Earlier, we covered two different approaches to defining KPI requirements: the strategy-driven approach and the process-driven approach. Today, we’re covering two other methods: the data-driven and the market-driven approach.
The data-driven approach
The data-driven approach defines KPI requirements using the registered data in the information systems supporting the business operations. External data sources could also fulfill the information requirements. An intelligent organization will also consider any other potentially interesting sources. Possibly even sources that don’t exist yet, but that have the potential to generate a lot of interesting data. For example, (IoT) sensors built into pills, trucks, or plane engines. This approach works as follows:
Look for the applications / information systems which the organization uses to manage its data. This could be the ERP system, the financial administration app, the CRM system, or a complaint system. If it’s relevant and affordable, you could look outside of the organization for data. For example, market research, demographic information, Twitter posts, the cadaster, the patent office, trademark register, a press release site, weather information, stock markets, the chamber of commerce, or a competitor’s website. All kinds of sensors can also function as a source. That quickly gets us into big data territory.
The data sources have a specific storage structure. Determine to what extent they’re suitable for distilling the available data into reliable information. Some sources can’t be used to take measurements. They may not be structured enough, or available in a format that doesn’t match the existing IT infrastructure. Some organizations are still working with legacy systems, which are often difficult to use.
Determining counters and indicators
When the structure of a source is clear, you can define counters and (performance) indicators of various shapes and sizes. This choice often partly depends on the degree of complexity. This creates a certain amount of variation. You could place a counter on the order table, for example, allowing you to monitor the amount of incoming and resolved orders. Or you can multiply the number of items in stock with the price from the order table and then add them up. That way you can find out what the revenue per item was on a given day. Counters and indicators on other tables work the same way. This approach is quite simple can be executed without too many obstacles, given that the sources are accessible and well-documented.
Example of an interesting big data source
A well-known oil company has thousands of oil pumps on the bottom of the ocean floor. When a pump breaks down unexpectedly, it costs the oil company nearly five days to repair it. The oil company decides to outfit pumps with various kinds of sensors, which generate a constant stream of big data. Irregularities in that data stream indicate that a certain crucial component is at risk of breaking down, which would cause a pump failure.
The data is monitored and analyzed in real-time, and any irregularities are immediately addressed. The company performs preventative maintenance on the pump, limiting the pump outage (a KPI!) to a maximum of two days. The result: huge savings for the oil company.
A word of caution
The big danger of this approach is that you can’t be sure if the indicators are genuine KPIs. Or that the produced information is actually useful. The data-driven approach feeds the system with a lot of information, and you probably won’t have enough real management information. As a result, the system isn’t usable, or only operationally usable.
The market-driven approach
There are also industry-specific systems, so-called analytical applications, on the market. You can use these to meet the first version of your information requirements in relatively short order. They can possibly even result in a first version of a management information system. These applications often only cover a part of the business, such as HR or marketing campaigns. This approach consists of two steps:
- Market research and (BI) tool selection: the available analytical applications are judged on several selection criteria. The most important criterion is of course whether the application and the indicators match your situation. See also, for example, Passionned Group’s Business Intelligence Tools Survey.
- Choice and adjustment: when you choose an industry-specific application, the system often has to be adjusted. The insights, indicators, angles, reports, and KPI dashboards have to be matched to the situation in your organization.
This approach could prove costly if it turns out that the application doesn’t match your own organization. Or if it proves very difficult to adjust the indicators and the required data links to the company’s systems. You run the risk of still having to choose another approach.
You also might have to build the management information system yourself after all. There’s also the risk of the organization focusing too much on the application’s functionality and thoughtlessly copying indicators, without thinking about how to use these for the continuous improvement of processes and strategy refinement. We don’t recommend using these kinds of analytical applications without adjustments. The application has to closely match the strategy and the key processes. And those should always be relatively unique to your organization.