Predictive Intelligent Targeting (PIT) Overview

PIT is a scoring service that assesses website visitors using the PIT model


    Predictive Intelligent Targeting (PIT), also known as “Visitor Scoring” is a Tier 2 chat and messaging system that’s complementary to the Predictive Dialer (PD) system. While PD is responsible for deciding what percentage of a brand’s traffic should be invited to chat/message, PIT is responsible for deciding which specific visitors to invite, based on visitor score. The visitor score is calculated in real-time by applying a data science model to visitor session information* collected by the real-time monitoring system (Shark).

    PIT maximizes the goals set by the brands for their campaigns by inviting the consumers most likely to achieve them using scoring models. Features such as visitor information, location, user history on the web, and browser information are used to drive decisions.

    How does PIT work?

    • Determine a visitor’s likelihood to convert on a website according to that visitor’s behavior. 
    • Mathematical behavior models to predict the probability of a conversion. The probability is captured by the visitor’s “score.”
    • PIT uses two components:
        1. Data – current and historical data located in Hadoop for present and past sessions.
        2. Test & validator set – a set of configuration files which produce different models.  Final model is a relative average of the models which produced a positive result.

    PIT Definitions

    • Score – A visitor’s score is a number that can fluctuate between 0 and 1. The higher the score is, the more likely is the visitor to convert with chat assistance. PIT will provide a score for every InPage traffic event (entering page, 10 seconds while browsing etc.)
    • Score floor – The threshold for a score to be considered as a desirable chat candidate. The score floor can be set manually (manual score floor) or left up to PIT to be determined and adapted throughout the session (dynamic score floor). The dynamic score floor is calculated every 2 seconds. Since the score floor is affected by agent availability, the Predictive Dialer is an essential part in the calculation process.  It is a best practice of PIT to use a dynamic score floor to improve agent utilization.
    • Model – The mathematical algorithm used to determine the score. In essence it is a formula of different conditions. The PIT model is a cluster of 3 working together to create a unified score based on their own individual score:
        1. A model to determine the probability of the visitor converting without a chat (self service).
        2. A model to determine the probability of the visitor converting with chat assistance.
        3. A model to determine the probability of the visitor to accept a chat invitation.

    Missing Something?

    Check out our Developer Center for more in-depth documentation. Please share your documentation feedback with us using the feedback button. We'd be happy to hear from you.