Sep 28, 2022
Schaun Wheeler

How many different timing preferences could users really have?

Sep 28, 2022
Schaun Wheeler

How many different timing preferences could users really have?

Sep 28, 2022
Schaun Wheeler

How many different timing preferences could users really have?

Sep 28, 2022
Schaun Wheeler

How many different timing preferences could users really have?

We wrote recently about the importance of getting the timing of a message right. This short post uses data from one of our customers' actual messages to show what we mean by that. 

They say a picture is worth a thousand words. This one is worth hundreds of thousands of active users:

To conduct this analysis, we mined the history of our “personalization scores”. 

Personalization scores are how Aampe records user preferences, so for every possible change you could make to a message—whether sending on a different day, or at a different time, or with different content—every user in the system gets a score that estimates that specific user’s probability of responding positively to that choice. 

The chart above shows the top 40 patterns of “strong” preference (users who scored in the top third of the personalization score index for one or more message timing slots), and the bigger the dot, the more users showed that preference. 

  • The very top row was a relatively large number of users who only like to get messages on Tuesdays—but not late night on Tuesday

  • The next row is a big group of users who only want messages on Sundays. 

  • The row after that is a group that wants messages on Tuesdays, but not in the evening—not even early evening. 

  • The next row after that is a group of users who are good with pretty much any time we want to send them something.

You can see there are a bunch of groups who are good with almost all times, but who really don’t respond during a few key windows, and there’s another group that’s only ok with Monday mornings, and another that’s only ok with lunchtime on Wednesdays.

Also note the note at the bottom of the plot: There are 482 other timing preference patterns that we didn’t try to show, because it was too much to fit into one picture. 

Those 482 other preference groups, however, represent a full third of the user base.

In other words, there are hundreds, sometimes thousands, of preference patterns you need to take into account if you want to send each user a message at a time when they’re likely to act on it (and if you want to not send a user a message at a time that annoys them). 

You just can’t manage all of these different timing patterns by hand. 

Thankfully, you don’t have to.

Email hello@aampe.com, and give Aampe a test drive today.

0

Related

Shaping the future of marketing with Aampe through innovation, data.

May 29, 2025

Schaun Wheeler

Discover how Aampe's semantic-associative agents adapt to varying user contexts, from fully known to unknown users, ensuring personalized experiences through continuous learning and contextual imputation.

May 29, 2025

Schaun Wheeler

Discover how Aampe's semantic-associative agents adapt to varying user contexts, from fully known to unknown users, ensuring personalized experiences through continuous learning and contextual imputation.

May 29, 2025

Schaun Wheeler

Discover how Aampe's semantic-associative agents adapt to varying user contexts, from fully known to unknown users, ensuring personalized experiences through continuous learning and contextual imputation.

May 29, 2025

Schaun Wheeler

Discover how Aampe's semantic-associative agents adapt to varying user contexts, from fully known to unknown users, ensuring personalized experiences through continuous learning and contextual imputation.

May 28, 2025

Schaun Wheeler

Explore how Aampe's semantic-associative agents differ from traditional multi-armed bandit models. Learn about their multi-dimensional action space and non-ergodic learning approach that tailors user experiences without generalization.

May 28, 2025

Schaun Wheeler

Explore how Aampe's semantic-associative agents differ from traditional multi-armed bandit models. Learn about their multi-dimensional action space and non-ergodic learning approach that tailors user experiences without generalization.

May 28, 2025

Schaun Wheeler

Explore how Aampe's semantic-associative agents differ from traditional multi-armed bandit models. Learn about their multi-dimensional action space and non-ergodic learning approach that tailors user experiences without generalization.

May 28, 2025

Schaun Wheeler

Explore how Aampe's semantic-associative agents differ from traditional multi-armed bandit models. Learn about their multi-dimensional action space and non-ergodic learning approach that tailors user experiences without generalization.

May 26, 2025

Schaun Wheeler

Traditional customer engagement tools often constrain strategies by bundling orchestration and analysis within fixed campaign structures. Adopting an agentic approach—separating orchestration from analysis—enables more dynamic, user-centered communication, allowing for nuanced decision-making and broader impact.

May 26, 2025

Schaun Wheeler

Traditional customer engagement tools often constrain strategies by bundling orchestration and analysis within fixed campaign structures. Adopting an agentic approach—separating orchestration from analysis—enables more dynamic, user-centered communication, allowing for nuanced decision-making and broader impact.

May 26, 2025

Schaun Wheeler

Traditional customer engagement tools often constrain strategies by bundling orchestration and analysis within fixed campaign structures. Adopting an agentic approach—separating orchestration from analysis—enables more dynamic, user-centered communication, allowing for nuanced decision-making and broader impact.

May 26, 2025

Schaun Wheeler

Traditional customer engagement tools often constrain strategies by bundling orchestration and analysis within fixed campaign structures. Adopting an agentic approach—separating orchestration from analysis—enables more dynamic, user-centered communication, allowing for nuanced decision-making and broader impact.

May 22, 2025

Schaun Wheeler

Agentic systems are designed to operate within a manageable state space, focusing on relevant variables to make effective decisions. This approach contrasts with traditional AI models that attempt to process vast amounts of data, often leading to inefficiencies and suboptimal performance.

May 22, 2025

Schaun Wheeler

Agentic systems are designed to operate within a manageable state space, focusing on relevant variables to make effective decisions. This approach contrasts with traditional AI models that attempt to process vast amounts of data, often leading to inefficiencies and suboptimal performance.

May 22, 2025

Schaun Wheeler

Agentic systems are designed to operate within a manageable state space, focusing on relevant variables to make effective decisions. This approach contrasts with traditional AI models that attempt to process vast amounts of data, often leading to inefficiencies and suboptimal performance.

May 22, 2025

Schaun Wheeler

Agentic systems are designed to operate within a manageable state space, focusing on relevant variables to make effective decisions. This approach contrasts with traditional AI models that attempt to process vast amounts of data, often leading to inefficiencies and suboptimal performance.

Load More

Load More

Load More

Load More