Are you a professional Data Analyst? Do you casually browse through data? Neither? There is a common ask of analysis tools and solutions that they be powerful enough to be useful, yet simple enough that novices can use. What does that really mean? How powerful do they need to be? How simple do they need to be? There is a dichotomy in those requirements that we need to really delve into and examine.
Powerful, Useful Analysis
We spend significant resources on data collection, aggregation, and analysis because it can help us gain useful insights for our business. We can track metrics to find out if we are being overcharged for goods and services. Using that analysis, we can also determine which suppliers are the best to choose. Other metrics may help us determine which types of customers to focus on, what products to build, etc.
Quite often, this analysis just isn’t wrapped up in a nice package and handed to us. We have to do some digging. We have to go through current records, past records, perhaps even records that we don’t own ourselves. This may be such a daunting task that we get a dedicated expert to work on it. Perhaps use dedicated tools for the analysis. The data is never in a nice and easy to use format. These powerful tools are powerful because they can do the data manipulation and aggregation for us, and show us visualizations that we want to see.
This powerful, useful analysis comes at a cost though. You must be an expert in the data, and at least proficient with the tools. That is why people dedicate their professional lives to this; because it is that big of a problem and requires that much expertise. Many times, the useful insights come from years of trends, and perhaps months of investigations. Those who are not in the data and in this field constantly just do not have the time to devote to this. However, these experts may be able to find the useful pieces of the information, but they may not know a) that this is useful information and b) what to do with it.
Analysis When You Need It
That brings us to put ourselves into the shoes of the people that can make use of that analysis that is handled by experts. Even though the analytical experts may not know exactly what they are looking at in this trend, we do. We can take actions off that newly found information.
Seeing that our headcount numbers keep increasing at a constant pace, we can see that we will need to allocate more space for our workers next year. That may be a large project that we need to start to get rolling now. Seeing that even though supplier A has better prices than B, they also have a higher defect rate, causing us to constantly loose time over the defective products. So in the long run actually choosing supplier B is the better choice for this widget.
However, for us to be effective with this data that the analysts found, we have to actually see it. Not only that, but we have to be able to see it in a context appropriate for making a decision or taking an action on that information. Seeing price trends on a product that I’m not looking at buying is not useful. Seeing satisfaction rates for a supplier that is no longer in business is not useful. The data must be brought in at a time when it can actually be effective.
We also do not have the time generally to go out and do the job of the analysts. There are times when we might want to, out of curiosity, or perhaps because we have a very targeted question to answer. But these rare cases are probably not something to build a system around. They would probably be better served by an email exchange, or a meeting, or a feature request to the “analyst team” to see if they can present the information we might want in a new way.
Keep The Uses Separate
These uses of an analytical “engine” are very distinct. Both are very targeted to different sets of users. One is exploratory, and the other is answering very targeted questions, at specific times. Trying to build a single entry point, if you will, for both cases is absurd. Keep them separate, and serve them with different tools. While both may actually make use of the same underlying platform or technology, they most definitely are going to be presented in very different ways.
The data analyst will need to see the raw data, to explore and play around with it. The analysis consumer doesn’t care about the raw data, just the summary. Combining these uses will give you an experience that is not satisfactory to either user. The data analyst will not be happy because the power of the system has been taken away. The consumer will not be happy because it is too confusing; he cannot get the data he needs.