Data-entry quality failure at a small ITeS vendor
A small ITeS vendor in Bathinda has a contract to digitise 1,200 applications per day for a state scholarship scheme. Five data-entry operators work in two shifts; each application has name, Aadhaar number, bank account, income, and category fields. The client's SLA requires 99.5% field accuracy and delivery by 6 p.m. daily. On Monday the client reports a 6% error rate and 43 duplicate applications. The supervisor finds that one operator uses a browser auto-fill tool to speed up entry, another copies previous rows in a spreadsheet, and the team has no second-level check because the verifier resigned last month. Aadhaar numbers are stored in a shared folder accessible to all five operators. The client threatens to invoke a penalty clause. The vendor must respond within 24 hours without stopping the daily delivery.
1. Identify the two most likely root causes of the 6% error rate and 43 duplicates, and explain how each cause produces the observed symptoms.
Read the worked answer
The pattern points to two separate causes. First, browser auto-fill carries forward previously entered values. When the operator moves to a new application, fields such as name, Aadhaar, bank account and income may be pre-filled from the previous record; if the operator overwrites only the visible name and not every field, the new application inherits the old Aadhaar or bank account. That produces mismatched fields and can create duplicate applications when the same Aadhaar is saved repeatedly. Second, copying previous spreadsheet rows reproduces the entire row, including application ID and Aadhaar. If the operator edits only name and income, the ID remains the same, which explains 43 duplicates. The missing verifier means neither error is caught before delivery. The 6% error rate is consistent with a small number of operators using shortcuts across a high daily volume: at 1,200 applications and five fields, even a 1% field error rate is 60 wrong fields. The shared Aadhaar folder is a separate privacy risk, not the cause of the duplicates.
2. Propose a containment and correction plan for the next 24 hours that meets the 6 p.m. delivery without compromising accuracy. Specify sampling, validation rules, and role changes.
Read the worked answer
Containment for the next 24 hours: freeze the 43 duplicate IDs and any rows entered by the two operators using shortcuts; do not delete them until they are compared with the source forms. Disable browser auto-fill on all five machines and block paste into the application ID, Aadhaar and bank columns. Assign a rotating verifier from the team, with the supervisor checking high-risk fields. For today's 1,200 applications, require 100% verification of Aadhaar and bank account and a 10% random sample of name, income and category; if the sample error rate exceeds 1%, expand to 100%. Use validation rules: Aadhaar must be 12 digits and pass the Verhoeff checksum; bank account must be 9–18 digits and match the passbook; application ID must be unique via a COUNTIF check. Reconcile the 43 duplicates against the source register and the client's list, then submit a corrected batch with an error log. Deliver by 6 p.m. in two waves: first the verified current batch, then the corrected duplicates. Record who verified each row so accountability is traceable.
3. The shared folder exposes Aadhaar numbers to all operators. Recommend immediate access controls and a retention rule, considering the client contract and Indian data protection expectations.
Read the worked answer
Immediate access controls: remove the shared folder's 'everyone' permission; create per-operator folders with access only to assigned batches; give the supervisor read-only access to all; enable logging and require individual logins, not a shared Windows account. Encrypt the folder and prohibit personal pen drives or personal email for Aadhaar data. Retention: keep source scans and Aadhaar only as long as the contract and the scheme require, then delete or return them; do not keep them in a general shared drive. Mask Aadhaar in displays and reports (last four digits only). If any unauthorised person accessed the folder, treat it as a suspected personal-data breach: preserve logs, inform the client and the company's grievance officer, and do not notify individuals until facts are confirmed. The contract may require a specific breach notification timeline; check it. These steps cost little and reduce the penalty risk more than simply warning operators.