High quality data can play a huge role in increasing efficiency and improving performance and can help managers in the decision-making process. Sometimes, it is acceptable to make decisions based on instincts and gut-feelings, but the majority of them should be backed up by numbers and facts.
Data-driven decision-making is a process of collecting measurable data, based on organizational goals, extracting, and formatting data, analyzing the insights extracted from it, and using them to develop new initiatives. Nowadays, advanced software is available to help with data gathering, processing, reporting, and visualizing, to support managers.
The main steps of the decision-making process
The first step to build a well-functioning, data-driven decision-making process is to clearly define organizational goals, and to identify the questions to which the answers we find can help reach these goals. For example, if our company’s revenue goal is to increase its portion of the market share by 20% until the end of the year, a good question would be: what are the most important factors which have influence on market share?
The next step is to identify data sources and to find custodians. The source of the data highly depends on its type. There are qualitative data, which cannot be expressed by numbers, and quantitative data, which can be measured by numbers. We can collect data from primary and secondary sources. Primary sources can be observations, interviews and surveys, whilst secondary data can be collected from external documents, third-party surveys and reports.
The third main step is to clean the gathered data. During the data cleaning process, raw data is prepared for analysis by correcting incorrect, irrelevant or incomplete data. There are six data quality dimensions which should be kept in mind, during this process: Accuracy (indicates the extent to which data reflects the real world object), Completeness (refers to whether all available data is present), Consistency (refers to providing the same data, for the same object, even if this data appears in different reports), Conformity (consists in ensuring that data follows a standard format, such as YYYY/MM/DD), Timeliness (indicates whether the data was submitted in due time, respecting the data gathering deadline) and Uniqueness (points out that there should be no data duplicates reported).
Only now, the data analysis process can start. Statistical models should be used to test data and find answers to the business questions identified beforehand. Descriptive statistics can help to quantitatively describe and summarize features of data and to describe, show or summarize data in a meaningful way. For example, monthly sales or changes in employee competency levels can easily be presented in a visual manner.
Interferential statistics can help find correlations between different variables and predict future outcomes. For example, by using regression analysis, we can make a prediction on how growth, experienced in the employee competency level, can positively affect the sales volume.
Even if the data gathered is cleaned and correct, and the data analysis process has respected all the recommendations above, if the data is not presented in a meaningful way, it will not be of much use. Well-presented information and the outcomes of the analysis can help in interpreting data, thus supporting the decision-making process.
From time to time, data should be updated and re-evaluated, to make the best decisions in today’s continuously changing business environment.
The advanced analysis techniques and software, which are available nowadays for the majority of organizations, make it possible to build up a data-driven decision-making culture, which leads to more prudent business decisions. These tools generate more thoughtful decisions that help performance improvement, which ultimately lead to organizational growth.
We already know that good quality data can help in the decision-making process. The first important step is to collect data from reliable sources. There are two types of data sources to consider: primary and secondary.
Data from primary sources are first-hand data, tailored to provide information on the firm’s own products, customers, and markets. It can be collected from both the internal (employees, board of directors, investors etc.) and external stakeholders (customers, suppliers etc.) of our organization.
Data from secondary sources are facts & figures already collected and recorded prior to the analysis done by others, and can be collected from internal sources, i.e., our annual report, sales data etc., or external secondary data, from government database and reports, national reports etc. This type of data includes both raw data and published summaries.
Primary and secondary data can be either quantitative or qualitative. Quantitative data refers to numbers and quantities like age, competency level, etc. Qualitative data is descriptive, observable and cannot be measured, i.e., clothing style.
Sources of primary data
The most widely used methods of primary data collection include the observation, interview, and survey. While these are not the only ones, most others are less popular than the former three.
The observation is the most used method of data collection in social and natural sciences. This method consists of gathering knowledge by observing certain phenomena when it occurs.
There are two types of observations: participant and non-participant. In case of the participant observation, the researcher watches the events and activities from inside, by taking part in the group he is observing. The researcher can freely interact with the participants. In the case of non- participant observation, this occurs when the researcher observes the events passively, from a distance, without direct involvement.
During this specific data collection process, chances of personal biases are high, as the observer interprets the situation in his/her own way.
When it comes to all fields of science, the survey is one of the most used methods of data collection in research. Questionnaires are formulated to acquire specific point information on any subject area. The questionnaire is an inexpensive method of data collection, when compared to other methods of primary research. Questionnaires can be submitted by vast audiences, at a time, and responses can be registered quite easily.
Lastly, the interview is another important method of primary data collection in all fields of science. During the exchange, the interviewer collects information from each respondent independently, making this process much more expensive and time-consuming when compared to other methods of data collection.
Sources of secondary data
We can collect secondary data from many sources, such as:
Text-based data sources, i.e., magazines, newspapers etc.
Non-text-based documents, i.e., TV, radio etc.
Survey research conducted by other entities, i.e., the government, NGOs etc.
What data sources should we focus on?
There are advantages and disadvantages to each of the sources, which is why choosing the appropriate data source is highly dependable on the research’s objective. Here are some considerations that might help deciding:
Advantages of data from primary sources
It is more reliable, because the source of the information and the data collection method are known
The collected information is up to date
The collected information is owned by the organization conducting the research
The organization conducting the research can ensure that it is addressing a specific issue, rather than investing in extracting relevant information from other sources
Disadvantages of data from primary sources
More expensive than secondary data
Advantages of data from secondary sources
It is easy to access, so it is less time-consuming, and the data collection-related costs are lower
A large amount of data can be collected easily
Disadvantages of data from secondary sources
It may be possible that it is not tied to the organizational needs
It is not as accurate as primary data, and it might be outdated
As we can see, data collection can come in many forms, types, and methods, almost as varied as the very object it desires to aggregate. Which method suits your needs is conditional on your research objective. Whilst some may require an in-depth, live approach via the interview or even observation, others could do with just a quick & easy fix, via surveys.
Moreover, carefully consider which sources will yield the most accurate and trustworthy data. Some research benefit greatly when you incur information from primary sources, whilst others yield surprisingly pinpoint results with just secondary references.
The PERFORMANCE Management team was present at the HR Directors Business Summit in Birmingham, United Kingdom, between 2 – 4 February, 2016. Consequently, the insights gathered from experts present at the event had been gathered and exposed here, in a series of interviews and articles.
In Part I and Part II of this article, we emphasized the need to base decisions on data, as this leads to improved results, and we discussed about the decision-making process and the techniques we can use to ensure relevant decisions are made.