LI meets the enterprise
Parallel development - LI and BI technologies have evolved in parallel universes. LI is the new term for what used to be called GIS. GIS has traditionally been deemed as highly 'proprietary' technology used and targeted at a niche. It has become the Cinderella technology that hasn't been invited to the ball and has yet to be accepted as part of the fabric of mainstream ERP, CRM, SCM and other enterprise applications. And there it has more or less remained. LI has been designed as a desktop application, not for today's high-volume Web applications, and carries the baggage of development complexity and specialized front-end. As such LI technology has been siloed and has not been brought to the attention of the wider organization.
In contrast BI technology has advanced much further, with deployment complexity and cost dropping, and usability and executive interest rising. BI is now a top spending priority for CIOs and is showing all the hallmarks of a mature and successful technology - i.e. consolidation. However, it still hasn't fulfilled its potential, with less than 20% of corporate users using it regularly. Marrying BI to the power of freely and widely available services like Google Maps with BI promises to solve the 'last mile' challenge in BI and get more people involved in analysis of the business.
Slow convergence -This marriage has never really taken off, despite their complementary nature. Even though estimates show over 80% of corporate data has some kind of geographic element attached to it, LI has thus far remained underutilized in corporate environments. Most enterprises have only paid lip service to the potential of adding LI to their enterprise applications and processes. For example, GIS has obvious benefits for micro-marketing optimization - using demographic data to sharpen up customer segmentation for more targeted campaigns. Yet many companies still continue to spend huge amounts of money simply blanketing offers based across postal/ZIP codes. Part of the reason is cost of course. Getting hold of detailed demographics relating to socio-economic groups, education, house prices, etc. doesn't come cheap. Plus the data is extremely volatile and slow to acquire, making it a significant barrier to successful micro-marketing.
Momentum in the BI community to incorporate a location dimension in analysis has been equally slow. Many BI vendors have taken tentative steps to bolt their tools and applications to specialized GIS systems like MapInfo (acquired by Pitney Bowes in March 2007) and ESRI. But uptake has been painfully slow, not least due to clunky integration and poor joint-marketing. Consider also, for example, that large BI systems integrators like IBM Global Services or Accenture, who employ tens of thousands of consultants, still don't have any location-specific competencies - as yet at least. These vendors are missing out on a valuable opportunity to put geospatial information into the hands of corporate decision makers.
Understanding the value of maps - So what's held companies and vendors back? The problem doesn't stem from an understanding of the value of geographic data visualizations. Maps have been around for ages and are everywhere. They're established as part of the fabric of everyday lives, as witnessed by the popularity of Google and Yahoo Maps.
Map-like visualizations also help users to intuitively gain quick insights into large stores of data that aren't easily interpreted in traditional tabular reports. Put spatial data onto a spreadsheet and you get a bunch of numbers. Put it on a map and you get an immediate trend. Add the power of BI analysis and reporting into the mix - i.e. the "how" and "why" piece - and you get a powerful tool for business management. Maps not only work with structured BI data, but also unstructured information as well. For example, connecting Google Earth to a hyperlinked Documentum document database provides a map-driven alternative to text-based search.
LI also has a role to play in the uptake and usage of visualization technologies, which arguably have a bigger role to play in bringing BI data and applications to a wider BI audience. So expect LI and visualization technologies like 3-D graphics, animation, heat maps and radars to come together in interesting ways to build more intuitive and insightful BI applications.
Maps and LI are now becoming a part of business life as well - over 350 million downloads of Google Earth certainly seems to be resonating with corporate CIOs and IT directors. Slowly but surely the "where" is becoming an important part of business information - useful in areas like strategic planning, territory definition, marketing analysis to answer classic questions like:
- where are my most profitable clients
- what are the underserved markets for my products and services and how can I expand into them
- where are my competitors?
Social and organizational awareness of the value of map-like visualizations has also advanced. The early attempts to integrate specialized GIS tools like MapInfo and ESRI with BI hasn't spurred widespread adoption for the simple reason that they didn't focus on solving specific business problems. Maps for maps sake - because they are fun and cerebral - isn't good enough to stir widespread corporate interest. Business users don't want to see a map and make a subjective decision based on it. Rather than building a "corporate GIS," they want a more application-centric view of location-data presented in that map. Hence the map becomes just another enabling technology for delivering a decision.
Other drivers - Add to that an increasing number of location-focused data sources - RFID, spatial databases, geo-coded documents, etc - that are multiplying almost daily and advances in wireless and remote sensor technologies that get data to and from virtually anywhere and you have a promising set of business and technical drivers for pulling LI technologies closer to enterprise data management for valuable business applications.
Ready for prime time
Web 2.0 and mashups - There are several factors that are edging LI into the enterprise. The most significant is the emergence of consumer-led BI platforms from Yahoo, Google and Microsoft, which have exposed more users to geospatial technologies, and at a low cost. The widespread adoption and acceptance of BI dashboards (graphical in nature) has also played a part in edging LI into the enterprise. At the same time Web 2.0 applications for the masses have also made location technologies more prevalent on the Web. And subsequent mashups - even narrowly defined as Google Maps on Craigslist - have raised more user awareness of LI than all the stuffy marketing around GIS for the past 20 years.
In particular, the advent of Web services and mashups has also simplified what used to be complex development of GIS with modular plug-and-play software that is cheaper to implement. That's now happening in the BI market with a next-generation approach of LI-enabled analytic applications. Mashups from Information Builders (WebFocus for Google Maps) and Pentaho (an open source BI firm) and Tibco Spotfire (demoing a prototype for pharmaceutical sales reps using Microsoft Virtual Earth) weld internal BI information with public web sources. Expect similar mashups from IBM Cognos, SAP/Business Objects and Oracle/Hyperion to follow this year. The key is that they virtually eliminate IT involvement and the huge costs normally associated with trying to GIS-enable BI platforms. Admittedly some companies will have no qualms about geo-enabling their BI environments. But many don't necessarily want to build or own the LI technology stack and are looking for it to be delivered as a service. The mashup option will appeal to them.
The rub is to integrate these mashups without reworking or compromising the existing IT and data infrastructures in place. For that reason they still tend to be created by IT on the server-side data access layer, rather the end-user tools layer. Additionally mashups don't mean the back-end issues around data management (integration and quality) go away. They are also at the mercy of the underlying mapping service, which (although relatively stable) isn't immune to downtime.
Laying a data foundation - Data warehousing firms like Oracle, IBM and Microsoft have all bought BI companies and are now working to provide the underlying database infrastructure for bringing location into their respective environments. Oracle and IBM have already location-enabled their respective databases (Oracle Spatial and DB2 Spatial Extender) so that location capabilities can be filtered up through the applications server to the ERP and BI application. Microsoft might have come late into the game of providing spatial capabilities in SQL Server (it is planned for the forthcoming 2008 release later this year), but it has by no means missed the boat and still has a chance to come in late and make a big play on the corporate desktop it already dominates.
The challenge for all three database vendors is how to make their spatial back-ends resonate in a world that seems more enamoured by the mashup route.
SOA - LI integration at an enterprise level will also depend on increasing adoption of service-oriented architecture (SOA). In BI environments SOA can help by enabling easy exchange of data between data warehouses and BI tools and applications, often in near realtime. The same services-based approach can be used to tap spatial databases, adding (for example) delivery route information to their analysis to map out late-delivery complaints and their causes - perhaps due to road construction for example.
Hurdles to overcome
LI must address business problems - Location-enabling enterprise applications like BI is clearly a white-space opportunity for vendors right now. However, LI must also overcome some technical challenges to break into the enterprise proper.
The first is making sure that LI adheres to standard IT infrastructure already in enterprises. It means moving away from the proprietary GIS front-ends of yesteryear and leveraging familiar interfaces. Using a proprietary LI to analyse geographic data can be a huge roadblock to adoption and use - in effect creating another unwelcome silo in the enterprise IT infrastructure. The key is making LI the norm for BI applications, instead of an unfamiliar and expensive alternative. It also means integrating business applications processes that workers are used to, rather than forcing them to switch to another mode - like the old spellchecker applications that required users to shut down their documents first. That's how GIS has largely worked up to now.
Affordable data access - work in progress - Data access, at least affordable access, is another barrier that has kept LI stuck in the mud. LI is unique in that it needs a full set of reference data (usually licensed) that's not always readily available or cheap. The main G8 countries are well-mapped but it costs millions to get the data from governments. Many small, mid-sized and even large enterprises are simply priced out. Fortunately companies like InterMap are stepping in to make this reference data more affordable so you don't have to be a country to get at it. The company is spending $120 million to map all of Western Europe - an area of 15 million square kilometres. By 2009 it hopes to put together a complete off-the-shelf database that tries to put a GPS address on every rock, tree and building that is accurate to 2 meters - much better than government maps that are usually off by 100 meters. Intergraph owns the IP of that data which it will license out to companies. It claims that interest is high.
Cleaning up dirty data - As geospatial data is integrated into multiple applications, it also faces a data quality management problem. Data quality, particularly getting spatial reference data to match, is another sticky issue for LI as it usually comes from a multitude of sources, each of which is formatted differently. If data is more accurate then better uses of it come naturally. However, the issue of data quality is something that vendors try to either cake over or push under the rug, hoping that it will go away. It doesn't and it is the rug being pulled that often trips up LI. Data quality doesn't go away in a map. In fact it becomes more apparent. That means connecting a visual interface to clean back-end data sources which becomes a data plumbing-cum-integration issue. That said, spatial databases naturally filter out bad data - it's not possible to get geospatial (polygon) data into a spatial database if it's not valid. Specialist companies like Safe Software are starting to address the issue of data integration (i.e. getting location data to match) and geo-coding techniques are also getting more sophisticated. That's making the underlying spatial reference layer more accurate and consistent.
Slowly evolving standards - Evolving standards for geospatial data is another hurdle. The Open Geospatial Consortium (OGC) continues to make steady, if slow, progress to provide codified standards for spatial IT infrastructures that enable interoperability between systems and data. The OGC has a broad set of specifications in its OpenGIS portfolio (GML, WSC, KML, etc). Some of these standards are particularly important for industries like insurance, which aim to provide a consistent set of services upon which they can communicate. Maturing standards like ESB and XML and flexible SOA architectures and APIs are also opening up ways for LI systems and data to interact with enterprise applications like BI and CRM. Hence the OGC is also working closely with other standards bodies like W3C and OASIS.
Change of mindset - Finally, and perhaps most important of all, is a need to break the away from the siloed "them and us" mindset of LI user and vendor community at large. They need to realise that GIS isn't the "center of the universe," in spite of the value location-based data can add to the enterprise mix. Lessons should be learned from the relative failure of GIS to rid itself of its proprietary tag by moving LI capabilities into mainstream business applications like BI and CRM via applications that a lot of people already touch today - the current LI mashup focus on Microsoft SharePoint is a good example of this trend. Nor should location data be siloed into a special database and specialized set of IT skills. Technically, processing geospatial polygons should be treated no differently from other enterprise data streams. Of course it will take time for enterprise DBAs to get really comfortable with geospatial data. Most important of all LI should be sold not as a technology - companies aren't looking to buy maps - but as part of solution that solves real-world business problems or pains they are experiencing today. Technically LI has a mature set of technologies supporting it. But pushing it to the enterprise without a business case or paying attention to customer demand/need is the wrong way to go. In short LI needs to break away from its clique mentality and start listening more to the rest of the IT industry.