What the sheet actually looks like
The DPDP-safe BPO scorecard is a single sheet. Fourteen rows. Five columns. Reviewed weekly per agent, calibrated monthly per queue. The discipline is in the row structure — agent signals, team rollup, coaching loop, and audit trail share the same page so the team leader and the DPO read the same evidence on the same day. The pattern below is what an 850-seat Pune BPO ran during a recent procurement diligence we worked through, with the consent and retention lines reviewed against DPDP Section 8.
The five columns are the same across every row. Column one is the signal name. Column two is the source system the signal pulls from — no new capture, no agent endpoint hook. Column three is the baseline window the signal is measured against. Column four is the amber threshold. Column five is the red threshold. Amber triggers a team-leader review. Red triggers a coaching note the agent can see and contest.
| Row | Signal name | Source system | Baseline window | Amber / Red threshold |
|---|---|---|---|---|
| 1 | Shift adherence | WFM roster + dialler queue-ready log | Rolling 30 days, per agent | Amber: -5 to -10 min/day; Red: > -10 min/day |
| 2 | Focus density | App-context signal (work tool in focus) | Rolling 30 days, per agent + queue | Amber: 8-12 pts below team median; Red: > 12 pts below |
| 3 | Hold-and-transfer pattern | Dialler call-state event stream | Rolling 14 days, per queue baseline | Amber: 1.3-1.6x baseline; Red: > 1.6x baseline |
| 4 | After-call cadence (variance) | Dialler after-call-work event | Rolling 30 days, agent-own baseline | Amber: 1.5-2 sigma; Red: > 2 sigma above own mean |
| 5 | Quality-loop closure | QA workflow tool flag timestamp | Per flag, in working days | Amber: 3-5 days open; Red: > 5 days open |
| 6 | Team focus density (median) | Aggregate of row 2 across pod | Rolling 30 days, pod median | Amber: 5-8 pts below site median; Red: > 8 pts below |
| 7 | Team adherence variance (spread) | Aggregate of row 1 across pod | Rolling 30 days, top-bottom quartile spread | Amber: 12-18 min spread; Red: > 18 min spread |
| 8 | Team quality-loop ageing | Count of QA flags > 5 days open in pod | Weekly count | Amber: 3-5 flags; Red: > 5 flags |
| 9 | Flag-to-acknowledge time | QA workflow timestamp deltas | Per agent, per flag | Amber: 24-48 hours; Red: > 48 hours |
| 10 | Coaching-note count | QA workflow tool | Per agent, per week | Amber: 3-4 open notes; Red: > 4 open notes |
| 11 | Appeal status | QA workflow tool appeal log | Per coaching note | Tracking only — open / acknowledged / resolved |
| 12 | Consent record reference | HRIS consent ledger | Per agent, per purpose | Must exist and be retrievable |
| 13 | Per-signal retention line | Data retention policy register | Per signal | Must be itemised, not blanket |
| 14 | DPO sign-off date | Compliance register | Per quarter | Must be within the last 90 days |
Two things to notice about the sheet. First, the source-system column is empty of any agent-endpoint capture — no screenshot tool, no keystroke recorder, no webcam stream. Every signal pulls from a system the operations leader can name without engineering help. Second, the audit-trail rows (12-14) sit on the same sheet as the agent signals. The DPO does not chase a separate compliance binder — the evidence is where the work happens.
Free: DPDP Vendor Risk Assessment Worksheet
Run the 16-question DPDP scoring sheet on the agent-monitoring vendor in your stack today. Section 4 proportionality, Section 8 retention, Section 10 SDF, Section 11 rights — line by line.
Open the worksheetWeekly cadence — the team leader's 30-minute slot
The weekly review is what makes the scorecard a coaching loop instead of a verdict. The team leader runs a single thirty-minute slot per pod of twelve to fifteen agents at the close of every week. The slot has three blocks — agent-level walkthrough, coaching-note capture, follow-through schedule. The team leader does not change thresholds during the weekly slot; thresholds are a monthly calibration job.
- Block 1 — Walkthrough (15 min). Open the sheet at the agent level for each amber and red flag in the pod that week. Three to five agents typically have at least one flag; the rest sit clean. Read the signal, the source system, the threshold, and the actual value. The walkthrough is signal-by-signal, not agent-by-agent, because the pattern across the pod usually matters more than any single agent line.
- Block 2 — Coaching-note capture (10 min). Every red flag generates a written coaching note inside the QA workflow tool. The note carries the signal, the threshold breach, the suggested intervention, and the follow-through date. The agent sees the note on the same minute the team leader writes it. Amber flags do not generate notes — they generate a verbal check-in, scheduled for the next 1:1.
- Block 3 — Follow-through schedule (5 min). The follow-through date for each note is added to the team leader's coaching calendar. The follow-through is the moment the loop closes — either the signal trend recovers, the note is appealed and adjusted, or the note is escalated to the Ops Head as a team-system gap rather than an agent gap.
The weekly cadence is short because it is run on a small surface. Twelve agents, five signals, a one-page sheet, a thirty-minute slot. The shorter the slot, the more honest the coaching tends to be — the team leader cannot dilute the signal with narrative when the surface is this tight.
Monthly cadence — calibration, not coaching
The monthly review is run by the QA lead and the Ops Head together, with the DPO on call for the audit-trail block. The discipline is calibration — recheck the baselines, adjust the thresholds against the team's actual range, refresh the consent and retention lines. The monthly review is not where individual agents are reviewed; that work sits in the weekly slot.
| Block | Owner | Focus | Output |
|---|---|---|---|
| Threshold recheck | QA lead | Rows 1-5 thresholds against actual 30-day range per queue | Updated amber / red lines, signed off by Ops Head |
| Pod review | Ops Head | Rows 6-8 across all pods on the same queue | Pod-comparison view, intervention list for outlier pods |
| Coaching-loop audit | QA lead | Rows 9-11 — flag age, open note count, appeal status | Loop-health report, escalations to People Ops |
| Compliance audit | DPO | Rows 12-14 — consent ledger, retention lines, sign-off date | Quarterly attestation, refreshed sign-off |
The monthly cadence is what keeps the scorecard from drifting into a fixed-target scorecard over a quarter. The amber and red thresholds are not absolute numbers carried forward — they are recalibrated against the team's actual 30-day range every month. The 850-seat Pune BPO we referenced earlier ran this monthly recalibration discipline for two quarters and watched the amber-flag rate stabilise around 15 to 18 per cent of agents — the band that produces coaching signal without flagging everyone, which is the band a fair scorecard sits in.
Per-agent and team-leader roll-up — the same sheet, two lenses
The roll-up pattern matters because team leaders and Ops Heads read different layers of the same data. The agent-level lens shows fourteen rows for one agent. The pod-level lens shows three roll-up rows (6-8) and a small table of agent names per signal. The site-level lens shows the pod-comparison view the Ops Head runs at month close.
- Agent lens. Read by the team leader weekly and the agent themselves on the same screen. Carries every flag, every coaching note, every appeal. The agent's right to see what is held on them under DPDP Section 11 is operationalised by this lens being agent-visible, not promised in a policy document.
- Pod lens. Read by the team leader monthly and the Ops Head at quarter close. Carries the three team-roll-up signals plus a small leaderboard-free table of agent names against each signal. No single-number score — the lens shows where the signals cluster, not who is ranked where.
- Site lens. Read by the Ops Head and the QA lead. Compares pods on the same queue against each other and against the site median. Used for resource and coaching investment decisions, not for agent-level decisions.
Free: Productivity Report Template (Weekly + Monthly)
The fill-in productivity report template managers use to wrap the weekly and monthly cadences into a one-page rollup the leadership cadence reads. Five-section weekly + seven-section monthly, anti-pattern callouts included.
Open the report templateWhat the team-leader review looks like for a multi-client BPO
The pattern shifts when the BPO runs multiple Data Fiduciaries — three or four clients in the same pod, each with their own contact-centre queue, QA criteria, and Master Services Agreement consent surface. The fourteen-row sheet still applies; the calibration overlay changes.
Rows 1, 2, and 4 — shift adherence, focus density, and after-call cadence variance — stay agent-baselined regardless of client. These three signals do not need a per-client overlay because they describe the agent's pattern of working, not the customer-side outcome. Rows 3 and 5 — hold-and-transfer pattern and quality-loop closure — recalibrate per client-queue, because escalation paths and QA criteria differ between accounts. A multi-client BPO runs the scorecard at the agent level and adds the per-client lens at the queue level for those two rows.
The audit-trail rows (12-14) become Data Processor obligations per client engagement under DPDP Section 8(4). The consent ledger reference (row 12) lives in the HRIS plus a parallel ledger in the client's Data Fiduciary system. The per-signal retention line (row 13) is documented per client in the Master Services Agreement. The DPO sign-off date (row 14) is per client, not per BPO — the DPO attests once per client engagement per quarter. Counsel review is required for cross-client signal blending.
The four anti-patterns to drop from the existing scorecard
Most BPO scorecards we have reviewed in 2026 procurement diligence carry at least three of the patterns below. Drop them before turning on the fourteen-row template.
- Single-number composite score. The composite hides which signal is firing and gives the agent nothing to coach against. Drop in favour of the five-signal panel.
- Daily AHT punishment. AHT is a queue-level operational metric, not an agent-level coaching metric. Move AHT to the queue dashboard the operations leader reads.
- Fixed organisation-wide thresholds. Thresholds calibrated against an internal target band rather than the team's actual range are how scorecards drift into theatre. Use the team's rolling baseline.
- One retention knob. One retention period across screenshot frames, keystroke logs, call recordings, and aggregate score histories is the DPDP Section 8 failure mode. Itemise per signal in row 13.
FAQ
Frequently asked questions
What does a DPDP-safe BPO scorecard actually look like, row by row?
Fourteen rows across five columns. The rows are five agent signals, three team-roll-up signals, three coaching-loop fields, and three audit-trail fields. The columns are signal name, source system, baseline window, amber threshold, and red threshold. Rows one to five carry shift adherence, focus density, hold-and-transfer pattern, after-call cadence, and quality-loop closure. Rows six to eight roll up to team level — team focus density, team adherence variance, and team quality-loop ageing. Rows nine to eleven track the coaching loop — flag-to-acknowledge time, coaching-note count, and appeal status. Rows twelve to fourteen carry the audit trail — consent record reference, retention line per signal, and DPO sign-off date. The template ships as a single sheet the team leader runs at week close and the Ops Head reviews at month close.
How often should an India BPO refresh the agent scorecard — weekly or monthly?
Both, with different purposes. The weekly cadence runs the coaching loop — team leader reviews five agents in a 30-minute slot, walks through amber and red flags on the five signals, captures the coaching note, and schedules the follow-through. The monthly cadence runs the calibration loop — Ops Head and QA lead recheck the baselines per queue, adjust amber and red thresholds against the team's actual range, and refresh the consent and retention lines. Weekly cadence drives behaviour change. Monthly cadence keeps the thresholds honest. Skipping the monthly recalibration is the failure pattern that turns a signal-based scorecard into a fixed-target scorecard within a quarter.
What does the team-leader roll-up look like for a 12-agent pod?
The team-roll-up is a derived view of the same fourteen rows, aggregated to the pod. Team focus density is the median of the agent line, not the mean — medians are less sensitive to one off agent. Team adherence variance is the spread between the best and worst quartile, not the average. Team quality-loop ageing is the count of QA flags older than five working days. The team leader runs the weekly review off the agent-level sheet and uses the team-roll-up to track pod health across weeks. The Ops Head uses the team-roll-up to compare pods on the same queue. The agent sees the agent-level sheet on the same screen the team leader sees, on the same minute.
How do you build a calibration baseline without screenshot or keystroke data?
Run a 90-day rolling baseline per agent on the five signals using the data the dialler, CRM, ticketing tool, and QA workflow already publish. Shift adherence comes from the workforce-management roster joined to the dialler queue-ready event log. Focus density comes from the application-context signal — which work tool has OS focus, rolled up per interval. Hold-and-transfer pattern comes from the dialler call-state event stream. After-call cadence comes from the dialler after-call-work event. Quality-loop closure comes from the QA workflow tool's flag-to-acknowledge timestamp. None of these requires a new capture layer on the agent endpoint. The baseline window of 90 days is wide enough to absorb shift mix and seasonal variance and short enough to refresh inside a quarter.
What does the appeal field on the scorecard actually do?
Every red flag on the scorecard generates a coaching note the agent can contest in writing inside the QA workflow tool. The appeal field carries the appeal status — open, acknowledged, resolved with adjustment, resolved without adjustment — and links to the appeal record. The appeal field is what operationalises the right of correction under DPDP and what evidences human-oversight requirements under workforce AI obligations. Without an appeal field, the scorecard is a one-way verdict rather than a coaching loop, and the audit trail breaks. The Data Protection Officer reviews the appeal log monthly as part of the consent-and-rights audit. Counsel review is required for each deployment.
Related reading on gStride
See the 14-row scorecard live on your dialler and QA stack
gStride reads the five DPDP-safe signals directly from your CRM, dialler, ticketing tool, and QA workflow. Agent-visible dashboards, per-pod roll-ups, per-client overlays. No screenshots. No keystroke capture.
See the platform Book a 30-min call