The clearest sign that an internship is going to fail shows up in the first week. The student arrives, and nobody has ordered them a laptop, created an email address or decided who they report to. They spend four days reading the website and reorganising a shared drive. By the time the work is ready for them, they have already learned the lesson you did not want to teach: that this place does not plan.
Hiring interns at a small business is not a scaled-down version of what a large company does. You do not have a recruiting function, a rotation programme or a cohort of forty. You have one or two students, a team that is already busy, and a summer that will be over before you have finished apologising for the slow start. That constraint is not a disadvantage. It forces you to be specific, and specificity is the whole game.
Scope the work before you write the posting
The test I apply is simple. Can you name one deliverable the student will hand over at the end, that a colleague would otherwise have to do, and that will still be in use six months later? If you cannot, you do not have an internship. You have a favour you are doing for someone's nephew, and you should be honest with yourself about that.
A deliverable is not "support the marketing team" or "assist with operations." It is a thing: a rebuilt pricing sheet, a documented intake process, a cleaned-up customer list with duplicates resolved and sources tagged, a set of written answers to the questions your staff get asked constantly. I have written before about the value of capturing the questions your team answers every day, and that exercise is close to a perfect intern project. It requires sitting with people, listening carefully, writing plainly and checking the result. It produces something the business keeps.
Write the deliverable into the posting. Students self-select on it, and the ones who are excited by a concrete piece of work are the ones you want. A vague posting attracts applicants who are optimising for a line on a résumé, which is a rational thing for them to do when you have given them nothing else to optimise for.
Scope more than one project, because the first will either finish early or stall. I keep a short list of the work that genuinely matters but never reaches the top of anyone's week. That list is where good internships come from.
Pay them, and know what the pay is buying
Pay the intern. I will not dress this up as a legal argument, because the rules around unpaid internships at for-profit companies are narrower than most owners assume and you should ask your employment counsel rather than me. I will make the operational argument instead.
An unpaid intern is a volunteer, and you cannot manage a volunteer the way you manage an employee. You cannot reasonably ask for a deadline, a full day or a redo. You will not feel entitled to give hard feedback, and they will not feel obliged to take it. The moment money changes hands, the relationship becomes legible to both sides. You are buying the right to have expectations.
Pay also changes who can apply. Students who need an income over the summer will take the paid job somewhere else, and in a market like South Florida that is a large share of the talent. Universities here serve a lot of students who work. If you will not pay, you are not competing for Miami student talent — you are competing for the subset of it that can afford you, which is both smaller and less interesting.
Benchmark your rate against what comparable roles in your market are posting, not against what you think a student is worth. And treat the cost the way you would treat any other allocation decision, which is to say honestly: this is a line item, it competes with other uses of the same money, and it should earn its place. I have written separately about how I think about where the next dollar goes, and an internship survives that test more often than people expect, because the output is durable and the recruiting option value is real.
Name a supervisor, and give them time, not just the title
The failure mode in a small company is that the intern reports to the owner, who is in the field four days a week. The student sits with the team, the team is kind to them, and nobody is actually accountable for whether they learn anything.
Name one supervisor who is not the owner if you can, and put the meetings in the calendar before the student starts. Thirty minutes at the start of the week to set the work, thirty minutes at the end to review it. That hour is the programme. Everything else is logistics. If your supervisor cannot protect one hour a week, you are not ready to host an intern and you should wait a cycle.
Tell the supervisor explicitly that their job is to give feedback that stings a little. Small companies are warm places, and warmth curdles into vagueness fast. A student who hears "this is great" for ten weeks has been robbed. The most useful thing I was ever given early in my career was a piece of work handed back with the parts that did not hold up circled. It was not pleasant and it was not ambiguous.
Make the first week boring and complete
Everything the student needs on day one should exist before day one: the laptop, the logins, the email address, a desk, a named person to ask stupid questions. Then spend the first morning on context rather than tasks. What the company sells, who buys it, what the margin looks like in plain terms, what breaks when things go wrong.
Give them a glossary. Every business has thirty words that mean something specific internally and nothing outside, and students spend weeks pretending to understand them. A one-page glossary saves a fortnight of quiet confusion, and the intern can extend it as they go — which becomes a second deliverable you did not plan.
Set the tool expectations the same week. Students now arrive fluent in AI tools and entirely unclear on where you draw the lines: what can go into a model, what cannot, what has to be checked before it leaves the building. You owe them that clarity rather than a raised eyebrow in week six. The same rules I would set for a small team using AI apply to the intern, and writing them down once covers both.
Recruit where the students already are
University recruiting at a small company does not look like a careers fair. You do not have the booth budget or the brand, and standing in a hall between two banks is a losing position. What works is relationships with the specific people inside a university who know which students are ready: programme directors, honours advisors, faculty who run capstone projects, the staff who run alumni and employer engagement.
I serve on a board at the FIU Honors College, and the thing that is most obvious from that seat is how much untapped matching capacity sits inside a university. There are thousands of students there, and the people who advise them are constantly asked by students for real-world work and rarely asked by employers for candidates. Being a small employer who calls the same advisor every year and reports back honestly on how the last student did makes you memorable in a way a job posting never will.
Do it on a cycle. Recruiting for a summer that starts in May begins in the autumn, not in April. Showing up once produces nothing; showing up every year produces a pipeline, which is the same logic that applies to any kind of community presence.
Decide about conversion early, and say the honest thing
Three weeks before the end, make a decision and communicate it. There are only three honest outcomes: an offer, an invitation to return next cycle, or a thank-you with a reference and no promise. All three are fine. What is not fine is the soft maybe that leaves a student waiting on a role you were never going to create.
If you are making an offer, make it in person and make it specific about the work, not just the title. If you are not, tell them why in terms they can use elsewhere — the skill that was missing, the thing that would make them a stronger candidate next year. A student who leaves with a clear, kind, specific assessment will send you their classmates. That is how a pipeline compounds for a company too small to have a brand.
And write down what you learned about your own process before you forget it. Which project landed, which stalled, how long the setup really took, what the supervisor wished they had known. Keep the notes in the same place every year. A small employer cannot out-recruit anyone, but we can out-remember them — and the second internship is always better than the first for exactly that reason.