How to Forecast Sales Performance

When I worked at BusinessObjects, our fearless marketing leader (and my personal hero) Dave Kellogg, introduced us to an impressively accurate forecasting formula.

He had found that the sales forecast, as measured in our CRM system on the second week of the second month of the quarter, would equal to the sales number our team would deliver that quarter…98% of the time.

His method was so wickedly accurate, it became a legend with the name of “Triple Witching Day” or TWD (“Triple Witching Day” is a financial term used to mark the expiration date of stock index futures, stock index options and stock options. That term is unrelated, although it occurs 4 times a year too!).

His method was so wickedly accurate, it became a legend with the name of “Triple Witching Day” or TWD.

This formula does not work for every company. It indeed depends on many factors: the reliability, consistency and discipline of your sales teams and of course, market conditions.

However, I became obsessed about the factors that predict sales performance many years after I left BusinessObjects and continued to use TWD whenever relevant and accurate.

When I joined Microsoft and helped build the execution engine that ultimately drove $1B in analytics revenue, I further learnt the intricacies of running a reliable and predictable forecast at scale. TWD is helpful in aggregate. But it can be harder to use team by team, geography by geography…etc. Microsoft runs a very diverse business, across many countries, through multiple channels and forecasting accurately can become quite complex.

So, we devised a system that combined TWD with my favorite visualization…the quadrant. And we called it, the Business Performance Quadrant (BPQ for short).

I’m a big fan of quadrants because they are great simplifiers. Working in tech with high I.Q. individuals, I’ve found that complexity is too often the rule. Complex design is comfortable for smart people.

Yet, I’ve found that truth only comes from simplicity and that there should be not pride in ‘complexifying’. Simplicity is harder to achieve. In the words of Leonardo da Vinci: “Simplicity is the ultimate sophistication.”

“Simplicity is the ultimate sophistication.” Leonardo da Vinci

The Business Performance Quadrant (BPQ)

The Business Performance Quadrant is eerily simple. It combines a simple set of leading and lagging indicators. My belief is that you can’t predict the future by using either in isolation, you should combine both.

Leading Indicator: Forecast

The X axis covers the % of the declared forecast in relationship to the expected forecast (say you expect 100 and the team is forecasting 90, you’d be at 90% forecast). You can pick any time of your quarter to measure this number and you can measure it as many times as needed. As mentioned above, if your performance resembles that of BusinessObjects’. TWD could be the right date for you.

Lagging Indicator: Past Performance

The Y axis covers past performance. Did the team miss, hit or surpass its forecast in the most relevant past performance. “Most relevant” here means that shouldn’t always pick the same quarter the year before for instance. Remember that you are trying to measure the progress of your team. You’re not worried about seasonality as much as you’re interested in assessing the ability for your team to develop its method.

The beauty of the BPQ is that it combines reality (past performance) with aspiration (forecast) to create judgment…in a repeatable manner. The boxes should be self-explanatory but here is a quick overview for how to read them.

1. Most Likely To Achieve: These are teams who have achieved forecast (or more) and whose forecast is set to meet or over achieve their goals.
2. Inspect for Over-optimism: These are teams who have NOT achieved forecast and whose forecast is set to meet or over achieve their goals. We used to call this category “HELP”. There are many reasons why these teams could fit this box: they might be overestimating their ability to execute or they might have inherited a strong forecast, have lost a leader and need more help to get across the line.
3. Inspect for Sandbagging: These are teams who have achieved forecast (or more) and whose forecast is set to miss goal. We also used to call this category “HELP”. There are many reasons why these teams could fit this box: they might be under declaring their forecast or lack forecast discipline (aka sandbagging) or they might truly be at risk. Regardless, they need help!
4. Most Likely to Miss: These are teams who have NOT achieved forecast and whose forecast is set to miss goal. They also deserve help. If you don’t trust their ability to reach goal because of their projections or their past, you ought to do something about it!

Now, that you have the model set, here is what you can do about it:

• Plot each division, country or channel on the quadrant.
• Assess the % of each ‘dot’ in relationship to your total goal.
• See the state of your overall and individuals goals, as well as how you should programmatically solve each box.

Below how this could work across multiple geographies. I hope this helps!

The Model Applied (Fictional)

Note that the below is an example of additive, absolute and non-weighted values. If you’re adding up percentages, you’ll obviously have to weight them to get your total. Oh, and don’t mind France’s position, that just happens to be ‘confirmation bias’. :)