How often to send email without annoying your list

The question is usually asked as though a number exists — weekly, twice a month, every Tuesday — and the reason nobody can supply one is that frequency is not the variable readers respond to. Value per message and expectation are, and cadence only matters relative to them.

A daily email from a business publishing genuine daily deals is fine, because each message contains something new and the subscriber signed up knowing that. A monthly email that has nothing to say is not fine, despite being twelve times less frequent, because the reader opens it, finds nothing, and learns that opening it is not worth doing.

This is why benchmark tables of average sending frequency by industry are close to useless as guidance. They report what businesses do, aggregated across senders with completely different value propositions and audiences, and the variance inside any one row is larger than the differences between rows. Knowing that the average retailer sends some number of emails a month tells you nothing about whether your list wants that many from you.

The largest single lever is the expectation set at signup. A form that says 'a short email every Friday' creates a subscriber who is not surprised on Friday and who notices if it does not arrive. A form that says 'sign up for updates' creates a subscriber with no expectation at all, which means every message has to justify itself from scratch and any increase in volume feels like an escalation.

Consistency compounds that. Arriving predictably builds a habit; arriving in bursts — silence for six weeks, then four emails in nine days because a campaign was planned — feels like more mail than the same total spread evenly, and it is the pattern most small businesses actually produce.

Relevance reduces perceived volume, which is why frequency and segmentation are the same conversation. Four emails a month, each aimed at people it applies to, is experienced as less mail than two untargeted broadcasts, because the reader is not repeatedly receiving things that have nothing to do with them. If frequency feels like the problem, relevance is often the actual one.

Watch rates per send rather than totals, because totals rise with volume and hide the trend. The specific things to track: unsubscribe rate per send, spam complaint rate per send, and whether click rate declines across a sequence of closely spaced messages.

The complaint rate is the one with an external reference point. Google's sender guidelines ask senders to keep the spam rate reported in Postmaster Tools below 0.30%, and the guidance is to stay under 0.10%. If your complaint rate climbs toward that ceiling as you add sends, the list is telling you something unambiguous, and it is the signal with the most severe consequences attached.

A softer signal worth taking seriously: replies asking you to stop, or asking how to unsubscribe. For every person who writes that, several thought it. A handful of those replies is more informative than a small movement in an aggregate rate, because it tells you the reader experienced the volume as a problem rather than merely losing interest.

The opposite failure gets far less attention and is common among small businesses, where the honest cadence is 'whenever somebody remembers'.

The deliverability consequence is not obvious. Mailbox providers build their view of a sender from recent history, so a domain that sends nothing for four months has little current reputation. A sudden campaign to the whole list after that silence has the shape of a compromised account — a dormant domain suddenly emitting bulk mail — and it is filtered accordingly. The mail that is meant to restart the relationship is the mail most likely not to arrive.

The human consequence is worse. A subscriber who signed up five months ago does not remember doing so. When your first email arrives, they read it as unsolicited, and the spam button is a reasonable response to mail from a business you do not recall. Reporting rates on a first send after long silence are predictably poor, and it is not because the email was bad.

The most direct way to resolve frequency is to stop guessing. A preference centre that offers a real choice — weekly or monthly, all topics or one, or a pause for a few months — converts a frequency complaint into a setting instead of an unsubscribe.

A pause option is undervalued. Somebody who is busy, travelling, or between purchase cycles is not rejecting you, and the alternative to pausing is usually leaving permanently. A subscriber who returns after three months at their own request is a better outcome than a clean opt-out and much better than a complaint.

There is a second benefit: what people choose is information you cannot get any other way. If most subscribers who visit the preference centre select the least frequent option available, you have a direct answer to the question this article opened with. If they mostly select a topic rather than a frequency, the problem was relevance rather than volume — and that is a different fix from sending less.

Frequency cannot be A/B tested on a single send, because the thing you are testing is cumulative. The correct shape is a cohort test over a longer period: hold back a comparable group at the current cadence, move another to the new one, and compare after a quarter.

Measure at list level, not campaign level. This is the trap that makes almost every informal frequency test come out in favour of sending more: additional sends reliably increase total clicks and total orders in the short run, while simultaneously raising unsubscribes and depressing engagement in a way that only shows up over months. Campaign-level reporting sees the first effect and is structurally blind to the second.

So track the metrics that capture the cost: unsubscribe rate over the period, complaint rate, active subscribers at the end, and revenue per subscriber rather than revenue per send. And change cadence gradually. A sudden jump in volume is itself a deliverability signal, so ramping over several weeks separates the effect of the frequency from the effect of the change.