🫡The pluses and minuses of multiplication

Designing 'smart salaries' Salary is the “gene of the company" which, upon examination, reveals much about the overall "health" of the organisation. The extensive experience (since 1993) of TRIZ-RI specialists in consulting indicates that there is scarcely any consultancy work that does not involve the creation of salary models for employees and/or the establishment of key performance indicators. More details ...

It is often declared, 'I shall pay based on the final outcome!' And just as frequently, we find ourselves puzzled by: 'Yes, but there are many outcomes…' What should we do?

Should we evaluate each one? How then do we associate each with a salary? How do we determine the 'primary criteria' from several? What sign should we place between them?

Furthermore, if we have to consider the results of several individuals, how do we prevent each one from being 'dissolved' in this 'collective assessment', and avoid favouring one over the other? How do we ensure that we don't miss the opportunity if an employee leaves just as a significant deal is being struck?

What will motivate the employee to consistently maintain a high standard? And what should be done when an employee 'falls short,' feeling that the punishment is too harsh?

These are the issues Alevtina Kavtreva and Ksenia Tkalich will explore.

Part 1. Signs

K. Tkalich: So, let's reread the announcement in the article and begin to answer...

A. Kavtreva: Let me remind you: we've faced a similar problem before. To assess not just luck but also effort, two categories were proposed for calculating the bonus of salespeople in the "Article about Salary": "Performance on turnover" and "Performance on number of sales".

This is because, in a shop, the sale of expensive items (like carpets and projector televisions) as well as simpler items (such as cheap wallpapers, coffee grinders, and kettles) are important.

To correlate these two criteria into one Overall Performance, we used multiplication:

Pluses:

The use of multiplication significantly diminishes overall performance when it fails to meet required benchmarks.

Furthermore, this operation compels the employee to maintain the overall level (it is highly improbable that anyone would be satisfied with the result of multiplying 200% by 10%, yielding 20% - please refer to "plus no.1" above).

It is an interesting fact that G. Altov and P. Amnuel employed multiplication to combine several criteria to compel authors to maintain the high overall level of ideas on the scale dedicated to evaluating science fiction books.

Minuses:

The employment of multiplication introduces the following risk: if the benchmark (e.g., for a startup) lacks an accurate forecast, it can lead to undesired mathematical outcomes.

Additionally, salaries might multiply due to significant overachievement of benchmarks, though the cause of this increase is not the employee's effort but rather a flaw in planning.

This is the flip side of the same coin: failure to meet a benchmark results in a drastic reduction in outcomes, whereas exceeding it leads to a significant increase.

The word "danger" is intentionally quoted to highlight that the issue stems not from the multiplication itself, but from inadequate planning.

K. Tkalich: And once more, we should ask: can we eliminate this "negative" effect from multiplying (refer to the "minuses" mentioned earlier), if the company is growing and at some point it becomes difficult to set accurate standards? I should note that an arithmetic mean does not have this drawback. It provides, as we saw earlier, "a smoother feedback" when a target is not met on one indicator. This means, whether there are significant increases for exceeding targets or, conversely, a decline, there will be neither "pluses" nor "minuses".

A. Kavtreva: However, this "virtue" of the arithmetic mean is also its drawback. We might "focus" on one of the criteria and become complacent with the same "smooth" (arithmetic mean) calculation.

K. Tkalich: What can we do?

Part 2. Benchmarks and Thresholds

A. Kavtreva: The solution to this issue is described above—by implementing minimum performance. The minimum ("threshold") performance is the lowest acceptable level of performance, below which performance is deemed unsatisfactory.

In the "Article about Salary" it is recommended to set the minimum performance at approximately 70% or higher (but certainly not below). Performance below the threshold value is considered unsatisfactory.

K. Tkalich: Counter question: "What if the employee does not meet the minimum performance threshold? Should they be deprived of a bonus?"

A. Kavtreva: In some situations—yes, and in others—no. Let’s consider two scenarios:

Situation 1:

1.1. Let’s suppose the fixed part of the employee's salary is 8,000 coins, and the variable benchmarks bonus part is 2,000 coins.

Note: when we refer to the "benchmark bonus," we mean the part that will be paid for achieving 100% performance.

1.2. One or more indicators have not reached the minimum performance threshold. In such a scenario (as described in point 1.1), the variable part shall not be paid.

Situation 2:

The fixed part of the employee's salary is 2,000 coins, and the bonus part is 8,000 coins.

2.1. One or more indicators have not crossed the threshold value. Not paying the variable part in this situation would be too severe a punishment. The employee would have no means of subsistence.

Solution: In such a scenario (as described in clause 2.1), the overall performance is equated to the lowest value.

Let's illustrate this with an example:

Let's suppose the employee has two performance indicators, for which, in the current month, they achieved the following results:

Performance on turnover = 65%, and
Performance on the number of customers = 110%. Then the
Overall Performance = the lowest indicator, which is 65%.

2.2. If the performance for both indicators was higher than the minimum value, the overall performance would be calculated using the "arithmetic mean."

For example:

Performance on turnover = 84%, and
Performance on the number of customers = 110%. Then the
Overall Performance = (Performance on turnover + Performance on the number of customers)/2, i.e., 97%. That is how we address the issue.

K. Tkalich: But this is only true for situations where the results of the indicators are equally valuable. What if they are not? What should we do if one indicator's result is more important than others? Or, for example, if the employee has more influence on it?

Part 3. Weights

A. Kavtreva: Then, for each result criterion, we shall introduce a "weight coefficient". The formula appears as follows:

Overall Performance = "Weight" (value) of the 1st KPI × result of the 1st KPI + "Weight" of the 2nd KPI × result of the 2nd KPI.

I shall give an example:

A manager sells goods with a price distribution ranging from USD 2,500 to USD 3,200 per unit. However, growth in turnover is achieved due to an increase in the number of sales. Consequently, the significance of the number of sales, in this context, is greater (let’s compare the number of sales to turnover - 70:30).

The overall performance will then be calculated as follows:

Overall Performance = 70% × Performance on number of sales + 30% × Performance on turnover.

This is not akin to "average" performance, and what is interesting is that:

When the "weight" of the indicator is significant (70%, 80%, or 90%), we encounter the same unwanted consequences as for multiplication (see Part 1, "Signs").
When the "weight" is lower or near 50% (in essence, it represents the same "average"), the disadvantages are the same as for the "arithmetic mean".

Of course, when calculating the "average weight", the same minimum performance should be used. That is, follow the steps described in the Part 2, "Benchmarks and Thresholds".

K. Tkalich: I have a question: "What shall we do if the "weight" of an indicator is very low, and the minimum performance has not been reached for that indicator?"

A. Kavtreva: Let me give one more example:

In some companies, there are minor delays in payment from the customers. These do not create significant problems, but we do not want poor practices to develop.

As a preventive measure, a performance on accounts receivable turnover is introduced into the salary structure of managers, which carries a low "weight". The introduction of such an indicator serves more as a reminder.

In this instance, we can agree that the Overall Performance will not equate to the "worst" (lowest) indicator if the minimum performance is not reached in an indicator with low "weight".

Let’s illustrate what we have said:

This employee has two performance indicators: KPI1 and KPI2.

Suppose

The Overall Performance = 90% × KPI1 + 10% × KPI2 (that is, the "weight" of KPI1 is high, and that of KPI2 is low).

If the indicator with a high "weight", KPI1, does not achieve the minimum performance of 70% (for example, it is KPI1 = 68%), and KPI2 is performing well (the indicator with low "weight" KPI2 = 100%), then we can equate the Overall Performance to the "worst" value.

Thus, in this situation, the Overall Performance = 68%.

If the situation is reversed (KPI1 = 100%, and KPI2 = 68%), that is the minimum performance is not reached for the indicator with low "weight", then the calculation for "average weight" is done using the formula:

Overall Performance = 90% × KPI1 + 10% × KPI2 = 90% × 100% + 10% × 68% = 97%.

K. Tkalich: Is this not a lenient penalty for a low value of KPI2?

A. Kavtreva: If such a problem arises, we shall increase the "weight" of this indicator.

Part 4. Steps

K. Tkalich: People often ask: "Can a progressive reward system address the same challenges as multiplication or a high 'weight'?"

A. Kavtreva: The "progressive system" aims to enhance the "sensitivity" of achieving each percentage of the benchmark. This is because employees, having reached a certain milestone, often feel that "enough is enough," relax, and do not aspire to exceed it.

K. Tkalich: Indeed, with a progressive reward system, the higher the productivity, the "richer" each percentage becomes. This is believed to foster a growing interest in high performance.

A. Kavtreva: However, in reality, people often reject the "progressive system."

K. Tkalich: I wonder why?

A. Kavtreva: For several reasons:

It is often misinterpreted not as an "increase in reward" for each percentage of task completion, but rather as an increase in remuneration based on a percentage of profit or turnover. Smart employers are hesitant to base compensation on these metrics. (Read more about this in the article "Nine Times You Can't.")
Moreover, in this instance, only one indicator is incentivised, whereas several are actually necessary.
In essence, only "luck" is rewarded. Any significantly large order results not just in a high, but a very high salary.
Furthermore, I dislike the "progressive system" because it fails to motivate the very "middle performers" for whom it was originally intended (see the chapter "Stimulating Angels, Not People"). Top employees—yes, they are affected. In fact, only their results fall into the premium range, where motivation is enhanced. However, the system was not designed for best performers, but rather to elevate the average ones. This does not occur.
Let me add another significant reason: the sharp disparity in wages during times of failure and times of growth, under heavy loads and during forced downtime. The "progressive system" is fundamentally similar to multiplication, but only "one-sided."

K. Tkalich: What does "one-sided multiplication" mean?

A. Kavtreva: "Multiplying" indicators, unlike the "progressive system," address problems using various indicators simultaneously. Not only "luck" is rewarded, but importantly, "effort" (achieving good results across all given indicators is not the same as excelling in just one).

K. Tkalich: A progressive-regressive reward system (if referring to the "increase in price" for each percentage of the plan) has one advantage over using multiplication for various outcome criteria: it makes the fluctuations in bonuses less "severe."

A. Kavtreva: Here, it's worth noting that results in multiplication "suffer" from a sharp increase even with the presence of a weighting factor. To some extent, this is the price paid for the desire to increase "sensitivity," compelling the employee to "endure" every percent of the benchmark.

K. Tkalich:

And sometimes, those for whom a rise was intended begin to 'practice deceit' by 'allocating' results to one specific person so this 'best' can earn a lot and then 'share' with the others, or begin to 'concentrate' the results within one reported month, not 'disclosing' the 'earnings' from the previous period. Though, this may occur in any system that strengthens the interest of the employee.

A. Kavtreva: Yes, but such 'manipulations' are not as concerning. This problem can be resolved.

K. Tkalich: I'm curious to know how?

A. Kavtreva:

By implementing a standard system of control and reporting;
By introducing 'counterbalances' (if there is an anticipated raise in salary for results, there should also be a reduction in salary if the results are lacking);
Often, introducing a quarterly bonus for 'consistent high performance' helps. If an employee’s average performance over a quarter is below, for instance, 90%, then the special bonus is not awarded.
Another solution involves categorising employees. The category depends on the quarterly (or biannual) performance and determines the base part of the salary from which all calculations are made.

Every quarter (or biannually), the category is reviewed based on work performance and assessment. An employee 'who loses customers' risks lowering his category, which leads to a decrease in the constant part (basic) of the salary (in the long run), and with it, the variable part (which is calculated as a part of the basic salary).

K. Tkalich: To conclude on a lighter note: What other arithmetic operations could we consider? For instance, why not explore the use of logarithms?

A. Kavtreva: The general rule is as follows: In a salary model, it is best to use only criteria that are understandable to the employee and arithmetic operations. The main principle of a workable salary system is its 'transparency.' …I recall the 'exotic' situation when, while developing a salary structure for his department (chemical-technological), a director analysed all possible arithmetic operations and chose 'the geometric mean' because 'Any technologist understands this term,' he explained.

The authors thank Sergei Sychev and Galina Vladimirova (TRIZ-RI Group) for their assistance in working on this article.

Designing 'smart salaries'

PreviousMinimum performance NextA linear way of getting efficiency

Last updated 1 year ago

Was this helpful?