Over the years I have seen a number of clients try to buy something to solve their analytics challenges. When I worked for Cognos, people bought Cognos thinking it would fix things, sometimes they threw Cognos out thinking it was the problem. Companies have purchased and thrown out various ETL tools, databases and reporting tools or all of the above, hoping to find the magic bullet to make reporting and analytics easy.
It has never worked.
But the most ridiculous strategy I have seen is buying code based solutions to deliver analytics. Usually packaged as a combination of ETL tool and ETL code, these “Analytic Applications” claim to deliver an out of the box data warehouse that will solve all your reporting and analytics needs. Examples: Noetix, PeopleSoft EPM, and Cognos had one of these a long time ago. Every single one started as a project for a client, and then someone got the bright idea to save the code and try to package it into a product.
Why don’t they work?
Imagine trying to write ETL or a metadata layer for Oracle EBS or SAP that works for every ERP implementation? Impossible. Even if it did work, the ETL or views are so complicated that they are impossible to use. Every time we have come across these applications, performance, and support are big problems, and users are not happy. The ETL is complex because it captures every possible data element. As a result, the ETL barely finishes in reasonable times, and delivers monolithic models that contain 75% noise – data that nobody wants. Using views is even worse, because in addition to forcing users to weed through hundreds of fields they don’t want, they also have to pay the price of terrible performance at run time. Oh, and when the underlying ERP changes? Haha! Good luck.
Why then, do people try to implement these things? For the same reason the supplement industry is $160 billion and the weight loss industry is $100 billion. Like software, many practitioners in these markets offer magic beans (pills) to desperate buyers who are unable or unwilling to do the work.
In the end these options cost way more than expected, and deliver much less value than promised.
The real answer is to do the work. In the words of David Goggins – “There is no hack Bro.” In the long run, your people are much better off learning how their businesses are run and managed, and better off learning how the systems that support operations work. They are better off learning the craft of data engineering, data architecture, managing IT processes and applications. Administering some black box and opening a support ticket every time an alert goes off or the business has a problem is not something talented people sign up for.
Who is running this place? Nobody talks about ROI anymore. The budget is set before the strategy is defined or the plan is made.
The people who manage analytics teams are better off learning how to develop strategies to improve the business, finding partners to collaborate with and making sensible investments to execute meaningful plans. All too often I see IT aiming to placate the business with the least amount of effort, so they buy a solution off the shelf, and they get the business to pay for it. And then managing budgets becomes their primary activity and they essentially behave like a procurement department dedicated to IT. Nothing disappoints me more than hearing an IT leader make a cost-justified choice between one option that promises to cost less and delivers little value, and a strategic investment that provides capabilities that are important but requires investment of time, effort and money.
If you can buy a product and install it, anybody can. How can this be a meaningful differentiator?
Well, I guess I am in the mood to pick fights today, sorry for that. It does seem like this is the best way to get engagement these days and I like to express my opinion anyway. If you are one of those IT leaders that wants to make a difference at your company, drop me an email or connect on LinkedIn. Whether we work together or not, I want to know you and you’ll probably like what I have to say from time to time.
elealos@quantifiedmechanix.com
https://www.linkedin.com/in/ericlealos/