwise go unnoticed, but such integration has previously required complex integration with specialised geographical information systems (GISes) that most companies do not have.
It may not always have looked that way, but the growing number of successful business intelligence deployments that can be seen suggests that BI is finally overdelivering and underpromising, rather than vice versa as in the past. Having climbed out of the trough of disillusionment -- both as a technology and as a management concept -- it's now time for managers to drive new uses for BI that will allow them to constrain costs and identify new business opportunities on an ongoing basis.
Contents Introduction From upstart to cornerstone Paradise by the dashboard Towards the future 10 ways to better BI Case study: NT Police put on the map | ||||
Business intelligence is no longer a black art -- for most of us. Here are a few tips you can use to help make sure your spell works.
- BI is not an IT project. Too many companies forgot this one in the first wave of BI, creating miniature turf wars as managers clung to their data and IT tries to introduce a new world order. Forget all that: BI is an operational tool that should be supported and used by business managers to support their requirements.
- Recognise different user needs. Not everyone needs a full-featured user interface; most employees may be more than happy with a few well-chosen dashboards to tell them what they need to know. Pumping BI through a user-customised portal is a great way to shed complexity.
- You work in real time; so should your data. Reporting after the fact is good for writing annual reports, but not for managing the business. Focus your BI system on gathering and analysing operational data in real time, so you can get the most benefit.
- Establish a context. Even the best business information can lose value in a contextual void. If information about customers, geographical details or even tide charts are relevant to your business, build them into your model.
- Empower customer service agents. If anybody needs real-time information, it's your customer service representatives. If they can get relevant historical information -- extrapolated to projections about possible customer churn and other future outcomes -- they will be able to work more proactively to service and keep customers that might otherwise switch to your competitors.
- Scrub your data. Business intelligence isn't worth anything if it's generated from unclean data -- often a symptom of system proliferation during the late 1990s. Centralise and clean your data regularly -- and redesign application screens to minimise errors -- so that you know all intelligence coming from the system is trustworthy.
- Customers and partners love BI. If you're spending too much time servicing enquiries from customers and partners, consider giving them limited access to your BI platform so they can get access to real-time reports themselves.
- Be proactive. BI is great for telling you what has happened to your business in the past, but it's even better for predicting -- and managing -- the future. If your BI system detects an impending problem with supply, help customers understand how to work around it in advance. Helping customers solve their own problems is a great way to keep them on your side.
- You may already be there. Early BI tools required integration of many disparate data sources with a standalone BI application. These days, however, ERP vendors have taken significant strides forward -- and, since an ERP system by definition has all the data in the one place, it can be relatively simple to add BI to your environment.
- Think laterally. Your business operates in the real world, and so should your BI system. Look for external data sources -- such as tenders published online -- that can be integrated with your own information to give your salespeople, customers, partners, and managers a context for BI.
Contents Introduction From upstart to cornerstone Paradise by the dashboard Towards the future 10 ways to better BI Case study: NT Police put on the map | ||||
Getting better business information is one thing, but without context, that information can often lose much of its impact. For the Northern Territory Police Fire and Emergency Services (NTPFES), overlaying information about crime trends onto geographical maps has proved invaluable in better understanding local trends in crime across the territory.
The service previously implemented Crystal Reports to generate regular reports from its central case management system (CMS), but last year upgraded to the Hyperion Intelligence BI tool. The addition of BI made it much easier to track individual types of crimes, or individual offenders, and to correlate factors such as a known offender's modus operandi to assist with generating potential leads on suspects.
"The goal was to try and do a bit more with the data and provide it in a more user friendly manner," says John Weippert, director of ICT with NTPFES. "Intelligence-led policing is enabling us to use the intelligence we have to do better policing, target offenders, target crime hotspots, and to look at how our strategies are affecting crime."
Although Hyperion Intelligence improved analysis of information from the CMS, something was still missing: trends were apparent, but it was hard to relate them to the real-world issues that the police faced every day on the streets.
The service found the answer in Integeo's Map Intelligence product, which sucks data from BI platforms and automatically overlays that information on geographical maps. A concentration of crimes around a particular shopping centre, for example, becomes blatantly obvious when geographical information is added.
Links with other external systems, such as the NT Department of Justice's Justice Information System, provide further context: for example, the system might pinpoint a particular known offender as a possible suspect, but that offender would be excluded if he was known by the Department of Justice to have been incarcerated at the time.
The ease with which Map Intelligence integrated with the BI system led it to quickly gain popularity at all levels. More than 200 of the NTPFES's 1000 officers are using the system regularly -- including high-level tasking and co-ordination groups, which use it daily to identify which areas have been experiencing higher rates of particular crimes, then shape their strategic responses accordingly.
Strong success within the police branch of the NTPFES has driven Weippert to talk with fire authorities about how the geographical BI system might help their own operations improve their strategies. NTPFES is also looking into building a central data warehouse that would enable aggregation of even more data.
"People are using the system more extensively now, and we're getting more and more information," says Weippert. "People are wanting more out of it, and it's become an essential tool for us. We could never go back now."
This article was first published in Technology & Business magazine.
Click here for subscription information.