Send-Time Optimisation
Deliver to each customer at the moment they are most likely to engage.
Braze AI
What it does
Send-Time Optimisation works out the best moment to reach each individual customer on each channel, then holds the message until that window. Rather than one blast time for everyone, Braze uses each person's own history of opens, clicks and app opens to schedule delivery when they are most likely to engage. It answers the 'when' for every campaign, with no manual scheduling, and it sharpens as engagement history builds.
What feeds it
- ·Message and engagement history per customer (opens, clicks, app opens)
- ·A channel with enough delivery volume for a per-person pattern to emerge
- ·New or low-history customers fall back to a sensible default time
How to switch it on in Braze
- 1Build the campaign or Canvas step as usual.
- 2In the scheduling options, choose Intelligent Timing (Send-Time Optimisation) instead of a fixed send time.
- 3Set the delivery window Braze is allowed to schedule within, so nothing lands overnight.
- 4Confirm a fallback time for customers without enough history yet.
- 5Launch, and let Braze stagger delivery across the window per customer.
How Fuse builds for and utilises this
- ·We make sure engagement events flow cleanly per channel, so the model learns from real opens and clicks rather than gaps.
- ·We warm it up: sensible delivery windows and fallback times while per-customer history builds.
- ·We monitor the lift against fixed send times and check the gain holds on conversion, not just opens.
- ·We tune the windows and fallbacks as the engagement picture sharpens, channel by channel.
What to measure
A capability that fires on people who would have converted anyway proves nothing. Hold a slice out, measure only the gap above them, and let the recipe earn its place on evidence.
- ·Open and click rates against a fixed send time
- ·Downstream conversion, not just engagement, so a better open still earns its place
- ·True lift measured with a holdout on /prove, not the headline open rate alone
Pairs with
On its own it is useful. Alongside these models it compounds.