There is a critical point in a spreadsheet’s development when that spreadsheet is first shared with someone that has never seen it before.
If the new user has a stake in the spreadsheet, then their response might be one of three: wow; howl; or shocked silence.
One reason that we work with the FAST Standard is to maximise the chances of a wow response. But, from experience, we have learned that often the wow response is a slow burn. F1F9 clients build up a sense of wow over time. As their requirements change and they get used to the FAST model upon which we collaborate, so they appreciate the flexibility, structure and transparency of the models with which we all work.
A howl is more likely from a client who has a non-FAST spreadsheet model that they offer up for rebuild. We understand their pain. However much our clients accept the long term benefits of FAST modelling, there is a leap of faith required in the short term. An existing model into which much time has been put must be offered up for dissection, analysis and reconstruction. That can be a painful experience.
So one thing we do to ease the pain is something called Interim FAST modelling. Interim FAST modelling is FAST modelling but undertaken in an existing non-FAST model.
There are three elements of Interim FAST modelling.
1. Make the existing model easier to review
Short formulas are easier to read than long formulas. One reason for long formulas is long worksheet names. So the first thing we might do is to shorten each of the worksheet names on each worksheet tab (Excel 2003: Alt O H R; Excel 2007/10/13: Alt H O R). The trick is to come up with something meaningful in three letters or fewer.
We will also identify the last column used in a worksheet and hide columns to the right of that column (Excel 2003: Alt O C H). We will also put in a placeholder label to mark the last row that is used on a worksheet. It is a simple psychological advantage to a reviewer if they know the boundaries of each worksheet – latitude and longitude.
2. Replicate existing model logic using FAST
Taking a particular section, we will replicate the model logic immediately beneath the original. We will go as far as we can to show how FAST modelling could work in the existing model: colour conventions, row anchored links and calculation blocks with one calculation at the end. We will also have a check to prove that the outputs of our interim modelling match the original outputs.
This is a powerful step for a client that is new to FAST modelling. They see their model logic made transparent; they begin to understand the use of flags to answer the question “when?”; they begin to understand why FAST models are so easy to review.
They will also see how long formulas are shortened and how few rows become many. In a recent training course, I took three lines of existing code and built an interim FAST model. The interim FAST model – built using FAST modelling techniques – used 30 rows. But it was far quicker and easier to review the 30 rows than tease out the model logic in the original three.
3. Transfer the interim FAST model to a FAST model
Non-FAST models tend to use columns in a different way to FAST models. So the final step of interim FAST modelling is to take the interim FAST coding and drop it in a FAST model. This tidies up the whole process: headings and sub headings can be set up; units and row totals can be slotted in and reviewed; quick graphs can be charted.
From our client’s point of view, the process of rebuilding a non FAST model is made clear. They can see, understand and challenge the journey that their original model is taking in becoming a FAST model. For us in F1F9, it is an effective means to achieve collaborative modelling and build an effective ongoing modelling relationship with a new client.
Comments
Thanks Andrew,
A very interesting article, particularly as I have joined an organisation that has several (non-FAST) models, and I’ve been trying to think about broaching the subject of change…