By: David J. Wardell

© 1989 By: David J. Wardell.  Reproduction or redistribution in any form without written permission is strictly prohibited.

The problem with travel auditing software (those software packages that purport to help travelers plan itineraries consistent with travel policies, find the "lowest fare", or verify that policy-checking and fare-searches were properly carried out) is lack of knowledge -- not necessarily the in the software, but in the user. If these products were better understood and expectations for them not so grandiose, everyone -- sellers and users -- would benefit.

Specific product knowledge important so that sales claims can be realistically evaluated; focusing upon results, rather than process, is critical. In the end, while it's appropriate to understand process in order to evaluate performance, whether or not the system delivers the result is much more important to you than how it works.

I'd like to consider some common features of auditing and planning software, illustrated by a few representative implementations employed in the auditing market.

Space constraints do not permit detailed reviews of any one system; these are general discussions, not in-depth analyses.

As I've previously mentioned, databases, both internal and external are the keys to successful travel auditing of any type. Without access to reliable data, which form the "knowledge base" against which decisions are made, the correctness of any analysis or judgment is obviously called into question.

External databases almost exclusively comprise fare, rule, and itinerary information from which specific selections are made as they pertain to trips under consideration. Beyond question, when it comes to fares and rules, the CRS databases are biggest and best.

In the world of auditing software, everything is tied, one way or another, to a CRS. This is the mechanism where data being audited are stored, and the transmission vehicle by which the auditor receives the data.

Since the natural marketing position of any auditing system says that the CRS aren't doing the job, most system designers try to distance themselves from CRS databases -- the alleged repositories of those errors and ambiguities for which the auditing software compensates. This leads to exaggerated claims of multiple and proprietary databases.

Unfortunately, this whole proposition arises from ignorance -- both of CRS databases and databases generally. There are really only a few primary sources of fare and rule data, which are the "publishers" of electronic air tariffs, such as ATPCO.

CRS (and other) databases are really secondary derivatives from these primary sources, in that all CRS' take their base electronic fare and rule data from the same, small group of sources. However, the seemingly logical conclusion -- that getting "closer to the source" (by removing the CRS intermediary) necessarily equates to better data -- is unjustified (the principal holds in archaeology, not computer science).

This is because CRS databases undergo substantial transformations as they move from the primary electronic source into the "production" version employed by the CRS. Many airlines employ hundreds of people whose jobs are to augment, enhance, and correct these electronic databases.

Thus the CRS databases -- the same ones available to anyone with an efficient airline reservation system -- are usually better than the sources from whence they came and, collectively, comprise the most effective body of information for auditing purposes. In fact, throughout some extensive research I conducted into the subject last year, it was difficult to identify enough instances where the reliability of the "primary" sources exceeded that of the "secondary" to the degree that would make creating yet another database (this time a proprietary one) based upon the "primary" sources commercially practical.

Most auditing systems (or services) rely upon the CRS as their exclusive data source. As you can see, while this is a commercial necessity it is also a fundamental limitation, as these systems don't use any "special knowledge" that isn't available to you. Their only value is in the way they apply it.

One notable exception is TravelMation's "Trip Planner" system, which is built around several proprietary databases that are unique to the system (and which don't specifically deal with fares). One is a specialized way of expressing fare-rules which is more usable (and therefore more accurate in application) by the system.

Any type of proprietary database activity is decidedly a two-edged sword: If the result is truly (and demonstrably) more accurate, then this is the result the buyer of the software or service should focus on; however, that result is only better as long as the supplier is committed to the difficult and expensive job of maintaining it.

Most auditing systems and services do a reasonable job of researching and correcting domestic U.S. itineraries. Performance is much more restricted (or non-existent) for international itineraries or trips wholly within foreign countries.

This is not only because of database problems, but also because most of the auditing logic employed is very straightforward and doesn't lend itself well to international fare construction, where there may be a number of "correct" answers to a particular problem, based upon the perspective you select.

One auditing technique, employed almost universally, is to select various traveler parameters that will be held in a traveler or company profile database and used as criteria to evaluate trips an agency may book, together with several alternate routings or carriers that also conform to the "policy". This is simple enough to be done without extremely complicated programming.

Again we run afoul of the database problem. Somebody has to set-up and maintain what can become a monstrous database that's constantly under revision. This is likely a job your corporate customers don't bargain for when they start shopping for auditing software.

CompuCheck goes a step further in its "Fare Sentinel" product. One the itinerary is booked, the product checks periodically to see if a more appropriate schedule may open. This is a nice feature, but it could get tiresome if the system were constantly identifying alternates you'd already dismissed for procedural or preference reasons it doesn't "understand".

Apart from identifying and correcting "exceptions", auditing software needs to report on what it's been doing. Most systems are sadly deficient in this area more than any other.

It seems as though the same people who write travel agency corporate reports and pile them so full of unintelligible "data" that the real "information" corporate travel managers are looking for can't peek through are writing auditing reports in their spare time. None of these people has yet figured out that managers want to see management-oriented reports attractively displayed -- which means graphic illustrations, one-page summaries, and reports-by-exception -- and no one (including computer specialists) takes time to read (or can understand) the detail.

The reports are what the system is all about -- from the user's point of view. This is the only tool a corporate travel manager has to measure the success or failure of the system. If you're properly focused on the result, how the good, reliable information got on paper becomes unimportant.

A few final thoughts: Preparing accurate and reliable reports and summaries is a highly mathematical and statistical job -- that is rarely undertaken by people with this specialized training. This leads to highly questionable results that can get everyone who uses them into trouble.

Next time you see some supposedly authoritative reservation or ticket auditor making bold statements (particularly about your accounts), ask some embarrassing questions:

• What's the size of the sample?
• How do you know it's representative of the population?
• What valid statistical technique (more than adding a few numbers or calculating an average) was employed to predict a trend or draw a conclusion?
• What's the uncertainty factor applying to the data in question?

Frequently you'll find that checking records and writing reports doesn't qualify someone to interpret reports. Statements people make about your business that might appear troubling are often cannot be defended.

Auditing software and services have come a long way, but there's still a great distance to be covered. The definitive auditing system has yet to be developed.

If these systems affect your business you should keep an open mind, but remember there is still great imprecision in many products.