Wednesday, January 09, 2013

Business Network Optimization

Some @ATKearney consultants have written an interesting article on Business Network Optimization

"Anyone thinking about rationalizing a network would naturally ask whether so many nodes are really necessary. Networks are a great deal more complicated than that, and managing them requires expansive strategic imagination."

A simplistic accountancy view of a network looks at the direct contribution of each node. From this viewpoint, some nodes may not produce enough direct value to justify their continued existence, and there will be calls for these nodes to be closed or merged with their neighbours.

For example, there are several proposals currently under consideration within the UK National Health Service to rationalize Accident and Emergency provision by closing some hospital departments and relocating staff. These proposals are based on arguments about the optimal size of an Accident and Emergency unit, and on claims that smaller units are unlikely to deliver value for money or clinical  excellence.

Opponents of these closures point to the indirect effect of these closures, including the likely consequences on non-emergency healthcare services at those hospitals that will lack accident and emergency provision, as well as the wider social impact on the local community.

The example given in the A.T. Kearney article is the French postal service, and the authors assert the indirect value of the village post office, using an almost untranslatable French term l'animation du territoire, the "animation of the territory".

The Kearney article identifies three types of business network, which it calls Production, Service and Distribution, and eight elements of network management which must be optimized together. It calls these KNOTs, which stands for Kearney Network Optimization Tools, and asks us not to think of them as merely a laundry list of best practices used to build an optimal network. 

The eight elements of network optimization (KNOTs)

The article illustrates the concept of indirect value in terms of the cross-over between physical and online retailing. If a customer views a product in a physical store and then orders it online, the physical store is providing some indirect value to the retail operation as a whole. It is therefore makes sense to optimize the entire online/offline network as a whole, rather than regarding them as two separate networks. See my post on Showrooming and Multi-sided Markets (December 2012).





The article also indicates the importance of Big Data in trying to make indirect value visible/manageable.

The general architectural point is that the business performance of a business network cannot be reduced to the sum of the performance of each individual node taken separately, and is an emergent property of the network as a whole. But you already knew that, didn't you?


Some network optimization problems can be tackled using Operational Research (OR). I found a case study published by Paragon Decision Technology, showing how its AIMMS mathematical modelling tool was used by Horoz Logistics, an integrated logistics services provider in Turkey, to optimize its nationwide distribution network.

Alternatively, it may be possible to build some degree of dynamic adaptation into the network. This approach has been explored by a number of technology companies, including IBM.



Eric Sauvage, Charles-Etienne Bost, Eric Gervet and Kevin McDermott. Tied in KNOTs - The right way to think about network optimization (undated)

Optimization of nationwide fertilizer distribution network with AIMMS (2008)

Grace Lin and others, The Sense-and-Respond Enterprise - IBM researchers develop integrated SAR system of global value chain optimization (OR/MS Today, April 2002)


updated 10 January 2013

1 comment:

  1. Richard, You might also like to refer to "The Antitrust Analysis of Multi-Sided Platform Businesses", by
    David S. Evans and Richard Schmalensee. http://papers.ssrn.com/sol3/papers.cfm?abstract_id=2185373
    It contains further evidence in support of the argument you are surfacing here...
    Philip

    ReplyDelete