Predictive Scoring
At its core, predictive scoring leverages advanced analytics and machine learning algorithms to assign a numerical score to individual customers or leads, indicating the likelihood of them performing a specific action. Unlike traditional lead scoring, which often relies on predefined rules and demographic data, predictive scoring dynamically analyses vast datasets—including historical interactions, behavioural patterns, demographic information, and even external market signals—to generate highly accurate probabilities.
Benefits of predictive scoring:
Prioritised Engagement: Not all customers are created equal. Predictive scoring identifies the individuals most likely to convert, churn, engage with a specific campaign, or make a high-value purchase. This allows marketing teams to focus their resources and personalised efforts on the most promising segments, significantly improving Return on Investment (ROI).
Enhanced Sales Alignment: Predictive scores provide sales teams with actionable insights, highlighting "hot" leads that are ready for immediate engagement. This leads to more efficient follow-ups, higher conversion rates, and a more harmonious relationship between marketing and sales departments, both working towards common revenue goals.
Proactive Churn Prevention: Identifying customers at risk of churning before they disengage is a critical application of predictive scoring. By recognising early warning signs, marketers can deploy targeted retention strategies, special offers, or personalised outreach to mitigate potential losses and preserve valuable customer relationships.
Optimised Campaign Performance: Predictive models can forecast the success of different campaign elements, helping marketers optimise channels, messaging, and timing. This iterative optimisation leads to continuous improvement in campaign effectiveness and better allocation of marketing spend
When you open predictive scoring in lemnisk, the dashboard will present you the following details.
Model Name: Shows the name of the existing prediction models
Created: Shows the model creator and the timeline of creation.
Last Updated: Shows the last update details of the prediction model.
Status: Shows the status of your prediction model.
Action: Allows you to perform different actions related to the model such as publishing, stop training, and deactivation.
Steps to create predictive scoring
Navigate to > Ramanujan AI > Predictions
Click + Create New Predictive Model
Provide the below mentioned model details:
Model Name: Give a unique name (Max - 50 characters) for your predictive model.
Start Event: Choose start events from the drop-down which is the starting point from which the prediction begins. There can be multiple start events. The scoring begins only after user performs this start event.
Goal Event: Select one or more goal events from the dropdown. Users will be continuously scored for their likelihood of performing a goal event, starting from the defined start event until the goal event is performed by the user.
Trigger Events: Select the events which when performed by the user will trigger an update in the prediction score.
Filters: You can use filters by clicking this icon
to add properties and filter events as required.
Additional Details:
Profile Attributes: Select a profile attribute from your account’s defined customer properties where the predicted score will be stored. This attribute will then be available for segmentation and other downstream use cases.
Email Recipient: Select the email recipients to which updates related to model training, or any other status change should be sent.
Click Save.
When configuring two models, you cannot save them if they have an identical set of lead, update, and target event configurations. To save both models, ensure that the combination of events and their associated property values is unique for each model.
While the models can share the same event types, their property values must differ to allow for saving.
Steps to view prediction analytics
Analytics can be viewed only for the published prediction model and the published models which is no longer in use.`
Navigate to > Ramanujan AI > Predictive scoring
Select the prediction model of which you want to view analytics.
Navigate to Analytics tab which shows the following insights.
Model Performance Score: Shows the overall performance score of your prediction model. This helps to know the accuracy of your prediction model in defining the conversion possibility for a user.
Predictive score vs Goal conversion rate: Shows the comparison between the score predicted by your prediction model and the actual goal completion by users.
Users Metrics
Calendar: You can choose the time frame in the calendar for which you need to see the user metrics.
Users Scored: Shows the count of the total users scored by the prediction model.
Users Converted: Shows the count of total users converted out of users who were scored.
User Conversion %: Shows the conversion rate in percentage.
Last updated