A programme is not a bigger internship. It is a set of decisions and documents that let the second cycle start from where the first one finished — which is why most companies run a good first cycle and then quietly never run a second.
This guide covers the five decisions that determine everything else, the mentor problem that sinks most programmes, and what to measure. It is written for the person who has been asked to set one up.
Step 1 — Decide what the programme is for, and pick one
Internship programmes are asked to do four things: build a hiring pipeline, add delivery capacity, build employer brand on campus, and satisfy a CSR or academic-partnership commitment. All four are legitimate. What does not work is claiming all four, because they imply different designs and contradictory metrics — a pipeline programme is measured on conversion, a capacity programme on output, a brand programme on campus reach. A programme measured on all three is measured on none.
- Pipeline
- Design for conversion. Fewer interns, longer internships, seats agreed before the cycle, senior mentors. Measure conversion and conversion-acceptance.
- Capacity
- Design for output. Scoped, self-contained projects; more interns per mentor is survivable. Measure deliverable completion.
- Employer brand
- Design for reach and experience. Cohort visible, showcase at the end, alumni network afterwards. Measure would-recommend and campus application growth.
- CSR / academic
- Design for the commitment as written. Be honest internally that hiring outcomes are not the goal, so nobody measures it as if they were.
Step 2 — Solve the mentor problem before anything else
The single largest predictor of whether an internship works is how many hours the intern actually gets with a mentor. It is also the thing every programme over-estimates, because volunteering to mentor and having four free hours a week are different things, and the gap only becomes visible in week three.
- Make a named mentor with confirmed capacity a precondition for hosting an intern — not a nice-to-have, a gate.
- Cap it at one intern per mentor for a first cycle. Two only for mentors who have done it before.
- Put mentoring in the mentor's own objectives. Unrecognised work is the first thing dropped when a deadline arrives.
- Brief mentors on what is expected: a weekly one-to-one, a written midpoint review, a written final review, and the certificate line.
- Measure mentor hours by asking the interns, not the mentors. Across a cohort the two answers routinely differ by half.
Step 3 — Require a scoped project before the JD is written
A JD written before the project is scoped promises work that does not exist, and the intern discovers this in week two. Make a written scope with a definition of done the price of admission for a hosting team. It takes a mentor thirty minutes and it prevents the most common failure mode in the entire programme.
Step 4 — Run a cohort, not a stream
| Cohort intake | Rolling intake |
|---|---|
| One onboarding session for everyone | A separate onboarding per intern, repeated |
| Interns support each other; questions get asked peer-to-peer | Each intern is alone and asks the mentor everything |
| A showcase at the end that the business turns up to | No natural moment for the work to be seen |
| Comparable reviews across the cohort | Reviews with no reference point |
| Feels rigid to hosting teams | Feels flexible, and quietly costs several times more |
Step 5 — Decide conversion seats before the cycle starts
Interns ask about conversion in week one whatever you say in the JD. A programme that cannot answer produces the worst outcome available: the interns most likely to convert are the ones with other options, and they take them. Decide how many seats exist, tell the interns the truth about the number, and communicate outcomes within two weeks of the internship ending — even when the answer is no. The failure mode is not rejecting people; it is telling them nothing for two months.
Step 6 — Write the artefacts down
- Internship policy and stipend bands
- Offer letter, and an agreement for longer or data-sensitive roles
- JD template and per-role variants
- Interview kit and scorecard per role family
- Pre-arrival and onboarding checklists
- Midpoint and final review forms
- Completion certificate template and an issuing register
- The anonymous intern feedback survey
Step 7 — Measure five things
| Metric | Where it comes from | What it tells you |
|---|---|---|
| Deliverable completion rate | Final reviews | Whether projects were scoped realistically. |
| Mentor hours actually delivered | Intern feedback survey | Predicts almost every other outcome. |
| Conversion rate and conversion-acceptance | Hiring records | Two numbers. High conversion with low acceptance means you decided too late. |
| Would-recommend to a friend | Intern feedback survey | Your campus reputation, measured before it reaches campus. |
| Cost per converted hire | Finance + hiring records | The number for the budget conversation. Cost per intern is not. |
What a first cycle should look like
Five to ten interns, one cohort, two standard durations, every artefact written down even if imperfect, and a genuine retrospective at the end using the anonymous survey. One or two interns never becomes a programme because nothing gets recorded; thirty in cycle one overwhelms an owner who is still discovering what breaks. Cycle two is where a programme is actually made — it is the first time the documents get used by someone who did not write them.
Frequently asked
How long does it take to set up an internship programme?+
About six to eight weeks of part-time work before the first cycle: the purpose decision and budget, then the document set, then mentor briefings. The documents are the slow part, and they are also the part that makes cycle two cheap.
How many interns should we start with?+
Five to ten. Enough to run as a cohort, few enough that one owner can see everything that goes wrong. Fewer than three never generates the artefacts; thirty in cycle one generates them under pressure and badly.
Do internship programmes need a dedicated owner?+
One named person with time actually allocated, yes. Programmes owned collectively by "HR", or by a manager as an added duty, run once well and then stop when that person is busy — which is the normal way internship programmes end.
Should every internship be paid?+
We would say yes. Beyond the ethics, an unpaid programme selects for candidates who can afford to work for free, which is a filter nobody would defend if it were written into the JD explicitly.
What is the most common reason internship programmes fail?+
Mentor capacity. Everything else — unscoped projects, disengaged interns, conversion decisions nobody can justify — usually traces back to a mentor who volunteered in good faith and did not have four hours a week. Gate hosting on it.