Main content

Digitizing the past or transforming for the future?

Den Howlett Profile picture for user gonzodaddy March 10, 2014
Summary:
SAP's provocative view that we have digitized the past has a certain truth but is repeatability the answer? Yes, but only to a limited extent.

When we founded diginomica we knew there was a problem that few media companies are addressing: we have an abundance of technology yet so many businesses seem rooted in the past. Over the last 10 months we've showcased many examples where genuine progress is being made but still these largely represent outliers. They are often the leaders or visionaries who see well beyond the walls of the enterprise. Sometimes they are businesses caught up in sweeping change. Why then is the pace of change so slow?

Yesterday, Sameer Patel, SVP and GM social collaboration software, SAP attempted to answer that question by presenting a slightly FUDdy presentation that talks about the work done to date to instantiate the digital enterprise. Check it out above. His thesis is that much of what has happened so far represents a digitizing of the past coupled with a failure to ensure repeatability. This slide (below) encapsulates it well:

digital transformation
Patel's argument makes a lot of sense in a process driven world and I congratulate him for the work he has done in embedding the social 'glue' into the fabric of certain SAP technologies. It is not job done by any stretch and SAP is not alone. Other enterprise vendors are doing the same or similar. They're all in market together so there's no shortage of coverage.

I have a different perspective that starts with accounting. I know it's not everyone's taste but hear me out.

The accounting bind

Like it or not, accounting is a necessary requirement in any business for both governance and management purposes. Every major enterprise vendors has this capability. But not one of them has truly rethought what it means in the 21st century. Or if they have then I haven't seen it.

Everyone starts with the correct notion that the set in stone rules established in 1494 must be the backbone from which everything else follows. And I agree. Up to a point. Those rules have served everyone very well and continue to do so in a world where you are looking through the rear view mirror and where regulation demands conformity to execution against a repeatable set of rules.

And for any new approach to  make sense, it has to at least adhere to established structures. But that's where it should end. So for example I asked a colleague something that has direct relevance to a project in this field upon which I have done some recent work. It goes like this: If we were to design a new way of management accounting designed to help line of business decision makers then how easily could I remove standard costing in manufacturing?

The answer was an emphatic 'no.'

I was brought up on that system. But here's the problem. Standard costing is a nice theory that is no longer relevant and is certainly not relevant to today's decision maker. Here's a definitional view that explains the problem:

Rather than assigning the actual costs of direct material, direct labor, and manufacturing overhead to a product, many manufacturers assign the expected or standard cost. This means that a manufacturer's inventories and cost of goods sold will begin with amounts reflecting the standard costs, not the actual costs, of a product. Manufacturers, of course, still have to pay the actual costs. As a result there are almost always differences between the actual costs and the standard costs, and those differences are known as variances.

It was a system made popular during the early 20th century period of mass mechanization and at a time when cost inputs were relatively static. You could predict with reasonable certainty how costs would work out. You could make reasonable assumptions that minimized the risks of running out of control. In fact the standard costing system was designed specifically for that purpose.

Roll forward to today and we live in a different world altogether. With so much manufacturing and process work outsourced, business can still predict costs to some extent but what it cannot do is reasonably flex production to suit changing demand without hurting the supply chain.

Viewed simplistically and other than in a few rare cases, collaboration in the physical supply chain is almost non-existent. The top down, supply chain master model still rules. Visibility is limited at best at a time when market volatility is the norm.

Patel's view implies that you can go some way towards solving this problem through repetition and through the recognition of patterns. But he caveats by saying that we need flexible systems. I think that's tinkering. The trouble is that the underpinning systems which provide some of the most important data - such as standard costing - are incredibly rigid and difficult to extricate from inside complex systems.

An ideal scenario

They have to change. If I had my way, I'd ban those rigid systems and replace with opportunity cost or marginal cost systems. For those who have the stomach for this stuff, here's an excellent Slideshare presentation that explains the difference between the two approaches.

I'd also argue that pattern recognition as implied in Patel's presentation is at best an optimistic view. Patterns are fluid. They can almost be thought of in the same way that we view the weather - carrying a high degree of short term probability in forecasting but with 10-15% wiggle room.

In other words, repeatability may only be of limited value if it becomes baked into the discovery systems Patel invokes. Rather, I see a much better application for the concept of barely repeatable processes as the basis for decision making. In that world, the row and column presentation that I sat through yesterday does not exist.

Instead, I access intelligent systems that only present me with the exceptions I need to understand but with the capability of seeing them in the context in which I need to take decisions. That cannot simply be financial data, however well constructed or accurate.

Context might easily mean the application of sensor data, unstructured snippets from email, internal chat systems, documents etc. What's more, I need this as close to real-time as possible so that I can course correct rapidly but with the confidence that hot data gives me.

My sense is that Patel has painted half a picture, one that is readily understood and which makes the SAP pitch palatable. But it needs to be so much more. Who will drive this?

Quo vadis?

I see snippets coming from tactical solution sprouting among the FOSS Hadoop communities. I see companies like GE attempting to take on complex operational problems among specified verticals. I see the massive potential among MongoDB customers. These are only scratching the surface.

I do believe the mega vendors have a strong play here and perhaps Patel is portending a piece of that. But...it cannot be at the expense of massive complexity and high cost. Those days are over.

The challenge therefore is akin to something a colleague said the other day: 'We need to make big data little enough to make the right decisions.' Yes - and we need to make little data comprehensible to those who make business decisions.

Oh yes - and we need a management mindset that not only embraces these changes but does so in a way that invigorates those who have to do the work. That, in the end, may be the hardest problem of all.

Disclosure: SAP is a partner at time of writing. Sameer Patel and I have undertaken joint consulting in the past.

Loading
A grey colored placeholder image