In 1812 Napoleon lead 422,000 men on his invasion of Russia. 10,000 came home (and only 4,000 of those had actually been to Moscow). By today's management standards, that would have to be regarded as a disappointing result. Someone might even get fired.

But would things have worked out better if Napoleon's numbers men had been using 90's style tools - did he need an Executive Information System (an 'EIS')?

What is an EIS?

This is a surprisingly difficult question. The term 'EIS' first became popularly used around the start of the 1980's, referring to on-screen displays of business data, usually with a graph or two thrown in, to be used by executives, to help them do whatever executives do. It was also sometimes called an 'ESS' (Executive Support System), and it was sometimes confused with a DSS (Decision Support Systems).

Here what it is not: it is not the toolset that EIS vendors sell. There are many fine, powerful, elegant, well-supported 'EIS packages' on the market that can be used to build an EIS. Packages like Comshare's Commander, the SAS Institute's SAS/EIS, Pilot's Lightship, Cognos' PowerPlay, and Holistic Systems' Holos - just to name a few. These packages are great building tools - but the EIS is what you build. To use an analogy: "... people don't want to buy drill-bits, they really just need holes ...".

We'll use this definition: an EIS is a system that helps an executive manage his or her business. And that covers a lot of different types of systems.

A Simple EIS: The One-Day Cricket Graph

The graph above will be familiar to one-day cricket obsessives. It shows the build-up of scores, over by over, and highlights when wickets fell. Now the surprising and interesting thing about this graph is just how much information it actually presents, in a very simple, understandable format: it actually conveys, directly or indirectly, measures of run rates by over, total scores and targets, wickets lost (and hence wickets remaining) required run rates.

But above all, this graph is a powerful executive management tool: the captain of the second side to bat constantly knows if he is ahead of or behind the required run rate - not the flat level average run rate typically quoted, but the real gradually-speeding-up curve-shaped target. The captain has a continuing indicator of whether his batters need to be more- or less-aggressive in the run chase, correspondingly accepting the higher or lower risks attached, balanced against the remaining resources (wickets in hand).

Despite its simplicity, the cricket graph is an EIS ... just not a typical one.

Could Napoleon Have Had Better Executive Information?

Definitely. Look at the wonderful graph above drawn in 1861 by Charles Minard (and brought to popular attention by Edward R. Tufte in his must-have work, "The Visual Display of Quantitative Information"). This one magnificent picture conveys, in a single sweep, all of the measures of the key enterprise-critical information subjects: troop numbers, geography, and the temperatures suffered along the way. It's hard to disagree with Tufte in describing Minard's graph: "It may well be the best statistical graph ever drawn".

Napoleon probably had broad counts of his forces, from time to time, during the march to Moscow. But just suppose he had had a young Minard in his maps tent, getting troop counts every day, and plotting this chart daily, highlighting the rivers still to cross, and the falling temperatures to come, and then -with no little trepidation - showing it to the boss. Might Napoleon have called it a day before Moscow?

EIS Generations

Before the emergence of the first EIS packaged tools, EIS's tended to be paper-based reporting systems, providing information interesting to executives: sales figures, accounts, expense analyses etc. Typically, the information was extracted from company-internal systems by purpose-built programs - a new information requirement meaning a new program needing to be written.

Since the 1980's, the nature of actual EIS implementations in businesses has been shaped largely by the toolsets available.

The first generation of EIS tools, in the early 1980's, provided somewhat clunky, but workable, ways of building nicely-navigable on-screen versions of previously paper-based information systems. Data extraction was still, typically, a multi-step read-write-manipulate-import sort of process, but it worked OK. The big hassle was that the EIS, once established, needed to be maintained, usually by the Information Systems people - every time new information was desired (eg a new product line was introduced), the EIS-building process needed to be re-visited.

The next generation of tools was able to address that problem by building on improvements in relational database technology. Most of the EIS tools were equipped with powerful 'data mining' capabilities: the raw data contained in relational databases could be queried and analysed, identifying critical features of the data (eg product lines, time periods, types of measures etc), that would then be used to logically shape the screens 'on the fly'. Drill-down capabilities allowed the users to look at different cross-tabulations and groupings of data, and then 'drill down' into lower levels of details if the summarised figures for the groupings looked interesting. Magic. Unfortunately though, the powerful data analysis capabilities came at a price: querying of relational databases was too slow (minutes, not seconds).

In broad terms, the solution to the 'time-to-query' problem was to pre-load data, extracted from the various sources, into multi-dimensional databases. These database forms deliver results just as if a giant multi-dimensional matrix had been set up, sitting there pre-populated with figures for all of the possible cross-tabulations, just waiting for a user to drill down to see the number in any particular cell. This EIS database would be updated, and if necessary (logically) re-shaped, at regular periodic intervals.

The most recent EIS tools also support powerful on-line analysis of the data presented - drawing trend lines from business data over time series, cross-relating data subjects to investigate relationships and causality, and so on.

And because most executive users (though not all) now routinely work with their own PC, connected to a LAN, running under a Windows or similar environment, the data extracted and presented by the EIS can be neatly re-used (and re-communicated) with all sorts of other PC-based tools. The EIS is slotting in as a component of an executive's work-desk, rather than being a standalone tool.

Pulling Data from Unexpected Places

Good EIS's obtain their source data from existing, operational sources. Accounting systems, and product order/delivery systems, are classic, generally reliable sources of data on many aspects of business performance, not just those applications they were built for. (For example, in a life insurance company case, with some manipulation the Policy Administration files become sources of surprisingly complex information on Customer and Sales subjects.)

A Good EIS Will Surf the Net

External data is also pulled into most EIS's, in some way. Industry sources may publish competitive data; the business press is a vast information resource on industry and competitor-specific subjects. The Internet and the commercial global networks (like CompuServe) are potentially bottomless wells of raw data.

The role of the EIS then is to gather this data from the many, disparate and varied sources, and to present that data conveniently and quickly, and to allow analysis of that data, to allow the executive to manage his or her business better.

The EIS as a Dashboard ...

To us, the critical question is not how the EIS tool works. It is "How does the EIS help an executive to manage his or her business better?".

There are two schools of thought here. We favour the 'Dashboard' approach: the EIS should show the critical business information, ideally live as it happens (or as close to live as possible), in an instantly-visible front-screen, just like the dashboard in a car (and that also means that the executive should not have to fight through layers of pretty but information-less icons, just to get down to the real stuff). The executive will do his or her work in a way that suits him or her - the job of the EIS is to provide data, and possibly some analysis, to support that way of working.

For example, I come from a life insurance background: if I were running a life company, I would like a series of colour screens on a wall, each one constantly monitoring one of the 15 or so critical business areas (Sales, Distribution Channels, Customers, Products, Market Share, Revenue and Budgets, Balance Sheet, Key Performance Indicators, Staffing, Projects etc). I would ponder this 'Big Board' each morning (while enjoying a nice Cafe Latte); and then initiate actions in particular areas, if any of the displays indicated problem or opportunity areas.

... or a Slice-And-Dice tool

The EIS toolsets strongly empower a 'Slice-and-Dice' activity, which works a lot like a diagnosis tool. An executive is (typically) provided with a powerful array of subject areas, categorisations, stratifications, timeframes and measures (units, values etc). He or she then selects whichever 'cross-tabulations' are of interest, and then 'slices and dices' various mixes and levels of these selections, to draw out pictures of aspects of the business - this allows the executive to consider and initiate actions to deal with specific business issues.

Of course most implementations will (properly) have elements of each of these approaches: perhaps a Dashboard as a starting point, allowing the user to go from there to Drill-Down or Slice-And-Dice into more intricate detail (if that is what he or she wants to do).

Is EIS an Important System?

Only if you care about managing businesses better.

Most business systems work to date has been all about administering data: recording, processing and reconciling transactions. Today's technology allows us to use data as information, in ways that were simply not possible a few years ago. So now it's not about technology anymore: it's about how to present and communicate that information, in ways that will help the executive manage a business better. The brilliant EIS designs will be insightful, artistic and (above all) directly and immediately relevant to the particular business.

Napoleon really did need his Dashboard - and it should have looked just like Minard's brilliant graph.