Frequency questions usually arrive framed as a number: is twice a week too much? The number is not the useful part. What matters is the relationship between what you send, what people expected, and what happens to your complaint rate.
Consistency beats cadence
The most reliable finding is that predictability outperforms frequency in either direction.
A monthly email that arrives every month is recognised, sometimes anticipated, and rarely reported as spam. The same total volume delivered as silence for eight weeks then five emails in ten days performs worse on every metric — recipients have forgotten who you are, the burst feels like a change of relationship, and mailbox providers see a volume spike from a domain that had gone quiet.
If you can only commit to monthly, send monthly. Aspiring to weekly and achieving it half the time is worse than either.
Set the expectation at signup, then honour it
Most complaints about frequency are really complaints about surprise. Someone who signed up for a quarterly product update and receives three emails a week did not get more value — they got something they did not agree to.
Stating the frequency at the point of signup does two things. It slightly reduces signups, which feels like a cost and is actually a filter. And it converts frequency from something people react to into something they consented to.
Watch the right metrics
Open rate is the instinctive metric and the least reliable one here. It moves for many reasons — privacy features inflating it, deliverability problems suppressing it — most unrelated to frequency.
The two that matter:
- Unsubscribe rate. Rising in step with increased sending is the clearest fatigue signal available. This is people leaving politely.
- Complaint rate. Rising means you are past politeness. Given the 0.3% ceiling under the bulk sender rules and the recommendation to stay under 0.1%, this metric has very little headroom to absorb an experiment.
A useful way to hold both in mind: an unsubscribe costs you one contact, a complaint costs you a fraction of your ability to reach everyone else.
Testing frequency without damaging the list
If you want to increase frequency, do it on a segment rather than the whole list.
- Take your most engaged segment — people who opened or clicked recently. They are the most tolerant of more contact and the least likely to complain.
- Increase frequency for that segment only, and hold everyone else steady as a control.
- Run it for at least four to six weeks. Fatigue is cumulative and does not appear in the first send.
- Compare unsubscribe and complaint rates between the two groups, not open rates.
- Roll out only if the difference is negligible.
The reason to test on the engaged segment is not that it flatters the result. It is that if your best contacts will not tolerate the increase, nobody will.
Different frequencies for different segments
Uniform frequency across a list treats a daily opener and someone who last engaged eight months ago identically, which serves neither. A reasonable structure:
- Highly engaged — your normal frequency, or slightly more.
- Cooling — reduce. They are drifting, and more mail accelerates it.
- Dormant — one re-engagement attempt, then removal under your sunset policy.
This is the same segmentation that protects deliverability generally, applied to cadence rather than inclusion.
The uncomfortable question
Teams asking how often they can send are usually asking how much they can extract from a list before it degrades. The more productive question is whether each message is worth sending at all.
A list that receives something genuinely useful every week will tolerate weekly. A list receiving filler will not tolerate monthly. Frequency is downstream of whether the mail is worth opening, and no cadence fixes content nobody wanted.