There is a new breed of systems integrator stalking the boardrooms of businesses up and down the country. They're not called systems integrators, or even consultants - they are marketing agencies. And they are about to cause a whole new class of IT screw up.

The marketing world is about making changes happen. It's a tried, tested, and remarkably inefficient model. You start with a big target audience, you market to them, and eventually you whittle that big group down through to a much smaller volume of people who make the change that you require - often, buying your product or servcie. Technology has made this process more efficient (email and web are far cheaper and quicker than mailshots and traditional media advertising), but the fatness of the funnel at the top of this process is still wide (if not wider) to gain the trickle out of the bottom.

For example, every quarter I run a campaign in which my self-selected audience of recipients get sent a report of, hopefully, some value to them. They must think that there might be something of value coming because they all personally went through the effort to sign up for the thing. Each quarter, about half of the recipients even open the message. And that, according to the stats of the email engine that I use, is really, really good when benchmarked against their general experience.

Now, the point here isn't the effectiveness or otherwise of email as a marketing tool - but that in that self-selected audience, getting them to take the action of just opening (let alone reading) an email is really quite a challenge, and I'm doing well against industry benchmarks.

This model of affecting change, with leakage throughout the process, is the way marketing works. And if you do enough of it, you get enough customers to change their behaviours to make some sort of business happen.

That's the DNA of marketing. It's part science, it's part art. And change results, but only on a minority of the people going through that funnel.

As marketing agencies build up software development capability, and with the client/provider relationship built on solid trust as a result of proven results, they are now starting to use their capabilities and methods to deliver internal systems rather than external marketing. Often shifting sideways (through services like content management) they are increasingly being given the opportunity to develop pure, internal business systems. And here's the rub: their change management methodology (the funnel) is utterly inappropriate when it comes to delivering internal business results. When you want to get something to happen within a business, you can't afford to make the change happen within only a minority of your staff successfully being processed through the funnel. You need everyone on board.

As we increasingly rely on mass-scale web services both within and outside of our working lives, the temptation to believe that technology alone will affect meaningful business outcomes within organisations. Look at Facebook - they've got a quadzillion users! No one bothered to do change management to allow that one to happen!

But there's a difference between a loose platform in which a tiny minority actively participate, many lurk, and even more sign up but barely ever use it, and a business change initiative when you need everyone on board. It's a funnel, again. A massively-scaled, hugely successful funnel, but a funnel nonetheless.

Imagine if you used a funnel approach to change the behaviours of your customer service group. What inconsistency would emerge? What would customer service be like? Actually, you can phone up most utility companies to find out the answer to that question…

Using a marketing method to deliver business change is, quite frankly, doomed. And to be fair, many traditional IT approaches (deliver the tech to specification, leave the change management to someone in the Comms team) haven't delivered that well either.

But delivering significant internal business change facilitated by technology is something that progressive IT management has got its head around in the past 10 years. Whacking in a system to see if it works was an approach in decline, but I fear we are seeing a new generation of tech-first projects emerging out of the agency space that will cause new parts of organisations to have to learn these lessons once again.