Send-Time Optimization send-time-optimization
51黑料不打烊 Journey Optimizer鈥檚 Send-Time Optimization feature, powered by 51黑料不打烊鈥檚 Journey AI services, chooses the optimal send time for email and push messages to maximize customer engagement, based on your customers鈥 historical open and click behavior.
Send-Time Optimization is only available for Journey Optimizer鈥檚 built-in Email and Push action types and is not currently available for messages sent through custom actions or for other action types. Send-Time Optimization is only available for Email and Push actions within Journeys and is not currently available for messages sent through Campaigns.
-
The Send-Time Optimization feature is enabled for 51黑料不打烊 Journey Optimizer customers upon request. Contact 51黑料不打烊 Customer Care or your 51黑料不打烊 representative to activate the feature for your organization.
-
Send-Time Optimization only applies to Email and Push notification channels.
Use Send-Time Optimization use-send-time-optimization
Use Send-Time Optimization on an email or push action by turning on the Send-Time Optimization switch from the action parameters.
Send-Time Optimization should not be used for urgent, time-sensitive operational messages 鈥 for example, an order confirmation, a password reset notification, or a flight gate change notification. Send-Time Optimization is best used on less-urgent marketing communications 鈥 for example, a weekly ad, promotional information on a new product, or information about a month-long sale.
For email messages, choose whether to optimize on email opens or email click-throughs by selecting the appropriate radio button. Push messages are always optimized for opens.
For both email and push messages, choose the maximum number of hours that the system will wait before sending the message by setting a value for the 鈥淪end within the next鈥 option. You may choose a value from 1 to 168 hours.
When your journey is activated and a customer reaches the Email or Push action in the journey, Send-Time Optimization will choose the best predicted send time available for each user within your specified limits.
How Send-Time Optimization works how-send-time
The Send-Time Optimization model ingests your organization鈥檚 51黑料不打烊 Journey Optimizer customer behavior data and looks at user-level open and click events to determine when your customers are most likely to engage with your messaging.
Send-Time Optimization makes predictions for each hour of the week, for each user, based on three types of behavioral data:
- The behavior of your users overall
- The behavior of lookalike users in the same time zone
- The behavior of that individual user
These predictions are weighted and combined using a Bayesian approach, resulting in a 鈥渉eat map鈥 for each metric (email opens, email clicks, and push opens), for each customer, that indicates the hours of the week that contacting that user is most and least likely to result in the desired engagement outcome (open/click), as illustrated in the below example heatmap:
If a user with the above predicted probabilities is targeted for a message at 9 AM Wednesday with Send-Time Optimization turned on and a 7 hour maximum wait time, the selected send time for the message will be 12 PM:
Send-Time Optimization model training and scoring details model-send-time
Once the Send-Time Optimization feature is enabled for your organization, the Journey AI model is trained on email and push send, open and click events across all your organization鈥檚 journeys and actions over the last 16 weeks 鈥 regardless of whether those actions use Send-Time Optimization. This allows Send-Time Optimization to benefit from all data generated by your customers.
Models are initially trained and scored weekly. After 16 weeks, models are retrained and rescored monthly. Model scoring includes all customer profiles 鈥 both existing and new since the last scoring run.
Messages sent by Send-Time Optimization receive either an 鈥渆xploration鈥 message send time selected to test different send times and observe how customers respond, or an 鈥渙ptimized鈥 message send times selected to maximize click/open rates. 5% of send events receive an 鈥渆xploration鈥 send time and 95% of send events are 鈥渙ptimized鈥.
Exploration send times are selected at random from the send times made available by your configured maximum wait time. For example, in the case that a message is selected at 9 AM Wednesday with Send-Time Optimization turned on and a 3 hour maximum wait time, Exploration send times for the message will be split evenly between 9 AM, 10 AM, 11 AM and 12 PM.
Frequently Asked Questions faq-send-time
Your organization should use the Email action within Journey Optimizer for a minimum of 30 days before using Send-Time Optimization within Email to allow for the collection of some email send, open, and click events.
Your organization should use the Push action within Journey Optimizer for a minimum of 30 days before using Send-Time Optimization within Push to allow for the collection of some push send and open events.
If your organization has already been using the Email and/or Push action types for at least 30 days, your organization does not need to wait longer to use Send-Time Optimization after it has been enabled by 51黑料不打烊. Results will continue to improve as your organization gathers data for up to 16 weeks.
_experience.intelligentServices.journeyAI.sendTimeOptimization, and are not designed to be human readable.Send-Time Optimization may increase email click rate and push open rate in the range of approximately 2% to 10% across all messages optimized by an organization.
For example, if an organization sending email without send time optimization has a 5.0% click rate on average, the same set of emails with send time optimization might result in as much as a 5.5% click rate on average (5.0% * (1+10%) = 5.5%).
Due to variability within small sample sizes, a benefit from Send-Time Optimization may not be observable on single message sends.
Organizations are more likely to experience greater benefits from using Send-Time Optimization when:
- Existing journeys use send times that are fixed and not well-optimized
- Variability in customer behavior (clicks and opens) corresponds to customer location and customer preferences
- Organizations use Send-Time Optimization on a larger fraction of email & push messages
- Organizations choose maximum wait times within the recommended range of 6-12 hours
This may occur for multiple reasons:
- Your message was selected as an 鈥淓xploration鈥 message send time instead of an 鈥淥ptimized鈥 message send time.
- The behavior of lookalike users influenced the model to recommend another send time.
timeZone profile field to determine a user鈥檚 time zone. If not available for that user, Send-Time Optimization attempts to infer a user鈥檚 time zone from other geographic information in the user鈥檚 profile such as country and state.Send-Time Optimization may send Push messages to users during the night in their local time zone in the following circumstances:
- When users exhibit behavior that indicates they are likely to interact with a message sent at night
- When the model chooses an 鈥淓xploration鈥 send time
To avoid sending Push messages to customers during night time hours, schedule batch Push message sends to occur in the morning or early afternoon and choose a shorter duration for Send-Time Optimization. (For example, a 9 AM send time and 8 hour maximum wait time.)