What is a system? 🤔
At the outset let us define what we mean by a system. A system is any IT tool that has an input, business logic, a process, a data structure and an expected output. There is also one other condition - the inputs & outputs are dynamic in the sense that new inputs come in periodically generating fresh set of outputs. In this sense anyone can design a system and it need not necessarily be created by people in IT. For managers today we have several tools for creating an MIS system such as spreadsheets and analytics platforms.
Why do we need systems? 🤔
Why do we need systems for analytical work? Today, much of the work happens in a virtual office with spreadsheets and electronic presentations. It is all knowledge work. When the work becomes repetitive it leads to boredom. If it is intensive it leads to manual errors. For professional growth, one needs improve skills so that the speed of work and productivity increases, and mistakes come down. This happens through a learning curve. However, by learning to create systems, speed and productivity increase dramatically at an exponential rate.
Systems can reduce work time and improve productivity 😎
A system enables one to drastically reduce the work time of a repetitive job. From our experience, we know that work time can be reduced by as much as 80% by implementing a system. This blog is all about how to create an analytics systems which is one of the key ways of cutting down work time and improving productivity.
Design the Input Structure
First choose your inputs carefully. Avoid manual inputs to the extent possible. Instead use standard reports or queries as your inputs. These are available in all organizations and are usually extracted from transactional systems such as ERPs. Separate and keep a separate area for the raw data. Keep a visible separator if you are using a spreadsheet. Use naming conventions in file names for analytical tools.
Clean the Raw Data
If you are using spreadsheets such as Excel or Google Sheets, there are a host of functions available for cleaning the data. Analytical tools such as Power BI are equipped with query functionalities to clean the raw data. Cleaning could involve harmonizing product codes or location codes, identifying headers, removing blank or irrelevant records, harmonizing date formats, segregating the data into manageable buckets or filtering & extracting relevant data. Never apply cleaning in the middle of raw data. Just as raw data is in a separate area, this cleaning also needs to be done in a separate area but needs to be closely linked to the raw data.
Process the Data
Apply business logic to the cleaned data to get the information you want in a separate area. If you are working on a spreadsheet, you may have to do this using lookups in a separate sheet. If the logics are complicated this may take several sheets. Whatever the tool used, keep the flow of logic in a unilateral direction and always keep the areas neat and segregated. Today, both spreadsheets and analytical tools are equipped with functions, queries, pivots, etc to handle any kind of logic.
Design the Output
To the extent possible, make the output visual. In some cases, a tabular output is preferable. Spreadsheets as well as analytical platforms have a number of visual graphics and charts to choose from. Two cautions to take care here. Usually the processed data needs to be filtered to give an actionable output. Make the filtering automatic to reduce the number of clicks. Secondly, do not ass manual comments to qualify the output in the output area. This messes the system. Remember that manual entries are an input. In keeping with the principles of building the system, always use a separate input area for this purpose.
Build a Story with the Output
It is not enough to just design a visual output and leave it to consumers to make sense of what is given. Visuals have to be simple and intuitively very simple to understand. They need to tell a story of their own. If the viewer wants more data, he or she should be able to drill down and get the required information. There needs to be a cue such as a “click” to go to the next visual which should continue the story to the next logical step. If a decision needs to be made, a good practice is to get all the relevant data on one page.
Have a process for updating the system
Updating the system involves all the above steps from putting in fresh inputs to getting the output as a storyboard. For a system to work well, it has to be updated in exactly one way every time because that is the best way. If it is updated in an ad hoc manner, chances are high that it will mess up the system. The best thing to do is visually build in cues so that one knows how to update the system with fresh data. Another good practice is to document the update process and keep it in a readily accessible place for reference.
Building a system is hence all about following a set of techniques and practices while working with everyday data. It also takes time and effort to build a good system. But the efforts are well worth the benefits.
Comentarios