Back

    "The Probability of Purchase" and How It Helps Find Users with Real Interest

    It’s no secret that you can try to predict users’ actions in e-comm by analysing their behaviour on the Internet. And the more we know, the more accurate the prediction may be, and thus the higher is the conversion into the necessary target action.

    Certainly, knowledge and data alone are not enough — it’s important to correctly interpret these data, identify patterns and relationships, and also to have the required production capacities to store and process large amounts of information in real time.

    To solve these tasks, we use various ML algorithms and unique self-learning models developed by Flocktory Big Data engineers. Today we want to tell you more about one of these models.

    What is probability of buying?

    This is a unique self-learning model that extremely accurate predicts the probability that a user will buy a product on a given site within the next 7 days.

    With the help of ML algorithms, the model analyses information about how many times and for how long the user visited the site, the history of the user’s interaction with the cart, their views of product categories, and many other factors.

    Based on this analysis, site visitors are divided into three groups: hot, medium and cold.

    What are these users?

    • A hot group user can place an order without additional motivation.
    • The chance of getting an order from a medium group user will increase if you offer a discount.
    • For a cold group, a discount or a gift with a purchase would potentially work best.
    • What does this information give us?

    Knowing who needs or doesn’t need motivation and what kind of motivation helps us to effectively customise communication with future buyers, and (!) save profit by not offering a discount where the probability of buying is high without it.

    The probability of buying model helps our partners save money by reducing the number of promo codes issued for discounts, while maintaining the conversion to order at the same level as before the predictive analytics were introduced into the communication.

    Where and how can we use the probability of buying model?

    Probability of buying can be effectively used in the following trigger scenarios: abandoned shopping cart, product abandoned view, category abandoned view, abandoned session with no useful action, site visit.

    We help our partners create trigger communication chains.

    We adjust them to the online store’s work features and the motivational methods used.

    And, of course, we offer hypotheses for testing to increase the communication’s effectiveness at each funnel stage. As a rule, recommendations for testing concern not only the technical side, but also the marketing component. We provide end-to-end scenario testing solutions with careful attention to all aspects of communication, including channel combination possibilities for omnichannel user experience strategies.

    If you have not yet connected Flocktory, we will be glad to meet you to discuss how we can solve your problems — contact us at sales@floctory.ru or via our TG channel.

    That’s all for today, thank you for your attention!