…rather use this rational approach
I have recently been party to yet another failed software implementation. This was specifically for a client’s service provider so, while I was not involved in the implementation, it did affect my client. It has caused much frustration for them and their customers and has had far-reaching financial impacts.
The worst part of this was that my client as well as their service provider were forewarned. Many suggestions and recommendations were made and, while we are now more than three months past go-live, the system is still not fully rolled out and is far from stable.
One of the reasons for this dismal roll-out is that a proper project approach was not taken, including not using a rational, objective approach to choosing the best possible software for their environment.
Science is quite clear: gut instinct is a good thing to base short-term decisions on, but our gut is not very good at taking the long-term into account. When the stakes are low – such as what colour shirt to wear today – go with your gut. But any more serious decision, such as implementing software, cannot be left to emotions. These types of decisions must be based on sound, rational, and objective thinking.
All hail the decision matrix
One way many people weigh up the various options is to list all the pros and cons. While this is a step better than going with your gut, this approach has one major pitfall: it often leads to double or even triple counting the same thing. For instance, one solution may be more cost-effective, so this is a pro, which implies that the more expensive option has this as a con. As you see: the same point, but counted twice.
So how do you ensure you select the best option for YOU and your organisation?
By using a decision matrix. This is a simple grid where you list all the possible solutions down one axis, and then rank them against a set of criteria.
So what should these criteria be?
Start with what the original requirements are, in other words, what need does the solution absolutely have to fulfill? Another approach is to list the problems that must be solved.
Let’s assume you are looking for a new warehouse management system. For certain environments, you would simply need to know how many of each type of product you have. For more complex environments, such as perishable fast-moving consumer goods, you would want to know how many of each you have, in which storage location they are, what their expiry date is, and which came in first, to ensure that you do not send the fresh stock out. So, two immediate criteria for a perishable FMCG environment are stock management at the bin location level and FIFO, or first in first out.
For complex industries, such as pharmaceuticals, you would want to know all of the same points as for the FMCG but also know what the batch number is for each item. You would also need to track which batch was sent to which customer. If, for instance, you manage medical devices, such as pacemakers, you need to know details down to the individual serial number for each single device. Thus you would have more criteria against which to judge the various options.
Something else you would want to consider to complete the list of criteria is to look at the risks associated with each solution, and other factors such as:
- labour impacts
- reliability
- quality
- environmental impacts
- safety and security
and of course, the total cost of ownership, not just the initial outlay.
The next step is the weighting of the criteria. Those that are most important are given more weight. You can now also add any other bells and whistles as further criteria, but ensure that their weight is never more than that of any of the absolute criteria.
Let’s use our warehousing management system as an example; perhaps one system uses AI to determine which products to count, for the daily cycle count. This is an amazing addition but should not be given as much weight as, for instance, the costs or implementation time.
Now that you have the grid for your decision matrix, you rate each solution against each criterion, with the solution that meets that criterion the best getting the highest rating and the solution that does not meet the criteria well rated lower. Then, once all the solutions are rated across all the categories, you can do a simple sum product of the weight of each criterion and the rating. A clear winner should emerge from this.
Of course, this is best done with all available stakeholders present. This allows them to add criteria that affect them and weigh in on the weight of each. This is very important so that the stakeholders know why the final solution was chosen. It will also ensure that no matter how cool the AI feature is, if you don’t need it, you won’t buy it, regardless of what the IT guy’s gut says.