Spatial Business Intelligence

Business intelligence tools have focused on producing detailed data and information and presenting it in the form of reports for decision makers within organisations. Typically these reports have come in the form of alpha-numeric information backed up by the occasional pie chart or graphical table.

That is fine for reports that concern a departmental process or a specific internal function, but if the information concerned covers a wider geographical spread - for example, a comparison of sales performance against targets across three continents - then the end result is all too likely to resemble alphabetti spaghetti.

The data that is supposed to be providing you with enabling information to make quick and sensible business decisions is being presented in such a format as to be either deeply off-putting or seen as something that itself needs to be translated and converted into something understandable.

But increasingly vendors are looking to build in functionality more typically associated with Geographic Information Systems (GIS) vendors, to make the presentation of BI data more user friendly and easy to understand by those without a background in statistical analysis. While many business people will be put off by the apparent demands of having to plough through reams and reams of numbers in a tabular form, reading a map with some bright colours on it is an entirely different prospect.

"Spatial information brings a rich view of fast-breaking trends that impact a business," says Henry Morris, group VP for applications and information access at analyst firm IDC. "Unfortunately analytics and GIS have emerged separately and have served different constituencies."

The business-to-business (B2B) applications vendors got in on the act several years ago, with location-related functionality being at the centre of development initiatives by a number of leading enterprise resource planning (ERP), supply chain management (SCM) and customer relationship management (CRM) providers. In addition, wireless service and applications providers have, by necessity, made location and geographic information management a key element of their product development.

For BI vendors, in-built location functionality has taken a little longer to come to pass, but is now appearing in mainstream products. For example, users of SAS can leverage ESRI technology to perform spatial analysis, while SAP and ESRI also have a commercial and technological relationship.

Business Objects is also working with ESRI to provide an ability to access geographic data and create new Crystal Reports with detailed map information from a web-based ESRI ArcIMS report viewer, while Information Builders is also working with the company to integrate ESRI's geographic mapping software with Information Builders' reporting software.

"Valuable insight is commonly hidden in the sheer volume of operational data and business information generated by organisations," says Steve Trammell of ESRI. "Forward-looking organisations realise that adding the dimension of geographic analysis to sophisticated business intelligence applications results in better insight, more informed decisions, and superior communication."

This merging of technologies has been a long time coming, but in reality the combination makes perfect sense. It is pertinent to recall that even the most basic data warehousing technology does in fact contain spatial elements. Every table is likely to have some geographic element, be it a mailing address or even just a country of origin.

"Much of the information stored in company databases today has some sort of geographic content to it, whether that is related to an address or a region or a post code," says Renaud Besnard, product solution marketing manager for SQL Server 2005 at Microsoft. "Unfortunately, this information is typically only viewed through a spreadsheet, a table or a graph. Rarely is it analysed in its most logical format: on a map.

"With solutions such as ours with SQL and MapPoint, businesses have an ability to visualise their information in a new way and get a new perspective on their performance, their competition and their customers. By understanding the location component of common business data, such as sales or customer lists, they can make more informed decisions. This viewpoint helps to identify trends or business performance issues that may otherwise have been overlooked."

So what is the basic principle? Geographical mapping of business data provides a more intuitive way for processing and comprehending certain business information, theoretically enhancing the ability to analyse critical trends and patterns. Users can aggregate real-time information from disparate databases and spreadsheets, build individual ad hoc queries and formatted reports, and then instantly transform their results into maps. The application of this can be seen across multiple business sectors from sales and marketing and CRM through retail and manufacturing to law enforcement and government analysis.

"Some excellent examples of geographical data that are useful to any business include sales data, competitor locations, customer lists and service records," says Microsoft's Besnard. "By adding context to your data in this way, a business can identify whether geographic or demographic variables may be affecting sales of a particular product.

"They can examine whether their business could become more efficient by reassigning sales or delivery territories. Or they can analyse different areas to determine whether they appear to be good prospects for expanding the business."

Government agencies have long been big users of GIS-related technologies. Worldwide spending on GIS software in 2004 was $1.8bn, of which $544m was spent by US federal, state and local government agencies, according to IDC. State, local and provincial governments account for 65% of government spending worldwide, meanwhile, and federal governments spend the rest according to Daratech.

Prior to the recent devastation caused by Hurricane Katrina, the Louisiana Social Services Department was a text-book example of how such combined BI/GIS systems could work in practice. The department wasted hundreds of man hours and huge amounts of money chasing down food stamp fraud, ploughing through thousands of transactions and receipts all produced in paper-based forms that had to be individually considered.

The solution was a web-based system that can access transactional data and display it in a geographical context, so that department officials had that same information in their hands in 30 minutes and could review food stamp activity in a designated region or neighbourhood. The ability to look at material in a geographic context also gave investigators a fresh way to view data and reveals patterns and anomalies that might not be apparent in tabular data.

This ability to zoom in is of obvious advantage to crime management organisations such as the police or homeland security organisations. "The Dutch police have an application where they can start with a map of the country, then zoom in to the 26 police districts," says Ad Voogt, Cognos European senior VP.

"All the districts are measured in the same way and get assigned additional resources based on how well or how badly they perform on the scorecard. What you can then see graphically on the map is where the crime rate is going up or down. You can see it at a regional level, then at a city level, then down to specific parts of that city.

"But the BI element really kicks in with the historical data. The scorecard tends to be a snapshot of the current crime statistics, but the police are also going to be interested in how things have changed over the previous 12 months, for example. You can run that as well using the same parameters and see how the colours change on your map."

Cognos recently announced an alliance to jointly market and sell MapInfo's location intelligence technology with its own Cognos 8 BI product, adding the ability to geographically query data. "The goal of business intelligence is to give users quick and easy access to the information they need to make strategic decisions quickly. Location intelligence extends this value for customers by uncovering critical location-relevant relationships in the data," commented Don Campbell, Cognos VP of product innovation and technology.

Critical to the success of combining BI and GIS functionality is the geocoding of data. That is the name for the relatively simple process of taking a geographic data attribute - such as an address - and translating it into latitude and longitude. This renders the alphanumeric information as a grid reference that can in turn be rendered as a graphic image.

BI vendors have spent millions of dollars on research and development to add new ways of 'slicing and dicing' data. One view of a set of data is no longer acceptable. Even the most basic reporting tools will allow some degree of drilling down or alternative analysis.

In the same way, location-based applications such as MultiMap or MapInfo will allow users to drill down or zoom in or zoom out of a particular screen. Put a post code into MultiMap and it will show you exactly where that address is within the confines of a set of geographic parameters. But it would be useful to know where that set of locations is in the wider scheme of things. Is it near to a mainline station or the town hall? So the ability is built in to zoom out and get a bigger picture to put the location you seek in a wider geographic context.

A combined BI and GIS solution provides the best of both worlds. The EMEA region's sales performance against targets can be rendered in tabular form or can be presented as a map of Europe with different colours for different countries, high performing regions, low performing regions and so on. After this initial view, the user can then decide to put EMEA's performance into a global context and zoom out to a view of all international operations on a global map.

Alternatively they might decide that there are clearly local problems bringing down the overall EMEA performance, and choose to drill down to a country-by-country level. Drilling down to the UK, they could then choose to zoom in on England, then the south of England, then the Reading area, and so on. In this way, they can apply the same performance parameters to a global, regional, countrywide or area specific view of the data, all presented in a simple, graphical snapshot that enables rapid absorption of the data.

"People have been looking for this sort of functionality for many years," says Donald MacCormick, VP of analytical applications at Business Objects. "They want to be able to take data and put it into a geographical context to make it easier to understand. From our point of view, geography is just another dimension that we support, another view of the data.

"If you put data on a map, then you get something different. You provide a level of insight that graphs and charts are simply never going to provide. It's amazing how much more insight you can get when it's laid out on a map for you. I can't think of a single area of business where they wouldn't want to be able to do this."

CBR Opinion

We all know about maps, but we do not all understand statistical analysis techniques. Faced with an array of tables, pie charts and flow diagrams relating to the regional sales performance of an organisation, many of us might be tempted to admire the amount of detail that has been provided, then run a mile from using any of it. On the other hand, if the data was presented on a map with a colour coded index to explain broadly what the major trends and market movements were, that would be a very different matter. The coming together of GIS and BI is a welcome move that offers better prospects for location-based data analysis.