Analytic tools "so easy a 10 year-old can use it"

If you search the web you'll find lots of analytic tools to support your business: tools to help with forecasting, inventory optimization, risk analysis, simulation for production lines and warehouses, production scheduling, supply-chain network design, vehicle loading, price-sensitivity modeling and planogram building - and that is very, very far from being an exhaustive list.

Some of these tools are bought as a service that includes expertise to prepare your data, do the modeling work for you and configure the system to meet your needs. These tools will be much more expensive than the 'roll your own' variety and the more frequently that expertise is required, the more you will pay.

If the business problem you are working on is relatively stable (meaning that the same issue comes up repeatedly and your business rules around how to solve it stay the same) such a system can have a long a fruitful life once set up. Vehicle loading applications are a good example - once configured to your needs they should need infrequent updates to work well.

Many analytic applications are more one-off in nature and if you use an expert to help you with every use it will cost you a substantial amount of money. So, do you really need all that expertise or is it simple enough that you can pick it up? Desktop analytic tools could help you deliver millions in savings and can be bought for a few thousand dollars. Many of them are marketed as being simple to use, and imply that you'll be able to pick it up as you go with, perhaps, a little training,

Let's look at an analogy - driving a car. Cars are very complicated things and yet almost every adult can drive one. Why would business analytic tools be any different? Cars are certainly complicated but I see no reason a 10 year old could not master driving one because they do not need to understand or master most of that complexity.
  • Cars have been around for a long time and have seen billions, perhaps trillions of dollars invested to make them robust and hide almost all the complexity from the driver.
  • Cars have relatively few controls that you need to understand to operate them: ignition, steering wheel, brake, accelerator, turn-signals, lights, mirrors, windscreen-wipers and gear selector. I know there are more gadgets but master these few (and hoping I missed nothing important) you can drive a car.
  • You get instant feedback when you use these controls badly and that really helps you learn
  • Most of us get a lot of practice driving a car and some of us are still quite bad at it :-)
In contrast most business users are ill-equipped to take on analytic tools where the complexity remains very apparent, there can be hundreds of options to choose from in building and executing your models and these must be understood to use the tool effectively.

Let's make a distinction here between getting a tool to work and getting it to work effectively. I've seen a lot of spreadsheet using Excel's Solver add-in which solves optimization models: you build a spreadsheet that models some aspect of your business then ask it to find the combination of input values that maximize your output (profit, sales or some such value). If you built a model with no errors and avoided a handful of functions that this optimizer can't handle it will run and most likely return an answer. It may even be able to guarantee (depending which algorithm it used) that this is the absolutely best answer available - the optimal solution to your model. Unfortunately, many of the models bear little resemblance to the real-world issue you are trying to optimize and consequently the optimal solution for the model is not particularly good (and certainly not optimal) in the real world.

There's one good example of an analytic application that really is "so easy a 10 year-old can use it" and it's not a business application at all. It came to prominence in the last few years and is now available in cars, on low-cost specialized devices, even on your phone - GPS navigation. Behind the scenes there is a large, accurate database of road networks and locations along these roads and an efficient routing algorithm (the analytic part) to find the shortest route between any 2 points The whole thing is wrapped into a well-designed user interface so that all this complexity is hidden from the user and we don't care because it just works, all of the time. Well, almost all of the time :-)

In my opinion, there are very, very few analytic tools that are truly accessible to business users without some configuration, packaging or guidance from an expert. If you are serious about solving a business issue that needs some analytic work, you need the tool and the mechanic to wield it.

No comments:

Post a Comment

Note: Only a member of this blog may post a comment.