Industry
Contents
Business and AI, with a European lens
Marketing & media

How to write a cold email to a scientist

A scientific marketing email guide based on a cold email experiment with 2,247 researchers.

Prefer this source
Illustration of a laboratory pipette releasing a single drop with an envelope inside it towards a dish.
Illustration, Industry Contents
TL;DR and context

A nine-step scientific marketing email process, based on our experiment in which problem-led emails beat statistics-led ones on open rate by 10.8 points.

  • In a randomised experiment involving 2,247 life sciences professionals, 49.6% of recipients opened first emails that led with a lab problem, compared with 38.8% for emails that led with validation statistics.
  • This guide turns that result into a working process for scientific marketing teams, from collecting customer language and scoring a researcher list to subject lines, follow-ups, measurement and a cleaner test.
  • Anything that goes beyond what the experiment measured is marked as practice.

Read the study behind this guide, What makes someone open a cold email?

Scientific marketing runs on proof. Product pages carry validation data, application notes and citations, and the reflex is to lead a cold email with the strongest figure available. A researcher scanning her inbox sees the subject line before she sees the evidence inside. In the experiment, emails that led with a question about a familiar lab problem had an open rate about 11 percentage points higher than those that led with a number. The gap held across all four campaign pairings in the US and EU. The statistic may still earn its place later, once a buyer is comparing suppliers, though the experiment did not test that.

This guide is for life sciences email marketing teams, growth teams and salespeople selling research tools, reagents, instruments, services or software to laboratories. About two-thirds of the researchers in the study worked at universities, 28% worked in industry and 5% worked in government labs. Those selling to clinicians, regulators or procurement departments should read the limits section before copying anything.

Sort your last ten cold email campaigns into two groups. Put subject lines that led with a number or validation term in one group and those that led with the reader’s situation in the other. Compare their open rates. Lists and timing will differ, so treat what you see as a hypothesis to test.

Then check the size of the list. To detect an open-rate gap of the size seen in the experiment, aim for about 660 contacts per segment, split into two groups of about 330.

Gap to detectRecipients per versionTotal for a two-way test
Open rate, 38.8% to 49.6%about 330about 660
Click rate, 1.4% to 3.0%about 1,300about 2,600
Standard two-proportion sample sizes at 80% power and 5% significance, using the rates measured in the experiment.

Industry averages are a rough guide. Woodpecker puts average cold email open rates between 27.7% and 44%, and Instantly puts the average reply rate at 3.43%, with top-quartile campaigns at 5.5%. Both are self-reported platform figures. The study list was filtered to recently engaged contacts. Compare your two versions with each other, since industry averages cover different lists.

Step 1. Collect the customer’s language

Use the words customers use to describe the problem. Pull the notes from your last 20 sales calls, your last 20 support tickets and any discovery interviews. Highlight every sentence in which a customer describes something going wrong. In a spreadsheet, record each phrase, how often it recurs and the speaker’s role. Keep the ten most frequent phrases and drop any that only your team uses.

Teams without calls or tickets can mine forums, review sites, protocol troubleshooting sections and the discussion sections of papers, where frustrations surface as workarounds. Practice, not study data.

Step 2. Build and score the list

You can use the study’s list-building process with a smaller list. Build three sets of keywords covering the core technology, the problem and the outcome. Rank a researcher who has published on both the technology and the problem above one who has published only on the technology. Require a verified current affiliation, relevant research interests and some sign of email engagement, such as an opened professional message in the past six months. Strip duplicates, invalid addresses and past unsubscribers. Contact data tools can fill gaps, and our comparison of Apollo and Hunter sets out what each costs.

FactorWeight
Publication 2023 to 20251.0
Publication 2021 to 20220.6
Publication 2020 or earlier0.3
First or corresponding author1.0
Middle author0.4
Senior titleHighest weight
Journal impactModest weight
Contact scoring used in the experiment. Within each weekly batch, higher-priority contacts received their emails first.

The experiment used separate segments for academia, pharma and biotech, and contract research. It added a regulatory and alternative methods segment for one product. For each segment, describe the problem using the phrases you collected in Step 1.

Step 3. Write problem-led subject lines

Name what goes wrong, where it goes wrong and what it costs.

Number-ledProblem-led
Validated assay with a 95% success rate across runsIs assay variability slowing down your screening?
99% purity, lot-to-lot consistencyDoes lot-to-lot variation keep forcing you to repeat runs?
Cuts sample prep time by 40%How many hours of sample prep did you lose this week?
Peer-reviewed protocol, 12 citationsStill troubleshooting a protocol that worked last month?
The first row shows a lightly edited version of the pair tested in the experiment. Rows two to four illustrate the pattern and were not tested.

Each line should describe something the reader has lived through, carry at most one number and make sense in under two seconds. In the experiment, subject lines were 8 to 15 words long, with the main idea in the first 40 characters. A phone inbox may show only 30 to 40 characters.

Step 4. Write the email

The problem-led emails in the experiment were 150 to 300 words long and followed the order below.

LinesJobIn the experiment
1One sentence on the recipient’s recent work, specific enough that it could go to nobody elseA reference to a recent publication, identical in both versions
2 to 3The problem, in two or three short sentences of customer language“Inconsistent results. Readings that vary from run to run.”
4 to 5The cause, in terms the reader recognises and you can substantiateWhy standard set-ups produce the problem
6 to 7The fix, as the answer to that cause, with one concrete outcome and no spec sheetThe product as the answer to the cause
LastOne call to action“Want to see the protocol and hear how other groups solved this?”
Structure of the problem-led email in the experiment.

Both versions in the experiment linked to the same validation material, so the proof was always one click away. Send from a named person, in plain text with light formatting.

Step 5. Prepare the send

Check delivery before reading anything into the response. For the experiment, we warmed the sending domains for four weeks, starting at about 50 emails a day, and authenticated them with SPF, DKIM and DMARC. We sent on weekdays during recipients’ local working hours, avoided academic holidays and big conferences, and put sender identification and a working unsubscribe link in every message. We emailed US and EU contacts in separate segments on different platforms, so compare gaps within a region, never raw rates across regions.

European rules on business cold email differ by country.

MarketBusiness cold email in practice
EU baselineGDPR legitimate interest (Article 6(1)(f)) can cover the data processing, but national e-privacy law decides whether the email itself needs consent
FrancePermitted without prior consent when the message relates to the recipient’s profession, with an easy opt-out
GermanyPrior consent generally required for advertising email, even between businesses
United StatesCAN-SPAM rules on honest headers, a postal address and a working opt-out
Summary only, not legal advice. Source, Fieldfisher, EU e-marketing requirements, January 2024.

Step 6. Follow up

In the experiment, non-responders got one follow-up three to five days later, in the same style but on a different symptom of the problem or a different data point. The sequence below builds on it.

SendTimingContent
1Day 0Problem-led email from Step 4
2Day 3 to 5Same style, a different symptom of the same problem, shorter
3Day 8 to 10A short example of how another lab handled it, one link
4After the reader asks for validation materialThe detailed validation material the reader asked for
Send 1 and the timing of send 2 follow the experiment. The rest is practice.

The fourth send only goes to readers who ask for the validation material, so the proof arrives when they want it.

Step 7. Measure beyond the open

Raw opens are now a weak signal. Apple Mail Privacy Protection loads tracking pixels whether or not anyone reads the message, and Apple clients accounted for 49.29% of email opens in January 2025, according to Litmus. Security scanners add more phantom opens and clicks. The experiment excluded opens generated by Apple’s privacy feature and by scanners. We identified scanner opens by near-instant timing, repeated patterns from one IP address and known scanner user agents.

Count unique opens and clicks over a fixed window. Divide each count by the number of delivered emails. The experiment used a 72-hour window for opens. Then add replies, meetings and qualified leads, since the experiment stopped at the click. The Wine Access email test shows how to judge copy on gross profit instead of clicks. Log which version each contact received so you can follow them down the funnel.

Step 8. Separate the subject line from the body

The experiment changed the subject line, opener and structure together, so it cannot isolate the effect of the subject line. A randomised two-by-two test can estimate the effects of the subject line and body separately.

Subject and bodyProblem-led bodyStatistics-led body
Problem-led subjectAB
Statistics-led subjectCD
Randomly assign a quarter of the segment to each of the four versions.

Compare A and C with B and D to isolate the body, and A and B with C and D to isolate the subject. Hold the sender, first line, offer, length and send time constant, and randomise within each segment. The experiment used a pseudo-random generator seeded by contact ID, which makes the split reproducible. Four versions need twice the list of two.

Step 9. Model the response

If the study’s rates held across 10,000 delivered emails, the two versions would produce the results below.

Per 10,000 deliveredProblem-ledStatistics-ledDifference
Opensabout 4,960about 3,880about 1,080 more
Clicksabout 300about 140about 160 more
Illustration only. The experiment’s list was filtered to recently engaged contacts, so most lists will see lower absolute rates.

Multiply the extra clicks by your click-to-meeting and meeting-to-sale rates to estimate additional sales. Multiply that estimate by your average gross profit per sale, then subtract any additional campaign costs.

The charts show the overall result and the split between the two product lines. One product was for regulated safety testing and the other addressed a common bench-method problem.

Bar chart of cold email open rates. Problem-led first emails were opened by 49.6% of 1,124 recipients. Statistics-led first emails were opened by 38.8% of 1,123 recipients. The difference was 10.8 percentage points (p < 0.001).
Open rate by first email style, 2,247 researchers in the US and EU. Data from our 2025 experiment.
Bar chart of open rates by product. Product A, problem-led 49.6% against statistics-led 38.8%. Product B, problem-led 49.6% against statistics-led 38.9%.
The problem-led advantage held for both product lines. Data from our 2025 experiment.

Limits of the framework

The experiment measured opens and clicks. Whether problem-led emails won more customers is unknown. The recipients were life sciences researchers. Clinical, regulatory and purchasing audiences may respond differently. Run a small version of the test on your list before moving a whole programme to problem-first framing, since the open signal arrives within 72 hours.

Pre-send checklist

  1. Does the subject line describe the reader’s situation, with at most one number?
  2. Does the framing land within the first 40 characters?
  3. Could the first line only have been written for this person?
  4. Are the problem phrases taken from real customer language?
  5. Is there one call to action, with validation material one click away?
  6. Is the message between 150 and 300 words?
  7. Is the sender a named person, with a working unsubscribe link?
  8. Does the target market allow business cold email without consent?
  9. Is the version logged against the contact?
  10. Is the sample large enough for the question being asked?

Related analysis

The companion study gives the full findings, limits and European breakdown. For another measured follow-up experiment, see Fischer Future Heat texted week-old leads with AI and booked 105 home surveys, and for a randomised test of retention messaging, The policy video that kept more customers after a price rise.


FAQ

What is a problem-led scientific marketing email?

One that opens with a frustration the researcher recognises, explains its cause, offers the product as the fix and gives one next step, with validation data a click away. In the experiment, these emails were 150 to 300 words long.

How many contacts do I need to test subject lines?

About 330 per version to detect an open-rate gap of 10.8 percentage points and about 1,300 per version to detect a change in click rate from 1.4% to 3.0%, at 80% power and 5% significance.

How should I follow up with researchers?

In the experiment, we sent one follow-up three to five days later in the same style, focusing on a different symptom of the problem. The third and fourth emails in this guide are suggested practice.

Is cold email to researchers legal in Europe?

It depends on the country. France permits business prospecting by email when it relates to the recipient’s profession, while Germany generally requires prior consent even between businesses. Check each market before sending.

Does the guide prove problem-led emails win sales?

No. The experiment measured opens and clicks, not meetings or sales.

Spotted an error? Write to corrections@industrycontents.com.

Industry Contents reporting draws on direct interviews, company statements and public records. Figures appear as published by the sources named. Nothing here constitutes legal or investment advice.

Written by

Yajush Gupta is the Founding Editor of Industry Contents, an independent publication reporting on business models, technology deployment and operating economics, with a European focus. He previously worked as a financial and business correspondent at Reuters, covering e-commerce, financial markets and international business. He holds an MSc in Data Science and Marketing and works across business journalism, growth analytics and search.

Keep reading

Get TalkHead

Every Friday, TalkHead checks what your preferred AI tools cost, how businesses test their claims and what buyers should check before committing.

Unsubscribe anytime. Your email stays confidential.