Digital transformation to nowhere in the age of AI

Digital transformation was always a scam.?

The idea that bits and bytes could bring company-wide transformation on the scale of electricity and steam power is flawed. And we conveniently forget that the companies who launched such technologies were rarely around to reap the benefits – US railroads mostly went bankrupt and the average Fortune 500 lasts less than 15 years today.?

The simple fact is: any technology is either an extension of an existing business model or a complete rewrite of the rules of the game. It’s rarely in the middle and it’s almost never possible to go from A to B through a planned transformation.?

For two decades we have used military or engineering metaphors for digital transformation, and in more modern terms, agile, kan-ban etc. to deliver the value of software transformations to teh enterprise But let's consider a different possibility:?

Most of the true digital revolutions have come from play and creativity.

The world wide web was only briefly the domain of academics before teenagers made it into their playground; the silicon valley myth of the college dropout is one of creativity, not industrial programs; and most of the stuff we consider digitally ground-breaking today was hacked together by passionate people in garages, not corporate labs.

The reason is simple: the unit economics of incremental digital innovation.?

As personal computers and then internet became cheap, everyone had the opportunity to experiment with new services. This created a boom in services-based GDP for the last 30 years – even if economists still struggle to calculate the full impact of free services like Google, Facebook and now OpenAI due to lagging models.

Returning our attention to the world of work, we can consider the world of enterprise digital transformation in three phases.

The initial phase was about foundation more than transformation. Moon landing scale projects landing IT within the world of business. The initial leap from analogue brought such immediate digital benefits that little creativity was considered. IT took control, installing suites of tools to keep things “simple”. Enterprise Architecture, 5 year ERP roll-outs and end-to-end standardization of on prem suites of wall to wall functionality. Project management and control models were all taken from military and engineering given the staggering costs of the projects and hardware.

Then the second big transformation after this was the move to SaaS. Single composed apps delivered over the internet, in a decade we grew from 900 to 35,000+ enterprise SaaS apps all with their distinct set of functionality and user-interfaces, best-of-breed was born combined with the SaaS suited that tried to span as broadly as possible. The added flexibility lets teams choose best of breed apps from different suppliers, thereby relating their choices more closely to their actual business decisions. This gave teams outside IT more interest and more to gain from getting involved in the buying process. The buying decision moved to the business side of the organization, but sales cost also went up dramatically, now there were more stakeholders to sell to and more objections to overcome than ever before, thankfully it could all be powered by zero interest rates and VC money.??

Since then, we’ve seen the move toward “product-led growth” (PLG). Tools infiltrating the enterprise bottom-up, like Slack, Google apps and similar, but still the adoption eventually tapers off as it meets the moats of enterprise resistance and the PLG sales cycle is long and expensive (average payback is 4 years for PLG or longer). The second problem for PLG is the dreaded pilot or sandbox. IT is happy to experiment and do a pilot, but these pilots very rarely scale beyond the initial first implementation as they don’t have buy-in or can meet enterprise level risk and compliance criteria.

So true transformations have a hard time in an inherently risk averse and static enterprise environment and more importantly an environment where evaluating value, committing and iterating is incredibly expensive, if we can’t be creative and experiment then we can’t transform, bytes are not bridges. To sum it up we are still trying to capture the revolutions of Silicon Valley using first wave enterprise project management and second wave SaaS sales cost, squashing all unit economics and ability to innovate.?

What can we learn from this?

We need to bring down the unit cost of transformation, so we can experiment and “play” with new ideas more easily, but at the same time we need to take the lessons from PLG or the dead-end “pilot” and make sure that as we play we can scale it up to production scale and real business value and not just do sandbox experiments or pilots to nowhere.?

This is even more true with generative AI, we are seeing lots of cool demos or hyped stuff, but how this is going to get from A to B in the enterprise world is an even harder question to answer without understanding how we fix the digital transformation journey.?

That is why I think we will need commercial models to reach the customer, as discussed in the death of a SaaS salesman, new ways of making software and new ways of creating customer alignment. This also requires changes in the organizations themselves. What happens when every employee has the power to automate their own workflows or create their own “software” just by prompting.?

The coming digital transformation will not be an evolution, but a revolution and it will be far more unpredictable than anything we have seen so far. CEOs and CIOs better be ready for it or their companies won't be around after.