Time the conversation: send window, day of month, follow-up rhythm, season
Generic studies point to midweek mornings. Learn the send window, day of month, follow-up gaps and season for each campaign cell from your own replies instead.
Jared Hawkins-Young · Founder, Lost Star TechnologyOctober 8, 20265 min readPart 11 of 14Key takeaways
- Published timing studies disagree: CIENCE's 2021 outbound data put 23% of replies on Wednesday sends, while Hunter's 2026 analysis of 31 million emails found the day of the week did not move reply rates.
- Treat those studies as starting hypotheses and learn timing per campaign cell: time of week, day of month, follow-up gap and season.
- Measure each cadence family on positive replies per enrolled lead and the average touch number before the first positive reply, plus opt-outs and exit reasons.
- Seasonal curves for a head-of-sales persona, such as quiet quarter-end weeks and busy planning windows, start as hypotheses that your own campaign data recalibrates.
There is no single best time to send cold email. There is a best time for each campaign cell, and you learn it from your own replies: use published studies as starting hypotheses, log the send time, day of month, follow-up gap and season on every touch, and let each cell's data replace the generic answer as it accumulates.
What do the cold email timing studies actually say?
The two most cited kinds of evidence point in different directions, and both are worth reading closely before you copy either one.
| Study | Data | Finding | Read it with care |
|---|---|---|---|
| CIENCE, best time to send cold emails | About 6 million rows of CIENCE outbound email data, January to May 2021; 600,000 with known time zones for the hour and day findings | 20% of replied emails were sent at 8 a.m. recipient time; Wednesday sends drew 23% of replies, Monday 16%; 87% of replies came within a day | Shares of replies, not reply rates per send. If more mail goes out on Wednesday, Wednesday collects more replies. |
| Hunter, State of Email Outreach 2026 | 31 million emails sent by Hunter users in 2025 | Sending on a specific day of the week was among the factors that did not move reply rates | Pooled across every use case, from sales to link building. Your cell is narrower than that. |
Neither study is wrong. Each is an average over thousands of senders and buyers who are not yours. A head of sales at a logistics software company and a founder at an HR tech company do not keep the same calendar, so a pooled answer is a reasonable first guess and nothing more.
How do you learn the best send window for each cell?
Log four things on every send: the recipient's local time, the weekday, the day of the month, and the week of the quarter. Then read positive replies per send inside each cell, the same unit part 8 uses for the portfolio.
Start coarse. A grid of six hours by five weekdays has thirty squares, and at a 2% positive-reply rate one square needs a few hundred sends before it says anything. So begin with two parts of the day (morning, afternoon) by three parts of the week (early, middle, late), read those six, and split a bucket only once it has the volume to carry the detail. Show the send count behind every square, and treat a square with too few sends as blank, not as bad.
In this illustrative cell the best window is late morning midweek, not the 8 a.m. that CIENCE's pooled data points to. That difference is the point of learning per cell. The day-of-month grid shows a second pattern worth testing for this persona: the last week of the month is quiet, because that is when a head of sales is closing.
Knowing the window does not mean firing everything at one hour. Google's sender guidelines tell senders to avoid sudden volume spikes and to raise volume slowly, and Hunter's 2026 data found that sending 20 to 49 emails per day per mailbox drew a 5.7% reply rate against the 4.5% average. Spread each day's sends across the cell's top window instead of stacking them.
What follow-up rhythm works for cold email?
Follow-up is worth testing because it clearly matters: Hunter found that three messages instead of one lifted total replies to 6.8% from 3.3%. The gap between them is the open question. CIENCE advises not letting the first follow-up drift past the third day. A slower rhythm that waits two weeks and arrives with something new is the obvious alternative. Test them as cadence families inside a cell, not as fixed assumptions.
| Cell | Cadence family | Positive replies per enrolled lead | Avg touch to first positive reply | Next |
|---|---|---|---|---|
| Logistics · head of sales | Signal → day 3 bump → day 10 new angle → day 24 close-out | 2.2% | 1.6 | Retest |
| Logistics · head of sales | Signal → day 14 value → day 35 close-out → re-entry on a new signal | 2.7% | 1.8 | Promote |
| Fintech · RevOps lead | Signal → day 21 new angle → day 42 close-out → seasonal re-entry | 1.8% | 2.4 | Retest |
| HR tech · founder | Signal → day 14 proof → day 35 close-out | 0.7% | 2.9 | Retire |
Measure each family on four numbers: positive replies per enrolled lead, the average touch number before the first positive reply, the opt-out rate, and the exit reason for every lead that finished without replying. Positive replies per enrolled lead keeps a pushier cadence from winning by producing more replies that say "stop". Re-entry happens only on a new signal, scored the way part 9 describes, never on a timer alone.
Does seasonality matter for B2B cold email?
For a head-of-sales persona, the calendar has an obvious shape to test: quarter-end weeks are spent closing, December combines holidays with year-end, and January and October are when next year's plans and budgets are being set. Those are hypotheses, drawn as the seasonal strip in the figure. Every send logs its month and week of the quarter, and the curve is recalibrated by size band and persona as replies arrive.
The curve has a second use: planning the reps' calendars. If your cells answer most in the October planning window, the meetings land in late October and November, and the reps need open time then. A timing model that only schedules sends and ignores who takes the meetings is half a model.
Timing is also the fourth layer of the diagnostic runbook in part 12. When replies drop in a cell, check whether the moment changed before you rewrite a message that was working.
Next in the series
Part 12, Trace a reply-rate drop from the outside in, is the runbook for the day the numbers fall: seven layers, checked in order, each against a matched control.
Common questions
- What is the best day to send a cold email?
- It depends on the cell. CIENCE's 2021 outbound data found Wednesday sends drew 23% of replies, but that is a share of replies rather than a rate per send, and Hunter's 2026 analysis of 31 million emails found the day of the week did not move reply rates. Log the day on every send and read it per cell.
- How long should I wait before following up on a cold email?
- Test it rather than copying a rule. CIENCE advises not letting the first follow-up drift past the third day; a slower cadence that waits two weeks and brings something new is the natural alternative to test against it. Judge both on positive replies per enrolled lead and opt-outs, not on total replies.
- Does send time matter if my emails are going to spam?
- No. Timing only matters once mail reaches the inbox. If bounces, placement or complaints moved, fix those first; part 12 of this series lays out the order.
Sources
Written by Jared Hawkins-Young from the system he builds and runs. Diagrams are drawn from the same specs as the software. Published October 8, 2026. Illustrative numbers are labelled as such; every other figure carries its source.