Credit where it is due. The framework in this guide comes from Ben Lang, who published The guide to getting a job with cold email on the Next Play newsletter on August 3, 2025. His piece is written for people looking for tech jobs generally. This one adapts it for students applying to internships, with our own examples and figures. Read the original.
There is a version of the internship search where you refresh a job board, find a posting, submit an application into a pile of nine thousand, and wait. That is the version most students run, and it is the version where you have the least leverage you will ever have.
There is another version. You pick a company you actually want to work at, you find the person who would hire you, and you send them about 150 words. No posting required. No career fair. No warm introduction from an alum you do not have.
This works better than it sounds like it should, for a boring structural reason: a startup founder gets far fewer good, short, specific emails from competent people than you would guess, and hiring is close to the top of their list of problems. A resume in an applicant tracking system is a row. An email in an inbox is a person.
You can email any company you would work for. Not just the ones with a posting. Not just the ones hiring in your city. Not just the ones that list your major.
The rest of this is three things: what makes an email fail, the format that works, and who to actually send it to.
The anatomy of an email that gets a reply
Every effective version of this email answers the same three questions in the same order, and then asks for one specific thing. That is the whole structure. It is not clever and it does not need to be.
A cold email from a student to a founder, annotated. The order matters more than the wording: identity, reason, evidence, ask.
The whole email above is 88 words. That is not a stunt. It is close to the right length.
Five things that kill a cold email
What you leave out matters more than what you put in. The list below is short on purpose, and following it removes most of the reasons a founder will not reply to you.
1. Length
Stay under about 200 words. Nobody wants to read an unsolicited career history from a stranger. The limit is doing more work than it looks: at 200 words you physically cannot include the three paragraphs about your personal journey, so the rule enforces itself.
2. Buzzwords
There are layers to this. "Executed on cross-functional initiatives" becomes "drove impact", which becomes "cut onboarding time from 9 minutes to 2". Keep going until the sentence contains a fact. Read the email out loud before you send it. If it does not sound like you talking, rewrite it.
3. No ask
An email with no specific next step puts the work of figuring out what to say back onto the reader, and the easiest response to that is no response. Ask for the call. Ask the question. Pick one thing.
4. Vagueness
Abstraction is where replies go to die. "I have worked on several projects" is invisible. "I built a Chrome extension that 400 people use" is a fact someone can react to. If you cannot attach a number, attach a name.
5. Fake personalization and volume
Do not open with a compliment you do not mean. Do not claim you have followed them for years. It is completely fine that you found them because they raised a Series A last week, and saying so is more credible than the alternative. Follow up once, maybe twice, then move on.
The upper bar is the budget, split by the four parts. The lower bar is roughly what a student sends when nobody has told them there is a budget.
The budget bar is a design target, not a measurement. The point is the gap: the parts that matter fit in half the ceiling, which is why the ceiling almost never binds.
Two examples you can steal the shape of
These are ours, written for students. They are deliberately unremarkable. Nothing here required a flash of inspiration, which is the encouraging part.
Hi Dev, I am Priya, a second year data science student at UT Austin.
I read your post about pulling Shopify data into Snowflake and ended up building something close to it for a campus nonprofit last term. It moves about 2M rows a night.
I am looking for a summer internship and would like it to be at a company where I would own a pipeline rather than watch one. Happy to send the repo.
Are you taking interns this summer?
Priya
78 words. The ask is a yes or no question. The evidence is one project with one number. Notice that she does not know whether they have an internship program, and asks instead of assuming.
Hi Sam, I am Jordan, a junior studying HCI at Georgia Tech.
I signed up for Fernpost last week and got stuck on the workspace step, so I redesigned that flow as a practice exercise. Three screens, nothing fancy: link.
I would like to intern on product design this summer. Previously I did a semester at a two person startup where I owned the marketing site end to end.
Worth a 15 minute call?
Jordan
76 words, and the strongest kind of opening: they did the work before asking. This is more effort than most students will spend, which is exactly why it lands.
Who to email, and how to find them
Notice that neither example references a job posting. That is deliberate. You do not have to wait for a company to publish a role before you write to them, and for small teams the posting frequently trails the intent to hire by months.
- Build a list of companies you would actually be glad to work at. Startup newsletters, your school's alumni list, portfolio pages of funds you like, and the companies behind tools you already use are all fine sources.
- Read enough about each one to have a real opinion. You need one specific true sentence about them, and you cannot fake it.
- Find the person. At a company under about 30 people, email the founder or CEO. Above that, find the engineering, design or data lead you would report to. LinkedIn is usually enough.
- Work out the address. Most small companies use firstname@company.com. If that bounces, try firstname.lastname or the initial plus last name. Tools like Hunter, ContactOut and Clay exist for this.
- Send it, then check it against the five rules above before you hit send, not after.
- Follow up once, about a week later, in three sentences at the top of the same thread.
You need to send more than one
Here is the part that is easy to skip and expensive to skip. Writing a good email removes one reason someone will not reply to you. It does not remove the others, and the others are mostly outside your control: hiring is not their priority this month, they saw it and forgot, they were on a plane, it went to spam, they simply did not feel like it.
Thirty companies, thirty emails. Most produce nothing, and that is the expected outcome rather than a sign you did it wrong.
Illustrative, not measured. We are not claiming a 10 percent reply rate. The claim is narrower and more useful: the outcome of one email is noise, and the outcome of thirty is a signal about your list and your writing.
So do it thirty times. Not thirty in one evening with the company name swapped out, which is the version that fails. Thirty over a few weeks, each one aimed at a company you read about first. That is a handful of hours of work, spread out, against an outcome that changes what you do next summer.
Run this list before you send
- Under 200 words
- Says who you are in the first line
- Contains one true, specific thing about their company
- Contains at least one number about your own work
- Has exactly one ask, phrased as a question
- Contains no sentence you would be embarrassed to say out loud
- Goes to a named person, not to careers@
Where this goes wrong
Two honest caveats. The first is that cold email works much better when you are already competent at something and can point at proof. If you have not built, written, shipped, organized or sold anything yet, the highest leverage move is not a better email. It is spending a month making something small and real that you can put in the third paragraph.
The second is that this is a probabilistic tool, not a spell. A great email to the right person at the wrong moment still gets nothing back. That is not a reason to skip it. It is a reason not to read a single non-reply as a verdict.
Start with five companies this week. Write five emails, send them, and see what the reading and the drafting actually cost you. Then decide whether you want to do the other twenty five.
Source
This guide adapts The guide to getting a job with cold email by Ben Lang, published on the Next Play newsletter on August 3, 2025. The three part format, the list of mistakes and the argument for volume are his. The internship framing, the example emails and the figures on this page are ours. His original is worth reading in full, and Next Play is a good place to find the kind of small company that answers cold email.