01 · What breaks
Design for the failure modes, not only the happy path.
Spreadsheet trackers break in predictable ways once more than one person edits them. These four are the ones that turn a usable sheet into an argument at reporting time.
Free-text states that cannot be counted
If one officer types working, another Placed - OK and a third yes, no formula can count them. Use data validation so every status column accepts only values from a lookup tab. Free text belongs in a single notes column that no formula reads.
Overwriting historical outcomes
Typing the 180-day answer over the 90-day one erases the trend and changes last quarter's report without anyone noticing. Give each window its own columns for status, date and attempts, protect completed windows, and keep the snapshot tabs as the record of what was reported.
Unreachable graduates deleted
A row deleted because nobody answered quietly raises the retention rate. Mark the row unreachable after three logged attempts and keep it. Hiding rows with a filter for a working view is fine; removing them from the sheet is not, because the base shrinks with them.
Evidence links shared too broadly
Payslips and appointment letters carry pay, addresses and sometimes ID numbers. A folder shared as anyone-with-the-link can travel to people who should never see it. Restrict the evidence folder to named staff, and keep only the document link and type in the tracker, not the details on the document.
02 · Minimum viable model
Lay out one row per placement, with a column group for each check-in.
Treat the sheet as a small database: a tab of candidate rows, a restricted tab of contact details, a lookup tab of dropdown values and a protected tab for each frozen snapshot. The MIS requirements template follows the same logic when the tracker later grows into a system.
| Column group | Columns | Allowed values |
|---|---|---|
| Identifiers | Programme ID, batch code, centre, course completion date, funder tag | Funder tag from the lookup tab |
| Placement claim | Employer, role, joining date, take-home pay, shift, claimed by | Dates as dates, pay as a number |
| Evidence | Evidence type, verification state, document link, verified by, verified on | Type: self-report, offer letter, joining call, salary slip, employer confirmation. State: verified, self-reported, disputed, needs review |
| Each window (30, 90, 180 days) | Due date (formula), attempts, status, contact date, reviewer | Attempts 0 to 3. Status: same job, changed job, not working, in education or training, unreachable |
| Change or exit | Reason, new employer, re-placement row ID | Reason from the lookup list |
| Next action | Action, owner, due date | Owner from the staff list |
03 · Implementation sequence
Move through reviewable gates.
Set the tracker up in this order and it stays countable after months of edits by several people.
Agree definitions and dropdown states
Before anyone enters a row, agree in writing what each status means and which evidence moves a placement to verified. Put the values on a lookup tab and point every status column at it. Adding a value later is a decision the MEL lead records with a date, not something anyone can type in.
Protect identifiers and contact fields
Keep names, phone numbers and any ID numbers on a separate tab or file that only coordinators can open, joined to the main rows by programme ID. Most people working the follow-up list need the ID, the due date and a way to call, not the full profile.
Enter the placement baseline
Enter each placement once, at joining, with the joining date, employer, role, take-home pay and evidence type. That row becomes the baseline every later window is read against. If the graduate changes job, fill in the change columns and leave the original placement untouched.
Schedule follow-up views
Calculate each window's due date from the joining date and build a filtered view of rows due this week, by owner. After three unanswered attempts, the window reads unreachable and the row joins the coordinator's list. On WhatsApp, a follow-up sent more than 24 hours after the graduate's last message must use a pre-approved template (as of October 2026).
Review conflict and nonresponse
Each week, a coordinator filters for disputed evidence, mismatched employer names and unreachable rows. Correct record errors directly, pass disputes to the placement officer, and note every decision in the reviewer column. A row where the graduate reports losing work gets a named owner and a call within about two days.
Freeze reporting snapshots
At each reporting date, copy the main tab as values into a dated snapshot tab and protect it. Reports cite the snapshot, never the live sheet. Pair two formulas on it, retention of all placed rows and the share of rows with a completed contact, so the base and the coverage always appear together.
Retention: rows whose 90-day status is same job or changed job, divided by all rows with a joining date. Coverage: rows whose 90-day status is anything other than unreachable or blank, divided by the same base. Keep both cells side by side on every snapshot tab, and never filter the base before dividing.
04 · Evidence
Measure whether the work became clearer.
These four checks show whether the sheet can still be trusted. When they slip, or when a second funder, a second centre or re-placements arrive, plan the move to an MIS and use this tracker as the clean migration source.
Several funders with different placement rules, more than one centre editing at once, graduates with two or three placements, or an auditor asking who changed a value and when. A spreadsheet answers that last question poorly. At that point the agreed dropdowns, the column spec and the snapshot history become the requirements for the system that replaces it.
05 · Responsible automation
Let systems prepare. Keep consequence with people.
Systems can support
- Calculating follow-up due dates from joining dates
- Restricting status columns to agreed dropdown values
- Filtering the rows due this week by owner
- Producing retention and coverage figures from a snapshot
People should decide
- Final verification
- Resolving evidence conflict
- Contacting employers
- Responding to support needs
06 · Common questions
Practical questions teams ask.
Questions coordinators and MEL staff ask when they set up, or inherit, a placement tracker.
How do you use a skilling placement and retention tracker?
Enter each placement once at joining, then work the weekly list of follow-ups that fall due. Record every attempt and answer in that window's columns, mark rows unreachable after three attempts, and have a coordinator review disputes each week. At each reporting date, freeze a snapshot and report from it, never from the live tab.
What columns should a placement tracker have?
Identifiers and a funder tag, the placement claim with joining date and take-home pay, the evidence type and its verification state, one group of columns per follow-up window, a change or exit reason with a link to any re-placement, and a next action with an owner. Contact details belong on a restricted tab, joined to the main rows by programme ID.
Who should maintain the tracker?
One coordinator should own the structure and the lookup values, while placement officers enter and update rows for their own candidates. The MEL lead approves any change to statuses or definitions and owns the snapshots. Only coordinators should open the contact tab and the evidence folder. Shared ownership with no named owner is how dropdowns drift back into free text.
When should a programme move from a spreadsheet tracker to an MIS?
Usually when several funders apply different placement rules, more than one centre edits at once, graduates have more than one placement, or someone needs a change history the sheet cannot give. The tracker then makes a good migration source, because its agreed states and column spec map directly onto the records an MIS needs.
How do you check that the tracker is complete and accurate?
Each month, count rows missing a programme ID or joining date, placements with no evidence type, windows overdue with no attempt logged, and any status value that is not on the lookup tab. Then pick a few rows at random and check them against the evidence folder. A short, regular check catches drift long before a funder audit does.