Deseasonalization of Strategic KPI Measures
Mihai Ionescu, Senior Strategy Consultant, Owner Balanced Scorecard Romania.
Difficult to spell as it might be, the term 'deseasonalization' (which Google learned to translate, as well ... 'désaisonnalisation', in French or 'dessazonalização', in Portuguese), is an important term in Strategy Execution. More subtle for many of us, but definitely important. If you are uncomfortable with the term, you can use an acronym, similar to the deoxyribonucleic acid, which we call DNA. So, you can use DSZ as an acronym for deseasonalization. The term is often used in retail management, describing the sale of items at discounted price, once their peak sales season is over. But in our case, it means
Taking the effects of seasonal, periodic or occasional external influences out of the Strategic KPIs measurement
In operational management, a KPI has to reflect all factors, internal or external, that may affect its measured values. That is simply because we need to react and take the required corrective actions, whatever the causes of the variations or fluctuations might be. Examples. We need to turn to alternate supply sources when our main supplier is having temporary outages or delivery problems, leading to lower levels of input stock for our operations. We need to employ new people when an important group of employees leave the company at the same time, leading to a decrease of personnel readiness level. We need to re-route the production flow when the spare parts for an important piece of malfunctioning machinery could not be delivered in time, leading to increased delays in production output. And so on.
But with the Strategic KPI Measures, the situation is different. The main reason for which we employ them is to quantify the success of (or trend towards) the accomplishment of Strategic Objectives. And we know that the accomplishment is based mainly on the effects of the Strategic Initiatives that are planned and executed for changing, improving or transforming the relevant operational processes that are targeted, for each objective. Before going further, let's clarify something. Certain Strategic KPI Measures have to reflect external influences. Or to be more precise, they have to measure processes that involve external parties: Customers, Suppliers, Partners, Regulatory Authorities, etc. But even in their case, we are talking about the quantification of our relationship with them, not about their actions or behaviors that are determined not by their interaction with us, but by third parties (upon which we have no control). Examples. The launch of our new product has been apparently less successful, but it coincided with the similar product launch by the market-dominant vendor. Our new supplier policy seems to give worse, rather than better results, but it coincided with the merger of our two main suppliers, which lead to the change of their negotiating power. Our bank savings product, in a certain currency, seems to attract fewer customers, but it coincided with a sharp variation of the exchange rate for that currency. Our compliance level to a certain regulatory authority enforcement rule has decreased, but it has been due to the change introduced lately by that regulating authority, valid for all players in our industry. And so on.
The need for Deseasonalization (DSZ)
So, what is the problem with external influences and the Strategic KPI Measures? Firstly, they may falsify the relevancy of our KPI measured values, which no longer show the effects of the changes, improvements or transformation resulting from our planned actions (the Strategic Initiatives), but a combination of those effects together with the external factors' influence effects. So we don't know anymore whether the KPI Measures show us the effects of our strategically planned change actions, or those of some external factors of influence.
Secondly, because we may decide to take corrective actions (mainly upon the Strategic Initiatives suspected to be responsible for the lack of measured effect), actions which are based on irrelevant measurements. And that happens not because we have picked the wrong KPIs for our Objective, or because they are not properly correlated with the corresponding Strategic Initiatives, but because we haven't deseseasonalized them (or DSZed them, to use the acronym).
It's clearer now that DSZ is about the separation of the effects of our change, improvement or transformation actions from the effects that have external causes. Or, in other words, applying of a DSZ correction to a Strategic KPI Measure.
The Deseasonalization Factors (DSZ Factors)
DSZ doesn't mean just seasonal. Of course, this is the first meaning that comes to our minds, but it has actually a wider applicability.
» Seasonal DSZ Factors
Examples. We know that the December sales in retail are on average three times the average of the monthly sales in the other months of the year. Or that we harvest crops once or two times a year, when our harvest storage levels are the highest. Or that for ice or barbeque products we sell around 75-80% of the annual volume in the summer/warm season (in the northern hemisphere, in temperate climate). Or that the seat occupancy rate of our planes is higher in certain holiday periods of the year, going to nearly 100% during four-five periods annually, irrespective of the occupancy rate during the rest of the year. Or that absenteeism for medical reasons occurs 2-3 times more often in the cold season, when flue or other seasonal diseases produce more significant effects.
» Periodic DSZ factors
The periodic DSZ Factors have an influence similar to that of the seasonal ones, but they are not related to any specific seasons or periods of the year.
Examples. We know that our products/services have a life-cycle, driven - if applicable - by the Law of Innovation Diffusion, with early adopters, early majority, late majority and laggards (see the second part of Simon Sinek's presentation, for more details). Therefore, depending on the overlapping of the life-cycles of our multiple products/services, we may measure sales volumes that are influenced by the timing of our products along their life-cycle timelines. We know that, depending on the specific price elasticity, when we run a price promo campaign for our products/services (may be price discounts or volume discounts), we'll get increased sales by a certain factor. We know that during the first phase of the introduction of a new manufacturing technology, or management system, or IT application, the learning curve will determine more errors/waste than in the remaining period of its use. We know that closer to personal evaluation deadlines, people tend to pay more attention to their rated performance parameters (what we call the toothpaste tube effect).
» Occasional DSZ factors
Whilst the seasonal and periodic DSZ factors can be anticipated in terms of occurrence timing and expected influence effect level, the occasional DSZ factors cannot be predicted in terms of timing, although for some of them we may estimate their influence effect levels.
See the examples given above, under the definition of DSZ (the operational KPI influence factors). In the majority of those examples, we can estimate the effect level(because it happened before and we know that it can happen again), but we cannot anticipate neither when the triggering event will occur, nor its periodicity (with a relevant level of accuracy).
An important note here is about the risk events, which resemble the occasional DSZ factors' triggering events, but are not the same thing. For example, we cannot consider that the launch of a similar product by a competitor is a risk event. However, it is an event that triggers an occasional DSZ factor (the reduction by a certain factor of our product's/service's sales numbers, due to the fact that the customers pay more attention to the novelty of the new product/service launched by our competitor).
The Correction Quotient for DSZ Factors
I'm sure that you have noticed that in all the examples above we are talking about the anticipation of the DSZ Factors, either in terms of (a) occurrence periodicity, (b) duration and (c) influence effect level, or at least in terms of influence effect level. This is an important aspect, if we want to take out the influence of the DSZ factors from our Strategic KPI Measurement.
In an ideal case, we could calculate a Correction Quotient (correction coefficient) for each affected Strategic KPI Measure, applicable during the active period of the influencing DSZ Factors, where the value of the Correction Quotient (CQ) is proportional to the influence effect level. But we have two problems here: (1) we cannot always anticipate accurately enough the periodicity and the duration of the DSZ Factors' influence and (2) we cannot always anticipate accurately enough the level of their influence.
But we don't have the luxury of capitulating and doing nothing about the DSZ Factors, nor can we take the approach of 'we'll figure it out on a case-by-case basis if and how much have external factors influenced the measured values'. So, what do we do? The historic data of the parameters that represent our Strategic KPI Measures come to the rescue. Or, at least, they can help.
True, sometimes it might be difficult to estimate, for instance, the increase in our sales volumes for a certain product during our planned price promotions that we have scheduled for this year, based on the discount levels, because we are talking, in most cases, about multiple factors that have to be considered. But something like this is not impossible, considering that we have run many such promotions in the past, for the same or for similar products, and we know by how much have the sales increased, in each case (our Correction Quotient - CQ). More importantly, in certain cases or in certain periods of the year, we have no other choice but to use DSZ, if we want to be able to understand our Strategic KPI Measurements.
The artificial Strategic KPI Measures trap
Taking the last example, let's assume that we can calculate that for the sales of our product/service A, we can apply a correction (CQ) of 1.10 and 1.15 for the sales volumes during the two discount promo campaigns that we have scheduled this year, by running a statistical analysis on the results of the the similar campaigns from the last years. Great! So we know that if we sell in April and in May (during the first scheduled promo campaign) 1,000 units of product per month, 100 of them are attributable to the effects of the promo campaign. So, 900 would be the volume sold if we wouldn't run the promo campaign.
But now we are in trouble! Why? Because we have two versions of the same KPI (# Sales Volume). A real KPI (1,000 units) and one that is an artificial KPI, resulting from the application of the 1.10 correction, a KPI that we use for our strategic performance monitoring purposes. So, one KPI version for operations and one KPI version for the strategy. This doesn't make sense! You could imagine hearing questions like 'Which Sales Volume KPI are you talking about, the operational one or the strategic one?'. We could drive people nuts if we take this approach ...
The solution: Deseasonalizing the Targets
That's when someone had a brilliant idea. What if we keep a single version of the KPI (the 1,000 units real value, of course) and, instead of applying the QC correction to the measured value of the KPI, for getting the strategically-relevant value, we would apply the QC correction to its Target. Bingo!
So, the Target for the KPI in our BSC Scorecard is of 925 units for April and 950 for May. Apparently, by selling 1,000 units, we have surpassed the targets in both months. But, wait a minute. The Target was supposed to be reached through the effects of a Sales Development initiative that included the improvement of our product positioning and of the sales procedures, the improved of shelf presence in more relevant store chains, not by harvesting the results of promo campaigns!Let's see how applying the correction to Targets works. Take a look at the diagram.
The animated version of the diagram can be found on YouTube.
The targets without DSZ applied are colored in blue. The gap between the January target of 900 units sold per month and the the December target of 1,050 is expected to be closed through the effects of the "Sales Development" Strategic Initiative that is rolled-out in the second and third quarter (Q2 & Q3).
However, there are two promo campaigns scheduled to run in April and May and between October and December, with CQ's of 1.10 and 1.15, respectively. If we would have planned no Strategic Initiative, and the targets would have been of 900 units for every month and if the targets would be met, we would get sales of 990 units in April-May and 1,035 units in October-December.
But we have a Strategic Initiative, planned to run from March until September and, for this reason the targets are increasing during that period with 25 units/month. The question is: How would we know if the initiative produces the expected effects in April and May? Should we wait until June to find out, when the effects of the first promo campaign would disappear and the KPI values measured would reflect again only the effects of the initiative?
Unfortunately, we cannot wait, so we take the DSZ route: we perform the deseasonalizing of the targets in April and May, by applying the Correction Quotient of the first promo campaign to the targets in those months, resulting the DSZ targets of 1,017 and 1,045 units sold, instead of the targets without DSZ, of 925 and 950 units. By doing this we can tell with better probability if the initiative has produced the expected effects during those months. Of course, the accuracy of DSZ depends on how well we have calculated the correction coefficient for the promo campaigns, but without DSZ we would be completely blind about the effects of our Strategic Initiative between March and June.
An anecdotal general is supposed to have said:
There is nothing worse than bad news but bad news late.
So, instead of finding out only in June that there is something wrong with our Strategic Initiative (it doesn't produce the expected effects), we can monitor it continuously, even during April and May, with a level of uncertainty, of course, depending on how well we have calculated the QC and applied the DSZ.
Instead of conclusion, I would say that the example above is a simplified one. In real life, we usually have a more complex mix of DSZ Factors that have to be considered. For example, in December we have the cumulative effect of the second promo campaign, combined with the effect of the increased demand that is specific for that month of the year, therefore a combination between of a periodic DSZ Factor and a seasonal one.
Fortunately, since we have no Strategic Initiative active from October onwards, the DSZ during the last promo campaign serve no practical purpose. But remember that this was just a simplified example.
Looking forward to feedback on the specifics of this topic.
P.S. One more question: Have you deseasonalized your KRI Measures (Key Risk Indicators)? If not, you should consider using the same methodology presented above for the KPI Measures. Integrating the Strategic Performance & Risk.
#DeseasonalizationofStrategicKPIMeasures #KPI #strategy #MihaiIonescu #BalancedScorecard #Management #strategicmanagement #strategyexecution #strategicKPI #MIhaiIonescu #KPImeasurement #deseasonmalization #DSZ #KeyRiskIndicators