01 · What breaks
Design for the failure modes, not only the happy path.
The value of 90-day and 180-day retention follow-up depends on what happens when records are incomplete, evidence conflicts or the next action carries consequence.
Different questions at each interval
Make this visible in testing, ownership and review rather than leaving it as an implementation surprise.
Unreachable graduates removed
Make this visible in testing, ownership and review rather than leaving it as an implementation surprise.
Job change counted as failure
Make this visible in testing, ownership and review rather than leaving it as an implementation surprise.
Sensitive reasons shared with employers
Make this visible in testing, ownership and review rather than leaving it as an implementation surprise.
02 · Minimum viable model
Start with graduate follow-up episode and the decisions around it.
The smallest useful model for this page connects the following records. Each field should have a purpose, owner and retention rule.
03 · Implementation sequence
Move through reviewable gates.
A complete small workflow is safer and more useful than a large collection of unfinished modules.
Define follow-up windows and states
Agree the owner, evidence and exit condition before moving forward.
Verify safe contact channels
Agree the owner, evidence and exit condition before moving forward.
Schedule multilingual outreach
Agree the owner, evidence and exit condition before moving forward.
Collect comparable evidence
Agree the owner, evidence and exit condition before moving forward.
Route support needs and conflicts
Agree the owner, evidence and exit condition before moving forward.
Report retention with nonresponse
Agree the owner, evidence and exit condition before moving forward.
04 · Evidence
Measure whether the work became clearer.
Usage alone is not success. Review the quality, timeliness and decisions the system is meant to improve.
05 · Responsible automation
Let systems prepare. Keep consequence with people.
Systems can support
- Verify safe contact channels
- Schedule multilingual outreach
- Collect comparable evidence
- Route support needs and conflicts
People should decide
- Responding to distress or exploitation
- Resolving employment conflict
- Offering programme support
- Approving the reported status
06 · Common questions
Practical questions teams ask.
Use these answers to clarify scope, compare approaches and identify the decisions your team needs to make before implementation.
Can AI safely help a nonprofit with 90-day and 180-day retention follow-up?
A retention follow-up mission schedules consented contact at defined intervals, gathers current employment evidence and prepares exceptions for human review. Denominators must retain unreachable and unresolved graduates so retention is not overstated.
What information does an AI mission for 90-day and 180-day retention follow-up need?
Start with graduate follow-up episode. The minimum useful record set covers graduate and consent state, placement baseline, follow-up window and attempt, current role, employer and wage evidence, change, exit or support need and nonresponse, conflict and reviewer decision. Remove any field that does not support a service, safeguard, decision or reporting obligation.
What can AI execute and what must a person decide in 90-day and 180-day retention follow-up?
People should retain responsibility for responding to distress or exploitation, resolving employment conflict, offering programme support and approving the reported status. Software or AI may prepare evidence and queues, but it should not silently make those decisions.
What evidence and audit trail should 90-day and 180-day retention follow-up produce?
Keep the source, definition, time, owner, status and review decision together. Monitor contact and completion rate, verified retention by window, job change and exit reasons and unresolved and support cases, and preserve limitations instead of cleaning them out of the report.
Where can an AI mission for 90-day and 180-day retention follow-up fail or require escalation?
The main risks are different questions at each interval, unreachable graduates removed, job change counted as failure and sensitive reasons shared with employers. Test these with representative cases and keep the consequential judgement with a named person.