Evaluating an intern is not a smaller version of a performance review. There is no baseline, the engagement is short enough that trajectory matters more than level, and the person receiving the feedback is often hearing workplace feedback for the first time in their life. All three change what good evaluation looks like.
This guide covers what to assess, when, and how to say it.
Assess against what you agreed, not against a standard nobody stated
The written project scope is the evaluation criterion. If it was never written down, the assessment becomes a comparison against an implicit standard the intern was never told about — which is unfair and, more practically, produces a conversion decision nobody can defend. Everything else in this guide assumes a scope exists.
Five dimensions, and what each actually means
| Dimension | What you are assessing | What it is not |
|---|---|---|
| Quality of work | Whether output needed rework; whether they checked it before submitting. | Whether it was perfect first time. |
| Learning speed | Time from "shown once" to "does it unaided". | How much they already knew on day one. |
| Ownership | Whether they raised blockers early and owned outcomes including failures. | Working late. |
| Communication | Whether a reader or listener could follow them and act on it. | Fluency in English, or confidence. |
| Coachability | Whether feedback given in week three changed behaviour by week five. | Agreeing with feedback pleasantly. |
The midpoint review is the one that matters
A review at the end is a verdict. A review at the midpoint can still change the outcome, which is the only reason to hold one. Keep it to four things: where the deliverable stands, two things going well with examples, at most two things to change with examples, and — the section most companies omit — what the mentor will do differently. Most first-half underperformance is a scoping or access problem rather than an intern problem, and a form that only asks about the intern will never surface it.
Writing feedback an intern can act on
- Behaviour, not personality
- "You went quiet for three days on the blocked ticket" can be acted on. "You lack initiative" is a label, and a first-workplace label is remembered for a long time.
- One development area, not six
- Six produces zero changes. Pick the one that would most change their next six months and let the rest go.
- Evidence for every judgement
- A score without an example is an impression with a number attached, and it will not survive being questioned by the intern or by a conversion panel.
- Say the hard thing plainly
- Interns told "you did great" and then not converted learn nothing and tell everyone. Kind and clear beats kind and vague.
- Separate the person from the project
- Projects fail for reasons that have nothing to do with the intern. Say which is which explicitly, or they will assume the worst reading.
Rating biases to design around
- Recency — the last two weeks dominate a twelve-week assessment. The midpoint review is the structural fix; it puts the first half on record while it is fresh.
- Halo — one strong dimension lifts all the others. Requiring separate evidence per dimension is what breaks it.
- Similarity — mentors rate interns who resemble them higher. Anchored rubrics and a second reader reduce it.
- Leniency — nobody wants to give a twenty-year-old a low score. Anchors where 3 means "yes, convert" rather than "exceptional" make honest scoring easier.
- Cohort comparison — ranking against this cohort rather than against the bar. Ask "would I hire this person externally today", not "were they in the top three".
Close the loop with the intern and with the programme
Every intern should leave with their written review, a clear conversion answer within about two weeks, a completion certificate and, on request, a reference. And every cohort should leave behind an anonymous feedback survey that you read across the cohort rather than one by one. Individual evaluations tell you about individual interns; the pattern across the cohort tells you about your programme — and the programme is the thing you can actually change.
Frequently asked
What criteria should we use to evaluate an intern?+
Quality of work, learning speed, ownership, communication and coachability — assessed against the written project scope agreed at the start. Without that scope you are assessing against an implicit standard the intern was never told, which is both unfair and indefensible at conversion time.
How often should interns be reviewed?+
Two written reviews — midpoint and final — plus a weekly one-to-one that is a conversation rather than a form. The midpoint is the one that earns its place, because it is the only one that can still change the outcome.
How do you give negative feedback to an intern?+
Describe behaviour with a specific example, name one thing to change rather than six, and be plain. Interns told everything is fine and then not converted learn nothing and say so publicly. Kind and clear is better received than kind and vague.
Should interns be compared with each other?+
Only when seats are genuinely scarce, and never as the primary criterion. Cohort ranking converts the best of this cohort rather than someone who meets the bar. The better test is whether you would hire them externally today knowing everything you now know.
What if the internship failed because of us, not the intern?+
Say so, in the review and in the reference. Unscoped projects, absent mentors and missing access are common and they are the company's failures. An intern penalised in writing for a failure that was not theirs is the version of this that gets discussed publicly, and rightly.