Dozens of programs each designed to produce a package of reports. Multiple copies of each report were produced, requiring someone to separate the copies of each report and collate everything into multiple copies of the entire suite of reports. By spooling the output of each program and adding a final output step which produced multiple copies of the suite of reports, what used to take hours each day was now performed in minutes.
The morale improvement was noticeable and it freed people up to take on other tasks.
At the time, MCI received its sales data from many sources, from Bell companies down to independent cold-calling outfits. All input programs reformatted the data and output it to tape (remember, the 80s), but many of the input files were very small and could be kept on disk. I changed most of the programs to use disk and added a program to bring those files into the main processing seamlessly.
Characteristic of such a hodgepodge set of data sources, input regularly was processed out-of-order, resulting in rejection. This rejected data was then queued for re-processing.
Cutting the run time to less than 24 hours meant that reprocessing could happen the very next cycle instead of waiting an extra day to be pulled in -- improved the cash flow for an industry disrupter.