It is 4:40 on a Thursday afternoon at an imaginary BPO in Kochi that handles insurance renewals. A team leader, Joseph, notices an agent’s spreadsheet open on a shared screen: policy numbers, names and mobile numbers of about 1,200 customers, with a filename that suggests it was meant to be emailed. When asked, the agent says she was sending it to her personal email to “finish the work from home”. The email has already gone.
Is this a data breach? Let us use it to see what a breach response involves, hour by hour.
What counts
A personal data breach, loosely, is any event where personal data is lost, accessed, disclosed or altered without authority. It does not require hackers. A misdirected email, a lost laptop, a screenshot in a WhatsApp group or an agent sending files to her own account all qualify in most definitions. Whether a particular event legally must be reported depends on the law and the contract. In India that includes the Digital Personal Data Protection Act, 2023 and its rules, plus sector regulators and, for cyber incidents, the CERT-In directions that ask for certain incidents to be reported within six hours of noticing. Legal specifics change, so the compliance team owns that call, not the floor.
Hour zero to one: report and stop the bleeding
Joseph’s first duty is to escalate immediately through the channel defined in the incident procedure: the information security team and his operations manager. He does not wait to investigate fully, and he does not tidy up. Speed matters because every later step depends on someone who knows the procedure being in charge.
Containment follows. The security team can disable the agent’s access, ask the email provider to recall the message where that is possible, and check whether the file went anywhere else: other external addresses, printers, cloud drives, pen drives. They also secure the agent’s workstation without wiping it.
Hours one to six: preserve and assess
Evidence is easy to destroy by accident. Email logs, access logs, the file itself and the workstation’s state are kept as they are. Joseph writes down what he saw, with times, while memory is fresh. The temptation to confront the agent at length, or to speak to the team about it, should be resisted; HR and security handle the interview.
Then the assessment questions. What data was in the file, exactly? Names and mobile numbers are one thing; if it had Aadhaar numbers or health details, the seriousness rises. How many people are affected? Who received it, a trusted personal inbox or an unknown party? Is there any sign it was used? A quick, honest first estimate is better than a perfect one next week.
Hours six to twenty-four: tell the client
The data belongs to the insurer, and the BPO is processing it under contract. Most contracts and security schedules require prompt notification of incidents, often within a fixed window of hours. The notification goes through the designated contacts, in writing, stating facts: what happened, when it was found, what data, how many people, what has been done, what is not yet known. It does not speculate and does not minimise. A client who hears about a breach from a third party, or two weeks later, loses trust faster than one who hears a bad but accurate account on day one.
The client, as the party responsible for the data, then decides on notification to regulators and to affected customers. A BPO should not contact those customers itself unless instructed to.
After the day: find out why
The agent’s act was the trigger, but the interesting question is how it was possible. Could she export a spreadsheet of 1,200 customers? Was the email gateway able to send it outside without a block or alert? Had she been trained, and was working from home for unfinished work a symptom of unrealistic targets? A root-cause review with corrective actions, owners and dates turns the incident into a fix. Where the cause is a person’s deliberate misconduct, disciplinary action follows the company’s process and any legal advice.
What goes wrong in practice
- Nobody is sure who to call, because the incident procedure is a document on an intranet nobody has opened.
- People fear blame, so incidents are reported late or not at all. A no-blame reporting culture, with consequences reserved for deliberate or repeated violations, produces faster reports.
- Teams delete things “to clean up”, destroying logs.
- The notification to the client is vague, then corrected three times.
Rehearsal helps. A short tabletop exercise once or twice a year, where managers talk through a scenario like the one above, exposes the gaps in a way no policy review does.
For Joseph’s team, the outcome was undramatic: the recall worked on the personal mailbox, the agent confirmed no onward sharing, the client was informed that evening, and export permissions were tightened the following week. Many real incidents end this way, which is exactly why prompt reporting is worth the awkwardness.
Figures and names in the examples are illustrative.
Leave a Reply