Most enterprises still haven’t figured out digital. That’s the bad news. The worse news is that they’re not spending the money necessary to figure it out on the form factors that matter most–namely, mobile devices.

This is according to several different surveys, all of which point to one big conclusion: we continue to epically underinvest in mobile, which is bizarre, given the general realization that mobile is the platform going forward.

The small screen keeps getting bigger

According to Ericsson’s June 2015 mobility report, 90% of the world’s population over the age of six will have a phone by 2020. This isn’t just feature phones, either. According to Ericsson, smartphone subscriptions will more than double by 2020.

While much of the growth is in developing economies, there’s growth everywhere, even in markets (like North America) that already have greater than 100% mobile saturation:

Given this growth, it’s not surprising to see mobile phones dominate how we access the internet:

This growth keeps accelerating, with mobile data traffic in Q1 2015 a whopping 55% higher than the year before. By 2020, 80% of mobile data traffic is expected to come from smartphones.

Given these gargantuan numbers, it borders on shocking just how tepid the industry’s investment in mobile has been.

What, me worry?

Granted, as an industry, we’re still coming to terms with digital, generally. One UK survey of 1,000 business leaders found that 57% acknowledge they don’t really understand the digital economy.

Even so, a much bigger number is pushing the desktop web as a way to reach customers, as a recent Econsultancy/Adobe survey found.

It makes sense that companies would invest in the desktop, given that it remains a significant channel for reaching customers. And, as Intercom’s Paul Adams insists, while mobile devices are critically important, they’re just part of a broader strategy that necessarily will include the desktop web.

But given the disproportionate share of mobile traffic, and the rapidity at which customer preferences are changing toward mobile, it’s not good enough that just 26% of companies are putting mobile websites or apps first.

Consider:

And yet… this:

According to this same Econsultancy survey, 62% of enterprises plan to increase their mobile investments in 2015, but only 33% of the companies surveyed have actually set aside budget to experiment with mobile.

And experimentation is precisely what is required.

Iterate. Iterate. Iterate.

Here’s the reality: companies are going to fail. We already know that most IT projects fail, particularly in new areas like big data or mobile. We simply don’t have the skills and experience built up to ask the right questions of our data (in big data) or know how to roll out a winning mobile strategy.

But sitting on our corporate hands won’t fix this.

One of the reasons that open source has taken off is that it facilitates iteration, as I’ve written before. The same is true of big data, where Amazon’s elastic infrastructure makes it easier for enterprises to experiment with the right questions to ask of their data.

Not surprisingly, mobile follows this same pattern. Companies that want to succeed in mobile should, of course, talk with peers, engage agencies, etc. to try to figure out the best way to begin. But the key thing is to begin. Until a company is actively learning mobile “on the job,” it simply won’t know how to best craft a customer experience that drives real business value.

So, how to begin? In my experience, the most successful companies build out a mobile center of excellence, which can start with a single person. This nucleus of mobile talent and expertise is then used to disseminate mobile knowledge, push the enterprise to move faster, and run interference with the executive team to make sure they support the mobile initiatives.

I’ve seen this model work at major retailers and smaller government agencies. But, again, nothing works when a company allows bureaucratic inertia to prevent it from starting. To get mobile moving at your company, therefore, the first thing to do is to designate a mobile team, then give them room to iterate.