Skip to content
Outreach guide

How to reach a data scientist

Reach a data scientist with a short, plain message: why them, one to three direct questions, and the exact time you need. Keep the ask small.

15 minutes of their time
$14.45

At the U.S. median for data scientists.

Median pay
$120,230 a year
Per hour
$57.80
Middle half earns
$85,660 to $158,880
Fair price for a reply
$20

BLS OEWS, May 2025. Full pay breakdown

What lands in a data scientist's inbox

A data scientist's inbox is mostly other people's urgency. Recruiter mail about roles they never applied for. Vendor notes about a platform they already use. Students asking for a project review. A stranger asking for "a quick call" with no agenda. The U.S. Bureau of Labor Statistics counted 262,440 data scientists in May 2025, with a median annual pay of $120,230 — which tells you the person reading your message has a calendar that other people price all day.

That price is close to literal. At the median, fifteen minutes of a data scientist's time is worth about $14.45, and an hour is $57.80. They know it. Their manager knows it. Your request has to be small enough to fit between two meetings. The messages that survive the preview pane do three things at once: they show you know what this particular person works on, they ask for a fixed slice of time, and they make replying cheaper than ignoring you.

What dies on arrival is anything that could be sent to a thousand people unchanged. A message that opens with "I hope this message finds you well" and ends with "pick your brain" tells the reader you did no work. The fix isn't cleverness. It's specificity and brevity, in that order.

Before you write

Name the ask before you name the person, because the ask decides who to write to. Requests to data scientists come in four shapes: a quick question (about five minutes), a considered reply (about fifteen), a review or short call (about thirty), an hour of advice (about sixty). At median pay those cost the recipient roughly $4.82, $14.45, $28.90 and $57.80 of working time. Write your number down before you open your email client. If you can't say which shape you're in, you're not ready to send.

Then pick one person. Read the last thing they published, shipped or said on a panel: a conference talk, a write-up, an open-source repo, a job move. You want one sentence that could only be written about them. "I saw you work in data" is not that sentence. "Your write-up on why the retention model kept flagging new accounts was the clearest explanation of leakage I've read" is.

Check the alternatives first. If a colleague of theirs could answer, ask the colleague. If the question is about team structure or tooling budgets, an analytics manager may be a better target than the person doing the modeling. And know the money context. Published medians run $185,080 in San Jose, $170,110 in the San Francisco Bay Area, $164,740 in Seattle, $135,980 in New York and $132,200 in Washington, DC. Someone at that level gets pitched constantly, so your message has to be smaller and sharper than the last ten.

What to say in the first message

Keep the body between 50 and 125 words. That range isn't a style preference: Boomerang's analysis of more than 40 million emails found messages of that length got the most replies. The same analysis found one to three direct questions did better than none, and plain, simple writing did better than formal, college-level prose. Write the way you'd write to a colleague on another team.

The order that works: why them, the ask, the questions, the close. One line of context that proves you read their work. One line saying what you want and how long it takes. Then one to three questions that can each be answered in a sentence. Then a specific time. "Could you reply by Thursday?" beats "at your convenience."

Here's the difference in practice. Weak: "Would love to pick your brain about our data strategy before we hire." No shape, no end, no reason it's them. Stronger: "You moved a recommendations model from batch to streaming last year. I'm weighing the same change on a much smaller team. Two questions, about ten minutes of typing."

Subject lines carry the whole message in a preview pane, so keep them under 70 characters and make them concrete: "Question about your churn labeling" or "Streaming recs: one question from a small team." Never "Quick question" on its own. And if your draft holds three asks, delete two. A reader who answers one question is worth more than a reader who closes the tab.

How to follow up

One follow-up, about a week later, in the same thread. Say what changed since your first message: a new detail, a result you got, a narrower question. Do not resend the same text with "just bumping this" on top, and do not add a second channel because the first one went quiet.

Keep the follow-up shorter than the original. It isn't a second pitch; it's a smaller door. If the first message asked for a call, the follow-up can ask for one line by email. If the first asked two questions, ask one. "One more try, then I'll leave it" is a fair and honest opener, and people respond to it because it respects their time.

Then stop. Two messages is the limit for a stranger. If there's no reply, you've learned something useful: the ask was too big, the timing was wrong, or the person wasn't the right target. Don't send a guilt-flavored close-the-loop note. Send nothing.

If the answer matters enough to pay for, say so plainly in the follow-up rather than hinting at it. "If a paid message is easier, I can send one through Introworth" is a clean sentence. It turns an unpaid favor into a transaction with a price and a boundary, which is often easier to accept than a favor.

When a paid message makes sense

A cold message competes with paid work, so paying removes one common reason to ignore you. On Introworth, members set a price for messages from people they don't know. Your card is held, not charged, and the payment is captured only if the recipient replies within the reply window. If they don't reply, the hold is released. The member keeps the bulk of the price and Introworth takes a commission.

Prices start at $15, and the suggested levels track the value of the time you're asking for: about $15 for a quick question, $20 for a considered reply, $40 for a review or a short call, $75 for an hour of advice. Set those beside the $4.82, $14.45, $28.90 and $57.80 the time is worth at median pay, and the numbers read as fair rather than generous. An hour from a data scientist working in San Jose, where the published median is $185,080, is worth far more than $75 — but a bounded request with a number attached is much easier to accept than an open-ended one.

Paid messaging isn't right for every situation. If you share a connection, a warm introduction beats a cold message: on Introworth a member can ask a connection to forward a note, and the connection can forward it with a line of their own or decline. Existing connections message each other with no per-message fee. Paying doesn't buy a yes either. It buys a clear signal, and if there's no reply the hold is released.

Before you send, put the draft in front of a reader, or run it past Introworth's message coach, which checks length, questions, why them, the time asked for and plain wording before a sender pays. Most weak messages fail on one of those five things, and all five are fixable in ten minutes.

Message templates for reaching a data scientist

Each one passes Introworth's message coach: 50 to 125 words, one to three questions, no stock phrases. Replace the brackets with your own specifics.

First message

Subject: Question about your churn model talk (10 minutes)

Hi [Name],

Your talk on churn models at [event] ended on a point I keep running into: the model was fine, the labeling wasn't. I'm rebuilding the churn definition at [your company] and stuck on the same step.

Two questions. How did you decide the window for "churned" — 30 days, 60, or something tied to contract length? And did you validate the label against support tickets before training?

If a written reply is easier than a call, that works. Could you reply by [date]? It's about 10 minutes of your time.

Thanks, [Your name]

Follow-up a week later

Subject: Following up on the churn label question

Hi [Name],

One more try, then I'll leave it. Since my last note I pulled two years of cancellations and found the 30-day window hides a chunk of annual contracts, which is what your talk warned about.

If it's easier, one line on how you set the window would settle it. Would this week work? I'm glad to pay for the time through Introworth if that's the cleaner route.

Thanks, [Your name]

What your ask costs them

AskTheir timeFair price per message
A quick question (5 min)$4.82$15
A considered reply (15 min)$14.45$20
A review or a short call (30 min)$28.90$40
An hour of advice (60 min)$57.80$75

Their time at the U.S. median pay for data scientists. The fair price is what lets them keep the full value of that time after Introworth's 15% commission, rounded up to a familiar price.

Where data scientists earn the most

CityMedian a yearPer hour
San Jose / Silicon Valley$185,080$88.98
San Francisco Bay Area$170,110$81.78
Seattle$164,740$79.20
New York$135,980$65.38
Washington, DC$132,200$63.56

Published BLS medians for data scientists, the highest five of the metro areas Introworth tracks.

Questions

How do you cold message a data scientist?

Short, specific and bounded. Name the one piece of their work that made you write, ask one to three questions, and say how long a reply takes. Keep the body between 50 and 125 words with a concrete subject line under 70 characters.

Is email or LinkedIn better for reaching a data scientist?

Email holds technical detail better; LinkedIn is fine when you only need a sentence or two. Pick one channel and send one message there rather than the same text on both. A paid message is the third route when the question is worth money to you.

How much should I pay a data scientist for fifteen minutes?

At the U.S. median annual pay of $120,230, fifteen minutes is worth about $14.45 of working time. Introworth's suggested price for a considered reply of that length is $20, and the lowest price on the platform is $15.

What should I never say in a cold message?

Skip the stock openers: "hope this message finds you well," "I came across your profile," "pick your brain," "touch base," "quick chat," "circle back." They signal a mass send. Replace them with one specific line about the person's work and one clear ask with a time attached.

Reach people who chose to be reachable

On Introworth, members set a price for messages from people they don't know. Your card is held, not charged, and you pay only if they reply. Members keep 85% of the price.

Guides for specific asks to data scientists

General outreach guides

How to reach other roles

Pay figures come from the U.S. Bureau of Labor Statistics and are computed by Introworth's price calculator. The guide text is drafted with AI from those figures and Introworth's product rules, and checked automatically before it is published. It is general guidance, not legal, tax or financial advice.

Also from the OneLifeVibe network