If you’ve been involved with a major enterprise capital procurement recently, you’re probably sick of hearing about ROI. A sales pro who wants to convince a customer to sign on the dotted line for a big chunk of money had better have worked up a convincing financial model showing how the customer is going to reduce expenses or increase revenues by an amount that’s several times what they’re being asked to pay. The same kind of dynamic applies between a line manager launching a project and the corner office; and between senior management and Boards of Directors. These days, people don’t want to hear about synergies and intangibles and brand leverage, they want capital-R, capital-O, capital-I, and they want it fast. So, is ROI potential useful in predicting the success of new technologies?
The Table · In this table, technology that any idiot can see will have a huge dollar impact gets a ten; something that requires a major leap of faith is a zero.
| Winners | score | Losers | score |
|---|---|---|---|
| SQL/RDBMS | 8 | OODBMS | 0 |
| Unix/C | 0 | 4GL | 10 |
| Open Source | 7 | AI | 8 |
| PC Client | 1 | VRML | 1 |
| WWW | 2 | iTV | 0 |
| Java | 5 | Ada | 5 |
| XML | 5 | SGML | 1 |
Discussion · 4GL is the only 10 here, because the premise was “develop software without writing code”—the savings potential is not hard to understand. AI—replacing expensive people with silicon—and SQL/RDBMS—do a better job of centralizing your information—are obvious business winners if they can be made to work.
The Unix/C example is instructive: who’d have thought that an operating system out of Bell Labs without records in the filesystem or a create-new-process API or a user-friendly command language would turn out to be just the ticket for your back-room computing? Similarly for the WWW; the return on viewing everything as net-hyperlinked text through a document rendering engine was far from obvious, perhaps until the fateful day that Fedex put their package tracking on the Web. And the PC Client; why on earth would anyone have thought, twenty years ago, that it would be worth investing in migrating computers out from the professionally-tended backroom environment to everybody’s desktop?
It shouldn’t be too surprising that this one doesn’t seem to be helping very much; just because something would be great if it could be built doesn’t mean it can be built (as in AI or 4GL). Another way to put it is: if something seems too good to be true, usually it is.
Conclusion · So far, the Technology Predictor Success Matrix is zero-for-three; apparent ROI has no predictive value, positive or negative, as to the success of new technologies. And, dear readers, I must admit that I decided to go through the candidate predictors in (roughly) increasing order of usefulness; so do stay tuned. Has anyone guessed the winner?
