Try a round

According to the Finance Sector Union, CBA reversed its plan to remove 45 customer-service roles after calls and overtime rose. A likely reason is that the voice bot took the easy calls and left the hard ones. Routine requests are what bots handle best, so the bot could look busy and successful while the calls still reaching staff became longer and more complex. Fewer people would then face a heavier mix of difficult work, which would fit the rising overtime the union reported. The union itself described the problem as demand forecasting, not the bot’s answers, and that is a serious alternative. To settle it, we would need the mix of work and the repeat contacts before and after the bot, which this packet does not include.

The Finance Sector Union disputed the savings claimed for CBA’s voice bot, reporting that calls and overtime were rising. That gap suggests the bank was counting the wrong thing. If success meant how many calls the bot handled, the bot could meet its targets while customers still needed a person to finish the job. Each unresolved call passed back to staff would look like bot activity on one report and extra work on another. That would explain how savings could be claimed while overtime grew. This is an inference: the packet does not say which measures CBA used, and we lack the bot’s resolution rate and repeat contacts. The union’s own account, that management underestimated how long higher demand would last, remains a strong alternative.

CBA planned to remove 45 customer-service roles, and the Finance Sector Union disputed the savings claimed for its voice bot. The union says calls and overtime then rose, and the bank reversed the plan. The order of events tells the story. Executives wanted the headcount reduction locked in before the end of the financial year, so the cuts were planned on a deadline rather than on evidence that the work had gone. Had the bank run the bot alongside full staffing first, the rising demand would have shown up before any roles were put at risk. The union says management underestimated how long higher demand would continue, which fits acting first and checking later. A competing view is that demand rose for reasons no trial could have predicted.

The voice bot may have done exactly what it was built to do, and that may have been the problem. A bot tuned to finish calls quickly, or to keep customers away from staff, could succeed on its own terms while the service as a whole got worse. If customers who could not get an answer simply rang again, calls would rise, which matches what the Finance Sector Union reported about rising calls and overtime at CBA. But this packet does not say what the bot was designed to achieve, and the union’s explanation is about demand forecasting and staffing, not the bot’s design. So this cause rests on an assumption. The facts needed are the bot’s objective, its resolution rate and how many callers came back.

Rising calls and overtime are warning signs, and the Finance Sector Union says call volumes contradicted the case for removing 45 customer-service roles at CBA. The question is why the plan went ahead until it had to be reversed. One answer is that nobody had clear authority to halt it: savings could be reported upwards while the people handling the extra calls had no route to pause the cuts. The reversal shows that someone could eventually say stop, but late. This argument goes beyond the packet. We do not know CBA’s approval process or who reviewed the plan. The union’s simpler account, that management underestimated how long higher demand would continue, explains the same facts without any failure of authority.