CASE: A Clue About How Self-Service Could Ultimately Evolve
Do the high-level lessons of the computer-aided software engineering craze of the 1980s apply to self-service? The effects of CASE tools are similar to self-service tools, and the fate of CASE suggests how self-service might ultimately evolve.
The early 1980s saw the emergence of a new kind of software that promised to empower less-skilled users and permit highly skilled users to focus on value creation. This new software promised to accelerate common, tedious tasks and to improve productivity for users of every skill level.
It might sound as if I'm talking about self-service of one kind or another. Actually, I'm talking about computer-aided software engineering (CASE).
At a nuts-and-bolts level, there seems to be almost no overlap between CASE and self-service business intelligence (BI), discovery, data preparation, advanced analytics, and statistical analysis tools. If you look more closely, however, you'll see that the high-level lessons of CASE do apply to self-service.
It isn't just that both technologies were hyped as silver-bullet fixes for particularly painful problems or that both technologies had major problems in practice and entailed significant costs as well as benefits.
It is, rather, that the effects of CASE -- i.e., the benefits of using CASE tools in practice -- correspond to those of self-service BI and data prep.
The larger, systemic effects of CASE likewise show how self-service might improve BI, discovery, data prep, etc. Finally, the fate of CASE -- i.e., the absorption of CASE-like innovations and methods into integrated development environments (IDE) and ERP applications, among others -- suggests how self-service itself might ultimately evolve.
Let's look at a few of the parallels in practice.
Self-Service Tools, Like CASE Tools, Reduce or Eliminate Repetitive and Time-Consuming Tasks
CASE tools couldn't and didn't automate application development. They weren't supposed to.
They did, however, automate many of the productivity-squandering tasks that used to bedevil programmers. These included error checking, development testing, and especially the creation of documentation.
In a 1990 article, Kathryn J. Hayley and H. Thaine Lyman (both of Deloitte & Touche) found that users of CASE tools were able to significantly reduce human involvement in repetitive or time-consuming tasks -- such as creating and maintaining documentation. (Hayley and Lyman's research was based on a survey of 2,200 private-sector IT organizations.) In a 1994 study, Charles Necco, Nancy Tsai, and Shyh-Jen Chang found a similar connection.
Granted, documentation, reverse engineering, and dev-testing aren't big issues for self-service BI or data prep practitioners. Nevertheless, there is no lack of tedious or time-consuming tasks in BI and analytics. Most have to do with data integration (e.g., identifying data sources, building data flows) or with information delivery (building business views, reports, dashboards, and other artifacts).
Just as CASE tools weren't supposed to completely automate application development, self-service tools aren't supposed to completely automate BI, analytics discovery, data prep, or other tasks.
They're designed to automate the tedious stuff. To this end, self-service data prep tools help accelerate many of the tasks (e.g., profiling data, cleansing data, joining data from different sources and automatically identifying attributes or key-value pairs) that otherwise squander productivity. Ditto for self-service BI and analytics tools, which automate the selection of visualizations, algorithms, or functions and expose what-you-see-is-what-you-get design interfaces.
Self-service is a tool as much for boosting user productivity as for user empowerment. Self-service tools centralize, rationalize, and automate many of the tasks users were otherwise undertaking on their own with a hodgepodge of different tools.
Self-service is a time saver. This tends to get lost when we focus on self-service's second and most obvious systemic effect.
Self-Service Tools, Like CASE Tools, Make It Easier for Humans to Work On Value-Creating Tasks
Back in the 1980s, IT organizations spent a disproportionate amount of their IT budgets and working time on application maintenance. Instead of building new software or adding new features or functions to existing software, coders spent most of their time fixing bugs.
In 1994, for example, organizations were able to dedicate just 20 percent of their programming resources to new application development or feature enhancement. The remainder was allocated for maintenance. CASE tools helped attack this imbalance, Necco, Tsai, and Chang found: "Organizations can effectively allocate more resources to creating new systems as the environment evolves, rather than fixing problems constantly for the existing systems."
Survey data from the CASE era puts maintenance costs at between 60 to 80 percent. Sound familiar? It's the same range we tend to assign to productivity-killing tasks such as data preparation.
To the extent that self-service tools simplify (and to a degree automate) key aspects of data prep and engineering, they likewise give skillful, creative human beings -- on both the business and the IT sides -- more time to focus on value-creating tasks.
They give business people the power and flexibility to do many of the things they want to do when they want to do them. They likewise give IT a release valve of sorts because they offload some of the BI/analytics workload to the business.
Self-Service Tools, Like CASE Tools, Really Do Make Rock-Star Users Even Better
Organizations didn't adopt CASE technologies to eliminate human coders. They did so primarily to make human coders even better. Historical survey data doesn't support the idea that CASE technologies made all coders better, but it does clearly demonstrate that coding rock stars disproportionately benefitted from using CASE tools.
Hayley and Lyman found conclusive evidence that "CASE tools do not decrease the need for high-quality resources." What's more, organizations tended to equip their best and most productive programmers with CASE tools first. "Because CASE easily automates relatively mundane activities, it allows an IS staff's best employees to do more creative work," they wrote. The duo recommended that CASE "training should be made available only to the best and the brightest employees ... who are ready to do things a different way."
We'll likely see a similar dynamic play out with self-service. In all likelihood, the empirical data will show that although self-service tools make it easier for less-skilled users to do their jobs, the highest performance gains are for skilled users.
Business analysts, statisticians, data scientists, and other highly skilled consumers already require self-service-like tools to do their jobs. More than this, these users report feeling empowered and emboldened by these tools. Self-service tools will help make these bright, creative, problem-solving people even more productive.
Self-Service Tools Will Eventually Be Incorporated Into Larger Systems
We don't hear much about CASE these days. This doesn't mean it was a failure, however. No technology can be a silver bullet, and in the case of CASE, the hype far outstripped the substance. CASE as a discrete category more or less went away, but CASE innovations and concepts were ultimately incorporated into integrated development environments (IDEs), ERP applications, and other kinds of tools.
We'll likely see the same thing happen with self-service tools. Self-service as a category will probably go away, but self-service innovations and concepts will be incorporated into larger BI/analytics, data integration, or advanced analytics tools. We're already seeing this happen.