Technology, more than any other single factor, is what drives an economy forward.

Ironically, technology is, more than any other single factor, what holds individual companies back.

Is it what drives economies forward? Of course it is. In the absence of technological change, products and systems eventually optimize themselves, whether through the natural selection imposed by the marketplace, or by the eventual impatience of ingenious employees faced with an inefficient status quo.

Yes, painful as the message may be, without new technology, new-and-improved starts to mean the same old products in new packaging, just as without new technology, internal process improvement starts to mean fixing what’s broken by breaking what’s fixed: Once the stupid stuff is gone, process changes involve trade-offs. Or, to put it a different way, quality might start out being free, but it doesn’t stay that way.

Without new technology, business change is mostly a zero-sum game.

But with new technology? New technologies means new capabilities, some of which might prove useful. Startups base new products on it; established firms might build it into existing products (or else buy the more promising startups). People and companies with money to spend decide to buy the new or improved products, which means that by definition there’s now more wealth in the system than there was before.

And so, the new technologies have moved the economy forward.

Meanwhile, an average CIO in an average business has just survived (barely) an ERP implementation. Was it a good idea? Maybe.

For companies that want an integrated business architecture, ERP is a necessary … not sufficient, but necessary … precondition. Why? Because that’s what ERP does — provide an integrated informational view of an enterprise — and if you don’t buy and implement a commercial alternative you’ll just go through the enormous effort of building your own.

For companies pursuing a more entrepreneurial or holding-company approach, ERP mostly puts all the business units on a common accounting platform — nice, but not a big deal in the greater scheme of things.

Or else it puts handcuffs on the internal entrepreneurs and no-longer-independent business units.

But never mind all that. What matters this week is that the company in question now has an ERP suite in place. Or, if you like cloud stuff better, the company has bought a bunch of Salesforce licenses, configuring it to support its sales and sales management practices and integrating it into its (you know this is coming) ERP suite.

Or … pick your poison: Software that does important stuff for a company takes time and effort to implement. And then, every year, the company in question adds a swarm of small enhancements to tailor and adjust the system so it fits the company even better.

All this represents more than just sunk cost, although the sunk cost is nothing to sneeze at.

No, even more than sunk cost, the software a company runs on represents a collection of ingrained habits — the software becomes embedded in the business culture.

And now, to pursue new opportunities, or to compete with businesses that have newer and better systems to support their operations, or because the company’s old business model has run out of gas and the company has to do something different or die … for one of these reasons or maybe a different one, the company has to do something its current suite of applications won’t support very well.

Faced with the need to rip out and replace major chunks of the company’s business infrastructure, in order to support a new and untried business direction, it’s easy to find dozens of reasons to coast along for one more year, leveraging the sunk costs instead.

What’s the solution? Is there a solution?

According to legend, when American Airlines first invented frequent flier, it ran the whole program on electronic spreadsheets for three years, until it was able to build a “real” system to handle the program. True or not, the concept points the way: When trying out a new business concept, don’t invest in a robust, scalable systems architecture. Doing so will take everyone’s eyes off the ball. Instead, cobble together capabilities out of whatever cheap alternatives will be good enough to get by.

This is, after all, a new business. Nobody knows what it will really need.

And when nobody knows what they really need, there’s no point investing heavily in it.