Introduction
Sending Time Optimization (STO) is an artificial intelligence tool used in customer relationship management (CRM) systems to determine the optimal time to send messages to contacts. By analyzing historical data and customer behavior patterns, STO aims to maximize communication effectiveness by sending messages at the most opportune times.
This feature currently applies only to automated and tactical email campaigns.
How It Works
The STO algorithm is based on the collection and analysis of a large amount of data related to customer interactions, including the timing of previous message sends and click-through rates.
Using machine learning and statistical modeling, STO predicts the likelihood of engagement for each customer at different time intervals. This allows the algorithm to recommend the most appropriate day and time to send a message to each recipient.
When the algorithm does not have enough data on an individual (a new customer or no interaction in the last 2 years), the timing will be determined by a generalized model based on overall past behavior.
To ensure deliverability, a safety threshold is applied. When the volume of messages sent for a single campaign exceeds 150,000 per hour, they are automatically spread out over time at a rate of 150,000 messages per hour. This regulation may result in messages being sent outside the time windows defined by the STO.
Each month, the STO trains on more up-to-date data to continuously improve its predictions. For all contacts, the algorithm automatically divides the entire set of individuals in the database into two distinct groups:
Test Group (approximately 70% of contacts): Receives emails at the optimal times calculated by the AI.
Control Group (approximately 30% of contacts) : Receives emails randomly distributed throughout the time window defined during configuration
This control group is essential for isolating and measuring the actual impact of the STO on performance.
Recommendations
To ensure the results are relevant, we recommend that, when setting up the STO, you configure a sufficiently wide time range so that the algorithm has a choice of several times of the day.
It is also essential to have historical data with substantial volumes to allow the algorithm to learn and accurately distinguish between good and bad sending time slots.
Since the STO algorithm selects the best time slot across the entire configured sending period, campaign scheduling must be finalized no later than the day before. This ensures that each contact is targeted at their ideal time, including profiles optimized at the start of the time window.
Conclusion
The STO algorithm improves customer engagement and campaign response rates by leveraging artificial intelligence.
STO can be enabled on the campaign setup page, as explained in the following articles.