01 · What breaks
Design for the failure modes, not only the happy path.
Placement figures are where skilling programmes are judged hardest and where definitions are loosest. Each of the four failures below makes a placement rate look better on paper than it is.
Counting offers as placements
Offer letters arrive on placement-drive day, so they are tempting to count. Many candidates decline later, once they learn the shift timing, the commute or the deductions from pay. Record the offer, then a day-one joining call, and count a placement only once the start is confirmed.
Removing unreachable graduates from denominators
Silence is not neutral: the graduates who stop picking up are disproportionately the ones without a job, so dropping them flatters the rate. Keep every placed candidate in the base, mark the silent ones unreachable after the agreed attempts, and publish follow-up coverage next to retention each time.
Treating any employer response as proof
A quick "yes, they work here" from a supervisor's mobile is weaker than it sounds. Confirm the person answering is authorised, that the candidate agreed to the contact and that the details match the record. A contractor or brand name change at the same workplace is a record update, not an exit.
Automated calls without language and privacy checks
Automated voice or WhatsApp follow-ups in a language the graduate does not use comfortably produce silence or wrong answers. A call that mentions the job can also reveal employment status to whoever picks up a shared phone. Test scripts in the graduate's own language, confirm identity first and route any lost-job reply to a counsellor.
AI can read a salary slip and propose the employer name, CTC and take-home pay, but a person confirms the figures before they are saved, and slips are never pasted into public AI tools. It should not reject applicants on a predicted dropout score. An automated call is no substitute for a counsellor's conversation with someone who has just lost work.
02 · Minimum viable model
One outcome episode per placement, followed through every check-in.
Retention problems often begin at mobilisation, when seats are filled with candidates who never wanted the trade, so the outcome record keeps its links back to the batch. Donors tend to fund the top of the funnel and judge a programme on the bottom, which is why the youth and employability playbook treats placement as one continuous record from enrolment onward.
| Whose definition | Counts as placed when | Rule applied to the shared record |
|---|---|---|
| DDU-GKY 2.0 (as of October 2026) | Six months of employment, continuous or not, in wage, self- or gig work, paid no less than the applicable minimum wage. Training partners are expected to place 70% or more of those trained, with each placement tracked for 12 months. | Verified employment months inside the 12-month window add up to six or more, with each job's pay checked against the minimum wage |
| PMKVY 4.0 (as of October 2026) | No placement target. NSDC tracks certified candidates for a year after certification, and candidate data sits on Skill India Digital Hub (SIDH). | Report follow-up status and the SIDH candidate ID, not a placement rate |
| A CSR funder (illustrative) | Joined within 90 days of course completion, shown by a payslip | Joining date within 90 days of completion and evidence state verified by payslip |
| The programme's own target (illustrative) | In work with any employer at the 180-day check | The 180-day check records work with the first or a new employer |
03 · Implementation sequence
Move through reviewable gates.
These steps run from the day a candidate is counselled to the last check a funder asks for. Each one closes a gap that otherwise surfaces during an audit or a donor visit.
Define placement and retention states
Fix the placement states first (offered, joined, verified, self-reported, disputed, unreachable, needs review), then the check-in outcomes: still with the first employer, moved to another job, out of work, or back in education. A job change counts as retention in work, not failure. Write each funder's placed rule against these states, and recheck scheme guidelines at the start of every cycle.
Capture consent and safe contact channels
At enrolment, ask which number and channel the candidate wants used after training, in which language, and whether the programme may contact an employer for them. Record each permission separately with a date. On WhatsApp, business messages need the person's opt-in, and outside a 24-hour reply window only pre-approved templates can be sent (as of October 2026).
Verify initial placement evidence
Make a joining call on day one, then collect the first salary slip or an employer confirmation within about 30 days, whichever the funder accepts. Capture both CTC and take-home pay, because they differ. For self-employment, look for repeated evidence over months, such as sales records or customer confirmations, rather than a single launch photo.
Schedule interval follow-ups
Generate check-ins from the joining date at the windows each funder needs, commonly 30, 90 and 180 days, and at a year where a scheme tracks for 12 months. Keep a core of identical questions at every window so results compare, then add one or two questions specific to that window.
Route conflicts and nonresponse
Send record conflicts, disputed evidence and silence down separate paths. A mismatched employer name goes to a data fix, a disputed salary slip goes to the placement officer, and three unanswered attempts mark the case unreachable and alert a coordinator. When a graduate reports losing work, open a re-placement linked to the first job.
Report denominators and verification status
Every retention figure should carry its base and its coverage: retention among everyone who joined, the share actually reached, and the share verified by document. Break exits down by employer, trade and shift. If one employer accounts for many early exits, that calls for a conversation with the employer, not a judgement about candidates.
Say 200 candidates complete a course and 150 receive offers. 120 join, and 105 have a payslip or employer confirmation by day 30. At day 180, 78 are in work (64 in the same job, 14 in a new one), 17 are not working and 25 could not be reached. Retention is 78 of 120 joined (65%), with follow-up coverage of 95 of 120 (79%). Quoting 78 of the 95 reached (82%) would overstate it.
04 · Evidence
Measure whether the work became clearer.
Read the four measures together: a high verified placement rate means little if coverage is falling or reviewers are buried in exceptions. Teams still on spreadsheets can start from the skilling placement and retention tracker. For a system-level example, see the EdZola case study on streamlining programme management with Zoho Creator.
05 · Responsible automation
Let systems prepare. Keep consequence with people.
Systems can support
- Scheduling 30, 90 and 180-day check-ins per placement
- Reading salary slips for a person to confirm
- Applying each funder's placement rule to shared records
- Showing early exits by employer, trade and shift
People should decide
- Resolving disputed evidence
- Contacting an employer
- Interpreting informal or self-employment
- Support after job loss
06 · Common questions
Practical questions teams ask.
Questions placement officers, MEL leads and skilling programme heads ask when funders start auditing outcomes.
How do skilling programmes verify job placements?
Through a ladder of evidence, each step stronger than the last: the graduate's own report, an offer letter, a day-one joining call, the first salary slip, and confirmation from the employer with the graduate's permission. Most funders accept one or two of these. Record which evidence supports each placement, so a self-reported job is never presented as a verified one.
What records does a placement and retention system need?
A candidate record linked to batch and certification, an outcome episode for each job or enterprise with start date, pay and shift, evidence documents filed against the stage they prove, a follow-up record for each window with attempts and answers, and a reviewer decision where evidence conflicts. A re-placement links back to the first placement instead of replacing it.
Who should own placement verification in a skilling programme?
Placement officers usually own the joining call and evidence collection, a coordinator owns follow-up schedules and unreachable cases, and the MEL lead owns definitions and reporting. Contacting employers and settling disputed evidence should sit with a named senior person. Counsellors, not automated calls, should handle graduates who have lost work or report problems at the workplace.
How should retention be reported when many graduates cannot be reached?
Report retention against everyone who joined a job, and show follow-up coverage beside it. Graduates who cannot be reached stay in the base as unreachable, since silence more often means no job than a settled one. Reporting retention only among those reached usually overstates it, and the gap widens as coverage falls, so the two numbers should always travel together.
Do government skilling schemes require placement tracking?
It depends on the scheme, and the norms change. As of October 2026, DDU-GKY expects at least 70% of trained candidates to be placed, counts a placement after six months of employment and tracks it for 12 months. PMKVY 4.0 has no placement target, and NSDC tracks certified candidates for a year through Skill India Digital Hub. Check the current guidelines before designing to them.