Appraisal That Punished the New Hires
At a greenfield auto-component plant in Khurda commissioned 14 months ago, 180 operators work three shifts. Sixty-five were hired in the last eight months. Monthly internal rejection is 12.6% against a customer target of 2%; scrap value last month was Rs 11.4 lakh. A five-day off-the-job induction was cut to one day when a customer pulled a delivery date forward. Production managers routinely refuse to release operators for training, and Rs 13 lakh of an Rs 18 lakh training budget is unspent. The plant uses forced ranking: 10% must be rated 'needs improvement' and 4% exit through a performance improvement plan. New-hire attrition is 31% against a plant average of 12%. The customer has warned that 20% of order volume will move to a Tamil Nadu supplier if defect rates do not fall this quarter.
1. Before any training is commissioned, how would you establish whether the 12.6% rejection rate is a training problem at all? Set out the analysis you would run.
Read the worked answer
Do not commission a programme yet. In a fourteen-month-old plant a 12.6% rejection rate has at least five candidate causes, and training is only one: (i) operator skill, most likely among the 65 recent hires; (ii) machine setting and tooling, since the plant is new; (iii) incoming material or supplier variation; (iv) process discipline and shift handover; (v) measurement - the rate may be counted wrongly.
Data to pull, and what each result would mean:
- Defects by operator tenure (under 1 month, 1–3, 3–6, over 6). A steep fall with tenure points to skill.
- Defects by machine and by shift. One machine dominating across all tenures points to equipment, not people.
- Defects by incoming lot or supplier; clustering suggests material.
- Defect type: dimensional drift on a new machine suggests setting, assembly errors suggest training.
- A sample audit of operators against the standard work instruction and time study.
The three levels of needs analysis. Organisational: the customer's 2% target and the threatened 20% of volume dwarf the entire Rs 18 lakh budget, so the spend is justified if correctly targeted. Task: list the knowledge, skill and error-cost of each operation. Person: use a practical skills test and structured observation, not the forced ranking, to decide who needs what.
Then act. If tenure is the driver, the remedy is certified on-the-job training with a buddy for the first 90 days, not a classroom course. If machine or material is the driver, no amount of training will move the number - and the 31% new-hire attrition suggests the larger loss is already elsewhere.
2. Design a training programme with an evaluation plan that will persuade the customer and the general manager that it worked.
Read the worked answer
Build the evaluation alongside the programme, not after it. Four levels, with a control.
Level 1 - reaction, day 1. A short form on relevance and confidence. Useful as an early warning only, never as evidence of value.
Level 2 - learning, day 5 and day 15. A practical on-the-line skills test against the standard work instruction with a pass mark (say 85% with zero critical errors), plus a defect-recognition test using photographs of real rejects from this plant. Measure before and after.
Level 3 - behaviour, 30/60/90 days. Supervisor observation checklist covering adherence to work instruction, correct tool setting, first-piece inspection and escalation. Two observations per operator per month by a trained observer, not the line supervisor alone.
Level 4 - results, monthly for six months. Rejection percentage and scrap value against the Rs 11.4 lakh baseline, first-pass yield, ramp-up time for a new operator, and new-hire attrition against the 31% baseline. Convert rejection into rupees so the return is visible.
Make it causal. Run the redesigned induction on one line or shift first, leave a comparable line unchanged for eight weeks and compare. Defects and attrition may share a cause - a poor first experience - so an uncontrolled before-and-after will persuade nobody.
Unblock the budget. Rs 13 lakh unspent is not a training problem but a release problem. Agree a written release calendar with the production manager, name the cover operators, and carry it as a plant KPI rather than a training-department request.
3. Is the forced distribution defensible for a workforce where 65 of 180 operators have under eight months' service? Advise management.
Read the worked answer
The curve is being applied to people who have not had a fair chance. Sixty-five of 180 operators have under eight months in the job. A forced 10% 'needs improvement' and 4% PIP exit assumes a mature population with comparable opportunity; among recent hires it largely measures time on the job. Four percent of 180 is seven exits a year, and with new-hire attrition at 31% the plant is losing people faster than it can train them - the 12.6% rejection rate is partly a symptom of that churn.
Legal and practical risk. Rating a probationer against a quota when his induction was cut from five days to one is hard to defend if challenged. Every PIP exit needs a documented support plan, communicated targets and a review period, or it looks like a shortcut to headcount reduction.
Recommendation. (i) For the first six months use absolute standards - a new-hire appraisal against skill milestones and safety behaviour, not a curve. (ii) Apply the forced distribution only to confirmed operators, and publish the criteria in advance. (iii) Keep the PIP but make it a genuine improvement plan with a buddy, extra practice hours and weekly review over a 60-day window; the 4% should be an outcome, never a target. (iv) Feed the appraisal data into the training needs analysis instead of using it as an exit filter - the same data that flags a struggling operator should trigger training, not a rating.