As an antidote to my "top 5 things to AVOID", here are my top 5 hygiene factors to look out for when building or commissioning a financial model.
1 - Integrity
This has to be number one, because if the model doesn't balance then it's clearly not ready to be relied upon!
A model should include checks for basic things like "Does the balance sheet balance?"
Also "Do reserves match the P&L?", and "Does the cashflow match the balance sheet?"
If any of these things don't work, or worse if the model doesn't include checks for them(!) then don't rely on that model until it's sorted!
2 - Flexible
If a model is not flexible then it's not much of a model, and may as well be static analysis.
A good model will allow the user to update their assumptions easily, but a GREAT model will include a scenario manager
Using a scenario manager is a perfect way to hold several sets of input assumptions within the same file, to allow for comparison. This means that several options can be compared and there's no risk of version control issues as they are in the same file.
3 - Simple to use
A model is often a decision support tool and should be designed to simplify processes and help support decision makers in their role. If the model is not easy to use then there is a risk it won't be used, or will be used incorrectly.
Standalone user guides can be great, but what's even better is a model which is simple and intuitive to use
If the model includes the relevant information in the right place, then questions about "what does this do" and "how do I do that" are immediately answered via a note adjacent to the cell, or just a click away in an integrated guide.
So, make the input areas easy to find, make the model easy to navigate and include some orientation information when the model is opened, and it'll be smooth sailing for your users.
4 - Clear outputs
Time is precious and when using a model under pressure it's vital that the outputs are clear and insightful. A good dashboard with appropriate column widths, font sizes and even number formatting can go a long way to making sure that a model is easy to interpret.
5 - Easy to update
If a model is not easy to update, then there's a risk that it will just not get updated often enough - leading to incorrect decisions based on out-of-date information.
If a model is designed to link to a source system, or to take a simple copy/paste of data in the native format from a supporting system then the time taken, and risks of incorrect inputs are hugely reduced.
Something as simple as having a paste area set up ready to take in source data from systems which can then be cleansed automatically by the model will be a big benefit to the model users.
Comments