Performance Indicators Design — Part 1
Prologue
When we need to monitor a process, whatever it is, we often use indicators: measurable data structures that represent specific properties of that process and its entities (inputs, outputs, and neighborhoods) in a well-defined space-time. The synergy between the business administration areas and the production and service processes “canned” the definitions and construction of indicators, being today the [Key-]Performance Indicators, standardized and prefabricated structures (market share, Lead Time, Productivity, Mean Time Between Failures, etc.) to which business and processes undergo to be monitored.
The specific objective of this series of articles is to establish a methodology for creating process-oriented indicators to be monitored, either to meet the demands of the organizational strategic plan or simply to meet operational goals. For this, we start with a downgrade of the pre-established concepts about indicators returning to the primordial state: the data.
Indicators as Data Structure
Indicators, from the existential point of view, are data structures that contain within themselves a quantity of information obtained and organized in their abstraction layers.
Primitive Indicator: Consists of a homogeneous data structure with directly measured values such as “temperature”, “pressure”, “number of pieces”, “amount of money in cash”, “production time”, etc.
Abstract Indicator Level 1: Consisting of heterogeneous data structures result of relationship between primitive indicators. The resulting value will retain the dimensions and domains of your primitive generators. For example, the abstract indicator “number of parts produced per hour [units / hour]” can be obtained by the arithmetic relation “number of parts produced [units]” / “production time [in hours]”, it is noted that they are the primitive dimensions of quantity and time are preserved.
Abstract Indicator Level 2: Consisting of heterogeneous data structures that combine primitive and/or abstract indicators level 1, giving a conceptual value. Case Study: In the soybean business, two important indicators in the buying and selling relationship are protein and oil in grain, average values usually expressed in terms of weight percentage of the sample. Depending on the application, minimum levels must be met, for example, the receipt of soybean units at a vegetable oil refinery is subject to a minimum oil content (MIN) of 8.0 [% m / m]. An abstract level 2 indicator can be constructed as follows: “IF ‘oil content [% m / m]’ ≥ 8 [% m / m] → ‘Accept’, IF NOT → ‘Reject’”, note that the ratio between two primitive indicators in the same dimension and domain resulted in a dimensionless conceptual indicator in a binary domain.
Abstract Indicator Level 3: Combine several indicators from various topologies, besides deriving from a conceptual value a behavioral value. Case study: Assuming that in previous case industry, the receipt of the soybean units is limited to a minimum oil content (Level 2 Abstract Indicator), and in addition to this, the amount paid per bag of the sampled lot is inferred by its financial return capacity (Level 3 Abstract Indicator). This capacity relates primarily to the primitive indicator “oil content” which imposes the maximum possible extraction yield. The primitive indicator “protein content” complements the prospect of profitability of the raw material by valuing the solid residue of the extraction. Data from Embrapa, the Brazilian Agricultural Research Corporation, show that oil and protein content in soybeans vary from 8.0 to 25.4 [% m / m] and 31.7 to 57.9 [% m / m ], respectively. Abstract indicators level 2 can be constructed to define normalization factors. These factors, when applied to the soybean market price (Primitive Indicator), through a defined correlation model, arbitrate a new value that encompasses the whole perspective of financial return directly linked to the quality of the lot (behavior).
Diagram 1 shows the construction of indicators across all abstraction layers for the case study of the “Strategy for Receiving Raw Material in a Soybean Oil Refinery”. In this diagram the indicators are represented by the flow terminators (rounded rectangles) and the correlation models used in the generation and consolidation of [new] abstract indicators represented by data containers (parallelograms).
Indicator and Information Kind
Since indicators are information encapsulated in data structures, it is important to define their nature. Most of the indicators used are quantitative, because they explicitly meet the first foundation of an indicator: being measurable. Nothing is as measurable as entities represented by numeric values or countable. These entities are further classified according to the numerical representation model and may be Discrete or Continuous Quantitative Indicators..
Discrete: Possible values are contained within a finite or enumerable set or range. Example: “number of pieces produced”, “evaded students”.
Continuous: Values expressed within a set or range or union of real numbers. Examples: “temperature”, “glycemic index”.
Qualitative indicators are less common because they are not numerical representations. Qualitative entities have their values represented by categories and to be measured they must be organized in appropriate scales.
Interval scale: when it is possible to quantify differences between the values, but there is no absolute zero, and it is impossible to establish a representative and ordered interval. Example: minimum temperature and maximum temperature.
Ordinal scale: When entities can be ordered, but it is not possible to quantify the difference between their values. Example: social class (A, B, C…), school concept (A+, A-, B+, B-, C …).
Nominal scale: when variables cannot be hierarchized or ordered, being compared only by similarity or difference. Examples: ethnicity, education, gender, nationality.
The adoption of new paradigms and technologies such as Fuzzy Sets, Big Data, Machine Learning and Artificial Intelligence to process large data volumes, makes the representation of results by numerical variables inefficient. Qualitative indicators have been gaining ground in this new scenario, as the results can be simplified by categorizing them and later displaying them. An example is the TreeMap data representation technique, which uses nested rectangles of different dimensions to represent hierarchical data. The Diagram 2 shows a TreeMap representing USA exports in 2017. Note the hierarchy by economic sector (color similarity) and by the share of export balance (rectangle dimensions), either in relation to the sector or in relation to the sector to all exported volume.
Processes, Systems and Business — Concept and divagations about
In this series of articles, we use the processes as targets of the indicators, since, whatever the complexity of the demanding structures of this tool, they can be split down into processes. Moreover, processes are the functional entities of a system, being responsible for the execution of each of the activities that constitute it. Diagram 3 represents the relationships between these entities. The business contains systems that in turn contains processes. Processes can be unique to one system, or shared across multiple (system intersections), some of which have functions to connect flow between systems.
The origin of the word process goes back to the Latin procedere → processus whose meaning is ‘move forward’ which corresponds to its metaphysical definition that process is a path, a method, a sequence of efforts to achieve a goal. The physical definition of process builds limits and positions to this set of energetic and material efforts in spacetime, and a boundary is defined around this region of action. Within this boundary, actions are performed on the entities contained therein to bring about or resist change. In the physical model, the tributary flows to the process boundary are its inputs and, consequently, the effluent flows are the outputs. Processes can be sequentially chained (pipeline process), so that the output of one is the input of another, being considered [subprocesses] of a larger entity. The evolution of the process model is represented in Diagram 4.
One of the best ways to represent processes is through Block Diagrams, as they simply express input and output flows and subordination relationships between subprocesses. The mechanisms and actions implemented in each block (process or subprocess) follow the black box model and are suppressed. Only the cause (input) and effect (output) relationships are considered.
In a real application the processes will be related as entities of a system, being responsible for the execution of each one of them in an established sequence. For example, considering a vehicle maintenance system, we can define processes and relationships between them to make the system work. In this example the Customer delivers the vehicle with a receipt process involving registration, fault history, defect reports, etc. The following inspection process has the function of identifying the condition of the vehicle, belongings inside, fuel level at delivery, mileage, visible defects, among others. This process has as its first output to scientificize the customer of the state of the vehicle. The other input of this process is precisely the customer’s confirmation of the state and the authorization for repair, which if it occurs, the final output of the inspection process will be a work order specifying the defects authorized by the customer to be remedied.
The following repair process can be divided into several subprocesses, depending on the specialty involved in the defect (bodywork, mechanical, electrical, etc.). It has as inputs the vehicle itself and the work order. The outputs will be the vehicle for testing against the initially found defects and a hidden defect diagnosis sent to the previous process for consultation with the customer, feedbacking the previous processes. The repair process may still involve ordering (receiving) and receiving (receiving) parts if this need is identified in the repair.
The parts warehouse is a separate system with its internal processes. The vehicle test process returns two conditional outputs, if the repair is correct, it proceeds to the next process, which is washing and cleaning the vehicle, otherwise it returns to the previous repair process. After cleaning, the next process involves generating the payment order to the customer. After the process of delivery of the vehicle to the customer, upon payment confirmation, a relationship process is initiated in order to provide guarantees regarding the maintenance performed. This system is graphically represented in block diagram form in Diagram 5.
Performance Dimensions
The performance of a process or system is related to the role of each in the business. Processes are functional units of a system. Thus, for the system to function properly, each process must perform its function as programmed. Efficacy is therefore the key performance dimension that a process must achieve. The system in turn contains all the processes necessary for the business, being responsible for organizing and modulating the flow between them in order to obtain the best end result, ie the main dimension of performance to be achieved is Efficiency. Note that efficacy is a discrete variable because the only values that matter are whether the process has done what it should do or not. The efficiency is represented by a continuous variable, that is, there are possible nuances between the most efficient (optimal) and the least efficient (bad) state of the system. These characteristics will be very important in defining the indicators as we will see later.
A system’s efficiency directly reports how organized resource management is within it; be human, financial, energy or material resources. This translates into the ability to accomplish the ultimate goal (benefits) by doing the least work possible (lowest cost). However, even if a system is only efficient, it does not necessarily indicate that it is doing the best job possible. We will cite some very different examples to introduce a new system performance requirement.
Example 1: Diesel fuel is a crude oil derivate, usually obtained by mixing several distillation streams at petroleum refineries to make a hydrocarbon mixture with a diesel compatible distillation range. The greatest efficiency in this process is achieved by incorporating as many low-added value hydrocarbons as possible into diesel. However, this is limited because diesel follows a standardization by a U. S. competent organisms, as EPA (Enviromental Protection Agency) and EIA (Energy Information Administration). One of these limitations is the sulfur content, which makes it impossible to massively incorporate lower value-added products that do not significantly affect diesel parameters, except for its sulfur content.
Example 2: The project for the construction of a viaduct by a public entity in Brazil is much more time consuming and delicate than the same project carried out by the private sector. This is because the public bidding processes (Law nº 8.666, of June 21, 1993), public notices and hiring are structured in a standardized manner ensuring transparency (non-prevarication by those involved in the approval) and privileging fair competition and lower cost. to the public purse. The criterion time in a project is essential making it an indispensable factor in measuring the efficiency of its execution. In the example, making the viaduct available as quickly as possible can offset the additional cost of faster project execution. However, if it is not an emergency classified work, a public bid will forgo better compliance efficiency.
What do these two distinct examples have in common? The sacrifice of efficiency in favor of legal and quality standards imposed by external agencies. This new performance dimension to be measured in a system we define as Compliance. In the examples we note that these standards are applied to enable the delivery of the project result, thus being the compliance related to the outflows of a system. It is a discrete boolean value, that is, its possible values are only the status of conforming or nonconforming process.
Compliance imposes new processes on the system so that it meets a set of standards set by external bodies. In Example 1, the diesel type 2-D.S500 has its specification limit for sulfur content defined by government agencies law to a maximum of 500 ppm (parts per million) ie, with any value equal to or below this, compliance condition will be met. If by chance there were two tanks of Diesel 2-D.S500, A and B, being A with 50 ppm sulfur and B with 350 ppm, both would be compliant, however, B incorporated a greater amount of lower value-added streams (with high contents of sulfur), having a production cost of less than A. Therefore, the diesel production project in tank B was more efficient than A, since both will be sold at the same price.
The lowest extrinsic production cost of B in relation to A is defined as Economicity. This new dimension of performance to be achieved by a system relates to compliance in order to meet compliance as and only as required, reducing room for over-specification. But it also relates to other processes in sanitation of resource acquisition costs.
Based on Table 1 and Diagram 3 we build Diagram 6 which graphs the relationship between the entities of a business and their performance dimensions.
Redefining and Repositioning “Indicators”
So far we have deconstructed the predefined ideas of indicators, breaking them down into data structures, and conceptually addressing their targets: processes [and ultimately systems]. From this point, free from any preconceived concept, we can redefine indicator, in its purest form, as specific, available, measurable and accessible information; encapsulated in a data structure with a certain level of abstraction that quantifies or qualifies an entity of a process.
[Re]Defined indicator, we extend the concept to performance indicators, only delimiting the fact that will be applied to the performance dimensions of each entity. Like this:
Process Performance Indicators: Measure their Effectiveness and Compliance;
System Performance Indicators: Will Measure Your Effectiveness, Efficiency, Compliance, and Economicity, syncretically called Performance Pillars.
What about the key-performance indicators (KPI) where they fit into this story?
These are related to the performance of the business as a whole. Regressing the abstraction of entities, a business is a system composed of subsystems, and these by processes. Thus, from a practical point of view, only a semantic repositioning occurs. KPIs will measure the same dimensions of system performance (Pillars of Performance), but at a higher level (business).
Conclusion and Next Steps
Os indicadores de desempenho, tal qual conhecemos, são consumidos como direcionadores de um negócio, não havendo muitas vezes uma via de mão-dupla para um melhoramento contíguo. Parte disso se deve ao fato de estarem encapsulados em uma estrutura rígida com poucos parâmetros de customização. No primeiro dessa série de artigos, começamos por trazer os indicadores ao seu estado fundamental para em seguida reposiciona-los em relação às quatro dimensões de desempenho das entidades de um negócio, Eficácia, Eficiência, Conformidade e Economicidade. Os próximos artigos dessa série discutirão os seguintes tópicos:
· Como construir na prática Indicadores de Desempenho a partir das Dimensões de Desempenho;
· Como construir Indicadores de Desempenho utilizando dados oriundos de diversos processos e sistemas — Conectores relativos;
· Indicadores de Desempenho em Exemplos — Casos de Uso detalhados.
Performance indicators, as we know, are consumed as driver of a business, and there is often no two-way for contiguous improvement. Part of this is due to the fact that they are encapsulated in a rigid structure with few customization parameters. In the first of this series of articles, we began by bringing the indicators to their fundamental state and then repositioning them against the four performance dimensions of business entities, Efficacy, Efficiency, Compliance, and Economicity. The next articles in this series will discuss the following topics:
How to build in practice Performance Indicators from Performance Dimensions;
How to build Performance Indicators using data from various processes and systems — Relative connectors;
Performance Indicators in Examples — Detailed Use Cases.
