Making self-service BI a success
Organisations increasingly rely on data analysis for taking strategic and tactical decisions. Crunching data has never been easier. Data is available from many sources. It can easily (and cheaply) be stored. And many tools exist for analyzing data sets. While for long the exclusive realm of data analysts and data scientists, a huge part of data analysis can now be performed by the people who actually understand and run the business. Self-service analytics tools such as PowerBI, Tableau, Qlik, etc empower just about anyone to crunch numbers to their needs. But do they?
This takes me back to the days when Powerpoint was let loose on hurds of business people who overnight got bombarded to be presentation designers. You’ll find some examples of what that lead to here by simply browsing the web for “worst powerpoint presentations”. Great as it may be, Powerpoint didn’t turn us all into great presentation designers!
The same is happening to self-service BI… Many companies have invested in the tools, some (!) have trained the users but many companies we speak with admit they're disappointed by the results. The problem lies not with the tool selection. Whichever they picked is more than likely overly capable of satisfying their needs. Compare this to a carpenter who just bought a new lathe; besides knowing how to use it, he’ll still need raw materials and understanding of wood before being able to churn out beautiful furniture.
For analytics, two capabilities are key before embarking on an effective analytics-based journey:
领英推荐
To continue our woodworker parallel; would a good artisan start building a chest of drawers without knowing he had the right material in stock? And, assuming the inventory is quite big, wouldn’t that stock either be well-organized or would there not be a reference system to easily locate materials? Well, in data inventory invariably is vast and almost always is it poorly organized (data is stored across different datasets, etc.). Therefore, some sort of “inventory management system” is required to allow users - in our case business users, not just IT people - to easily locate the raw materials for their analytics reports.
Once materials have been located, we need to make sure people have a correct understanding of what exactly they’re looking at. Very often do people compare reports and don’t understand why they end up with different results. Our woodworker could be looking for walnut but there’s sanded, burl, crotch, etc. And then there’s Black walnut, English Walnut, etc. Not knowing exactly which one is needed may lead to a hodge podge chest of drawers. It’s exactly the same in business: if one finds sales data in a dataset, should one assume all sales-labeled data means the same or could it in some dataset refer to local data, in another to aggregated sales data, in another one to sales forecast data, etc. You get the gist.
Know AND understand what you have. It’ll avoid a lot of frustration (for those building the reports) and avoid a lot of discussion (for those reviewing the reports). In the end, not unlike physical inventory management, it’s just about good business hygiene. After all, investment in time, tools and people should be efficient and effective.