Go to Top

How to build an efficient and meaningful What-If analysis

What-if analysis

What should a What-If Analysis do?

We are all trying to predict the future. The most valuable type of knowledge you can obtain is the outcome of events that have yet to unfold. No one has a crystal ball; however, businesses can take steps to plan for future outcomes through data and analytics. Alternatively, a poor scenario analysis can be catastrophic to business planning. Any time action is taken to capitalize on a future environment and that future environment is not realized, it becomes a huge waste of time and money.

What steps can be taken to ensure the what-if analysis is measurable and (more importantly) explainable? Without getting extremely technical with platforms and situational modeling, here are some general steps you can take to ensure that your analysis will have a strong backbone on which you can rely.

1. Keep reality in the analysis.

A scenario analysis should do 1 of 2 things:

  1. Apply a scenario to historical data to show what might have happened if we (the business) had done something differently and where we might be now.
    1. This approach is useful when historical data is very repeatable (cyclical), and therefore past decisions and current environments can be modeled against this data.
  2. Apply a scenario to a forecast to provide alternative futures that can be used for planning.
    1. Usually this is either a “best/worst case scenario” analysis, sometimes with varying levels of probability for each scenario.
    2. This approach is useful when you have a dialed-in forecast in which the data driven 6 months from now is very important for making business decisions today. Functions that benefit from this include resource planning, inventory planning, etc.

You’ll notice in both of the aforementioned situations the common thread is the start data: historical actual data or dialed in forecast.  The latter is obviously tougher to obtain, and I’ll get into the challenges of that example in a bit. But I want you take away the importance of BOTH situations relying on a baseline. In science, they call this the control group.

Having an understandable baseline implemented into a scenario analysis is key, because it helps largely with the goal to have a measurable and explainable what-if analysis. You need a point of reference to explain how the scenario played out, i.e., measurability. You also need a point of reference that you currently understand, as it will undoubtedly help explain the scenarios. It’s important to note here that this segment applies to a wide range of audiences, as it’s essential in the building of a what-if analysis (from an analyst’s level) but also important in the presentation of the analysis (management level).

2. Keep the variables minimal and conservative.

Complexity is sometimes a necessity with any analysis, but it can also be an enemy. A what-if analysis is literally an analysis of variables, but to keep the analysis measurable and explainable, we need to be careful with the questions we ask.

  1. If possible, keep the variables short term.
    1. The further out in time we ask “what-if” the less likely the environment around that question will be aligned with the current data at hand.
  2. Less is more
    1. The what-if analysis is about asking questions. But too many data alterations to satisfy these questions along the way end up making the outcome tougher to explain. Even if the outcome is accurate, it’s likely tough to explain and can always be made more concise. Granted sometimes there is no way to get around a complex scenario model without many variables, but the majority should and can be minimized on this front.

3. Ensure the outcome is visualized with the audience in mind.

Steps 1 and 2 apply to this last guideline. Whether it’s simply data in a table across a timespan, or you utilize data visualizations, such as charts, graphs, dials, etc., it’s incredibly important to know who you are presenting the findings to and that the visualizations are also measurable and explainable.

  1. Make it clear where the baseline exists in the end.
    1. Echoing Step 1, it needs to be very clear in the visualization where the baseline (reality) exists in comparison to the outcome of the scenarios (measurable).
  2. If possible, show where the data altered in a timespan due to a what-if change so that you can visualize the path the data took after that point.
    1. Again, more what-if questions make the overall task more difficult. However, if you are visualizing the outcome in a literal data table with a timespan, or in some sort of chart, try to show where the change occurred and how it affected data from that point forward (explainable). If the questions you are asking would apply a variable to the analysis as of now, then this piece isn’t needed to visualize as much, as it is an assumption of the analysis.

Following these generalized steps will ensure, regardless of the analyses complexity, that you have a strong foundation to build on from start to finish. Apply your situation to each topic. There will certainly be quirks and variability to these types of analyses across the board, however sticking to these steps will generate efficiency more than anything else – efficiency in the build of the analysis and in understanding and communicating the results.

Have you been tasked with completing a what-if/scenario analysis? What intricacies are involved that might put one or more of the guidelines above in jeopardy?

Request info about financial analysis and reporting tools from our finance and business intelligence team.

, , , , , ,

Leave a Reply

%d bloggers like this: