Message frequency optimisation

Message frequency optimization is the practice of finding the send cadence that produces the most engagement per recipient while keeping unsubscribes and complaints low. Instead of asking how many messages you can send, it asks how many you should send before each extra message stops helping. If two sends a week lift total clicks but four sends a week lift unsubscribes more than they lift clicks, the optimal point sits somewhere between the two.

Frequency matters because attention is finite and easily exhausted. The same message that lands well at one send a week can feel like noise at five. Recipients do not file a complaint the first time they feel over-messaged. They quietly stop opening, then mute, then unsubscribe, and by the time the unsubscribe lands the engagement was already gone. Optimising frequency is about reading the early signal, not the final one.

The goal is not the lowest frequency or the highest, but the cadence where the marginal send still pays for itself. That point differs by audience, channel, and content type, so frequency optimization is a continuous calibration rather than a one-off setting. A cadence that worked last quarter can decay as the list ages or the content mix shifts.

Group recipients by how many messages they received in a window, then compare engagement and fatigue across those groups. If recipients who got one send a week reply at four per cent and lose one in 500 to unsubscribe, while those who got three sends a week reply at six per cent but lose one in 120, you can weigh the extra replies against the extra churn and find where the trade turns negative.

The core decision is how you value an unsubscribe against an engagement. A lost subscriber forfeits every future send, so a single unsubscribe is worth far more than a single open. Assign a value to a retained recipient and a cost to a lost one, then the optimal frequency is simply the cadence with the highest net value per recipient.

Use a holdout or a staggered rollout rather than changing everyone at once. If you raise frequency for the whole list in the same week a new product launches, you cannot tell whether engagement moved because of cadence or content. A controlled split keeps the cause clean.

1. 1

   Define the recipient window

   Pick a window such as seven or 30 days and count sends per individual recipient inside it, not sends per campaign.

2. 2

   Bucket recipients by send count

   Group people into cohorts by how many messages they received, for example one, two, three, and four or more sends.

3. 3

   Measure engagement and fatigue per cohort

   For each cohort, record engagement per recipient and fatigue signals such as unsubscribes, complaints, and mutes.

4. 4

   Find the net-value peak

   Value each retained engaged recipient and subtract the cost of each loss, then pick the cadence where net value per recipient is highest.

Optimal frequency is the balance of three forces: how much engagement each send earns, how much fatigue each send costs, and how relevant the content is at that cadence. Decomposing it shows whether a falling response is a frequency problem or a relevance one, which are fixed in opposite directions.

Optimal cadence varies sharply by channel because tolerance for messages differs. Email tolerates more sends than push, and push tolerates more than SMS. The ranges below give a realistic starting band per channel for a permission-based audience, to calibrate before testing rather than treating any single number as fixed.

The biggest gains usually come from sending the right cadence to the right person rather than one cadence to everyone. A single list-wide frequency over-messages the casual reader and under-messages the engaged one. Move toward per-segment cadence and let recipients signal their own preference.

1. 1

   Optimising for opens alone

   Opens can rise while unsubscribes rise faster. Judging cadence on engagement without weighing fatigue picks a frequency that quietly burns the list.

2. 2

   Counting sends per campaign

   The load that matters is on the individual inbox. Three campaigns where each person sees one is a very different frequency from three campaigns to the same person.

3. 3

   Changing cadence and content together

   If you raise frequency in the same week you change the offer, you cannot tell which moved engagement. Isolate cadence with a holdout to keep the cause clean.

4. 4

   Treating the optimum as permanent

   List composition and content mix drift over time. A cadence that was optimal last quarter decays, so frequency needs periodic recalibration, not a one-off setting.