For a period of time we would now measure in decades, heuristic modelling has been the go-to approach for logistics network consultants and solution designers to establish the best locations for distribution centre placement. But is this approach still the right one?Sadly, in most cases, the answer to that is almost certainly going to be no.
The first thing we must clarify is what is heuristic modelling? In its most simplistic terms, when considered in the supply chain analysis world, heuristic modelling is an approach that enables a large number of potential outcomes to be tested against a specific set of defined variables to produce an answer. Usually, such models are defined in quite complex algorithms usually built in modern terms into fairly complex IT solutions.
A simple question that may be posed to a heuristic model, is ‘based upon a certain predefined data set, where is the centre of gravity of my demand?’ Complex heuristic models pose the same question across multiple demand points and over a prolonged period of time. However, the larger the data set you try to work with, the more nodes and demand points you seek to test versus a larger number of supply points, and over a longer period of time can critically diminish the value of the outputs of heuristic modelling, as the answers simply become too complex to decipher within a reasonable time period. This then leads to compromise on the level of detail in the analysis, with all the knock-on effects you expect this to bring.
This is a trade off that is well known in the logistics consulting world, and one often faced when we are attempting to design the ideal logistics network for a client. Taking the heuristic modelling approach alone will rarely enable solution designer to establish the ideal network scenario for a business. There are too many other factors that heuristic modelling struggles to take account of, such as:
- Changes in labour and land availability;
- Volatility in fuel and utilities costs;
- Changes to customer behaviour and route to market expectations;
- Changes to commercial strategy, product channel and segmentation.
These are just a few of the things that can impact heuristic models making them less viable for supporting future site selection.
We contend that the first thing you need to do as a business is establish your service demand criteria, i.e. your customer service aspirations, and understand the types of fulfilment centres you need to satisfy that criteria. Once you have the right mix of types of facility mapped out, then you can consider heuristic modelling to develop the ideal logistics network for the future.
There is a very big change on the horizon however. The rise of AI will likely see the development of new ways of creating dynamic heuristic models that are exponentially more capable than traditional modelling approaches. Whilst this type of technology is as-yet untested in the market, there is little doubt that as AI grows, it could have a significant impact to the modelling approaches in network evaluation programmes.
Looking at the next 5 to 10 years, the best approach will continue to be based upon expert judgement and analysis of businesses commercial objectives and strategies, aligning those to greater macroeconomic challenges rather than working on simple heuristic modelling techniques.
It is fair to say that beyond a time horizon of a decade or so, who knows what we will actually be able to use as modelling technology. But one thing is for certain…the consumer market place, the global supply chain, and technology will lead us to a very different place from where we are now.