He disagreed at first. "Nobody gets applauded for raising expenses," he told her. "Headquarters wants numbers that show improvement."
"Then give them honest numbers," Jessica replied. "Otherwise we'll be right back where we started, only with more impressive presentations."
The conversation unsettled him because he recognized his own old reasoning in it. For years he had treated employee complaints as information that could be managed until a better quarter made them irrelevant. He was tempted now to treat the costs of reform in the same way, and Jessica had caught him doing it before the habit became a policy. He apologized without attempting to explain how hard his job was.
During the sixth week, a regional analyst noticed that customer satisfaction surveys were being completed at a much higher rate than before. At first the increase looked encouraging. Then a technician discovered that sales associates had started giving the survey only to customers they thought were pleased, leaving dissatisfied visitors with no clear way to comment. The behavior had not been ordered by management, but it fit perfectly with the company's old preference for attractive figures.
Jessica called a team meeting and explained the problem without naming individual employees who had followed an ambiguous process. "A higher score means nothing if we only ask people who are going to praise us," she said. "We need feedback from the people who leave disappointed, especially when we don't know why."
An associate asked whether that would lower their bonuses. She answered honestly that it might change the apparent scores in the short term, and that HR was reviewing bonus measures so employees would not be punished simply for making the reporting system more complete. The issue mattered to Kesha when she reviewed the report. A company could teach employees to welcome people fairly and still reward them for manipulating the evidence of whether it happened.
Kesha joined a follow-up call from her office in Atlanta, having traveled east for another set of meetings. She listened as Jessica described the revised survey distribution and the confusion it had created. Rather than reprimand the entire team for failing to predict every consequence, she asked which incentives had made the incomplete surveys attractive. Then she authorized an independent baseline review so the next improvement figures would be compared against something real.
"Don't make me a beautiful graph that hides the people we still disappoint," she said. "If a measure gets worse because we're finally asking the right questions, tell me that."
The words spread through the regional operations office in a different way from the viral video. They were not exciting enough for millions of strangers to share, but the analysts remembered them when they prepared the next month of reports. Several uncomfortable numbers appeared in the dashboard for the first time. Complaints rose temporarily because customers were finally being offered a reliable place to submit them.