The quality of your model depends on the quality of your drivers

When building a cost model, the main concern is ensuring that the model "balances," meaning that all inputs match the outputs, ensuring nothing is lost along the way and all costs are allocated. While this is a basic validation for a cost model, a balanced model is not necessarily an indicator of quality. The quality of a model is closely related to the logic applied to it, the cost allocations among each stage, but most importantly, to the quality of the drivers.

But what are cost drivers?

A driver, also known as a conductor or allocator, refers to the criteria used for distributing a cost in an allocation. It always points to a causality or a demand for the cost. The best way to understand this is with an example. Suppose you have a human resources department that wants to distribute the cost of the recruitment process among the organization's ten departments. It could simply divide the total recruitment cost by ten, giving each department one-tenth of the cost. Does that seem fair? Probably not, as not all departments needed to fill vacancies. A more objective approach would be to use a driver that assigns the cost proportional to the recruitment requirements of each department. In this case, the driver would be the number of people recruited per department, meaning the department that recruited more people would receive more cost, and vice versa. Departments that did not recruit anyone during the period would not receive any cost from the recruitment process.

>> What is ABC costing? what advantages does it have over other costings? <<

How does the driver help direct the cost?

Driver values are used to distribute the cost proportionally in an allocation. For example, the cost of Internet service in a company could be distributed based on the consumption in GB that each department has. This makes the cost each department receives fairer and objective, with Department C absorbing more due to its high Internet consumption.

Which are more important drivers or allocations?

If you investigate the literature related to this topic, it is likely you will find authors suggesting that allocations are more important than drivers. The reasoning is that the allocation determines whether a cost should reach a destination, which is correct, and I agree. However, there is a difference between academic theory and reality. Practically, although allocations are decisive as they form part of the cost model structure, we, as modelers, can add or remove allocations at will. The allocation is entirely within our control. If, after completing a model, a certain allocation is determined to be inappropriate during the review phase, it can simply be removed, resolving the issue.

What about the drivers?

Unlike allocations, where we have full control, driver information is not always available. Over the years, I've seen that when designing a cost model, we often hit a wall called "lack of data." Sometimes, the necessary information has never been collected, making it impossible to find the driver's value when needed.

The ideal driver vs. the real driver

Suppose we select the driver called "Hours dedicated to training per department." It's the ideal driver as it distributes the cost in the most objective way.

When searching for the data of the ideal driver, we encounter several scenarios:

1. The information may not exist at all,

2. It may exist but is on paper and the effort to collect it exceeds the benefit,

3. It may exist in a system from which it can be extracted but is outdated, or

4. It may be up to date but too aggregated and not detailed enough.

All these cases force us to switch to another driver that meets the following rules:

The information exists, is up to date, is detailed enough, and is easy to obtain.

A driver that has these characteristics is a driver we can use, in other words, a real driver.

In the example above, the ideal driver "Hours dedicated to training" could be switched to the real driver "Number of people trained per department." This wouldn't be as accurate as the ideal driver, as in Department A, one person could attend a three-hour training, while in Department B, another person attended several courses totaling 50 hours.

By using the driver "Number of people trained per department," since each department trained one person, both departments would receive the same cost.

We see that in both cases, the model balances as the initial cost is distributed 100%. Moreover, in both cases, the allocations remain the same, with no variation. However, the quality of the distribution is compromised because the real driver significantly changes the cost distribution compared to what would have been with the ideal driver.

>> Is Excel Cost Modeling a Path to Failure? <<

What to do then?

In the early runs of the model, we will have several ideal drivers that we will need to replace with real drivers for the reasons mentioned earlier. Therefore, it is advisable not to make decisions based on the first versions of the model. In such cases, it is recommended to select the real driver that most closely matches the ideal driver and simultaneously start collecting information for the ideal driver. This may require investments in software or equipment that can capture this information.

>> How to maximize the Return on Investment (ROI) in your Projects <<

Conclusion

It is important to recognize that although a cost model is correctly balanced (meaning that the inputs and outputs match), this alone.

does not guarantee the quality of the model. The true measure of a good cost model lies in the precision and relevance of the cost drivers used. These drivers must not only be suitable for the task of cost allocation but also be supported by accurate and up-to-date data. This becomes even more critical in an environment where information is not always

available or is difficult to collect. Modelers often face the choice between an ideal driver and a real driver, with the latter being implementable with the available information.

Therefore, it is advisable not to make hasty decisions based on the first versions of the model. Instead, it is preferable to select the real drivers that most closely approximate the ideals and work parallel to collect the necessary data for these ideal drivers. Investing in technologies that facilitate data collection and analysis can be a wise strategy to improve the quality of the cost model in the long run. In the end, a well-founded cost model not only provides transparency and objectivity in cost allocation but also becomes a strategic tool for decision-making in the organization.