Optimal send time prediction

Optimal send time (OST) prediction calculates the best hour to reach each customer. The time is based on their past campaign engagement, so every message lands when they're most likely to interact.

Packaging

If you're unsure which packages you have, contact your CSM.

How OST prediction works

OST optimizes for clicks by default. The previous "optimize for open vs. click" selector has been removed because open event data can be unreliable. When click data is insufficient, OST uses opens as a fallback if applicable for the selected channel.

The prediction relies on historical campaign event data. There’s no minimum number of campaigns required. A single click at 8 AM sets 8 AM as that customer’s predicted send time, and accuracy improves as more data comes in.

OST prediction workflow

1. Go to Analyses > Predictions, click + New prediction, and select Optimal send time.

2. Select the channel to optimize send time for. Select from the following options:

   • Email
   • SMS / MMS / RCS
   • All channels: Used to set universal send time.
   • Other: Used for custom channels, for which you specify the event and filter. For example, any campaign_open where channel = WhatsApp.

3. Set the default send time. This serves as a fallback for customers without a calculated OST. 0 represents midnight. OST is always calculated in UTC-0. The timezone defined in user settings doesn't affect it.

   

4. If the prediction is part of an initiative, mark it as a global object to display it in the customer profile. Click the Initiative icon next to the prediction name and check the Global object box.

   

5. Click Save, then Start to launch the calculation.

The calculation typically takes only a few seconds. The Results tab won't be populated for this prediction type. To review results, check the OST value on any customer's profile, or create a report to view OST across your audience.

Use the OST property

The OST attribute can be used across:

• Segmentations: Filter customers by OST to build time-of-day audiences. For example, evening-only email campaigns.
• Scenarios: Use the attribute with Jinja in a Wait node to hold each customer until their OST.
• Reports: Compare click rates across send time buckets to validate whether timing improves engagement.

Limitations

The predefined channel options only capture default campaign events, not webhook-based channels with custom event tracking. For those channels, create a separate prediction, select Other, and specify the event and filter. For example, use any campaign_open where channel = sms.

Related articles

Optimal send time in scenarios and campaigns: Learn how to use OST predictions in scenarios and campaigns.

Prediction use cases: Explore real-world examples of how predictions can personalize customer experiences and drive engagement.