Consumers today expect their interactions with various brands to be personalized, contextually relevant and responsive to their behaviors. As marketers, you must anticipate customer interests and intentions and decide on the next-best action. Blueshift’s customizable predictive scores utilize the power of AI to help create 1:1 experiences and real-time interactions that adapt to each customer based on their data and behaviors. This helps you to optimize who to target, with what content, on which channel, and at what time throughout the customer lifecycle.
Predictive scores help you to automate audience selection by identifying high-value customers for each of your marketing goals, as in those most likely to purchase, churn, engage with content, respond to an offer, renew a subscription, upgrade, and so on. Using AI and machine learning models, Blueshift’s predictive scores calculate each customer's likelihood to perform a business goal or action. Each customer gets a score from 0 to 100, with higher probabilities translating to a higher chance of performing the desired goal.
Predictive scores are continuously updated with the latest customer data and behaviors so that you can be confident that you are always reaching the right audience.
With the Predictive studio, you have full visibility into which customer attributes and behaviors impacted the model the most and can track how the performance (i.e. conversion rate) of the predictive scores cohorts changes over time.
You can export predictive scores to any 3rd party application, paid media channel, or back to your internal systems for deeper analysis.
Note
Predictive scores are an advanced Blueshift capability and are included in certain Blueshift packages. Contact your Blueshift CSM to start using AI powered predictive scores in your campaigns.
Advantages of using predictive scores
With Blueshift’s predictive scores, you can automatically identify the most valuable customers by using always-on predictive models that determine the right customers to target based on their likelihood to perform desired actions. Since scores are always up-to-date, it eliminates the need for manual segment building and maintenance. Using predictive scores, you can improve marketing performance through greater targeting accuracy across campaigns and strategies.
- Customizable inputs and goals - You can select what events, customer attributes and goals go into the models.
- Automated feature engineering - Blueshift's fully automated feature engineering takes away the grunt work of cleaning, de-duping and normalizing raw data to compute a rich set of features that feed into learning models. Blueshift also automatically removes noise features that do not contribute to the accuracy of the model.
- Automated inference of training data - Blueshift's automated inference of training data from past customer behaviors takes away the manual task of curating and labeling training data to feed the learning models. We do this by advanced sampling techniques over past customer behaviors to select sufficient positive and negative examples for the required outcome.
- Automated model building and validation - Blueshift's automated model building and validation uses state of the art learning models to build an ensemble of models that generalize well without overfitting.
- Easy to use outputs - The output score for each customer is converted into a percentile rank between 0 and 100. You can pick subsets of the range that best suit your campaign needs.
- Rich visualization - The model visualization, showing how the goal completion rates vary by each percentile, is built into the dashboard along with the relative importance of each feature.
AI algorithms used by Blueshift
Blueshift uses ensembles of boosted trees (GBMs) and logistic regression (LR) algorithms to train classifiers for model building and uses auto-encoders, PCA, Matrix Factorization for feature representation and selection.
Use Cases
Marketing campaigns aimed at driving specific KPIs often involve targeting customers who are likely to perform a specific action. For example, customers who are likely to upgrade to a paid subscription, purchase a product, submit a form, or download specific content.
Using predictive scores, you can create precise segments that target customers that are likely to perform a business goal or action and then send very targeted and personalized messages to nudge them towards specific parts of your site or apps that are expected to yield the highest follow through or conversion.
For example, you can create a segment of customers who have an 80%+ likelihood to purchase.
Predictive scores are most often used for the following purposes:
- Identify the most valuable customer
- Improve marketing performance
- Reduce churn
- Optimize conversion
- Activate customers
- Subscription upgrades
- Cross-sell and upsell
- Driving repeat engagement
- Paid media targeting/ improve ROAS
- Increasing deliverability
- Increasing offer redemption
Here are a few sample predictive scores for your reference. You can use the Predictive studio to create these based on the individual account data and required business outcomes.
| Predictive score type | Description | Use cases |
| Conversion | Conversion predictive scores model the likelihood of each individual customer to reach a specific state in a funnel. |
Travel, e-commerce, 2-sided marketplaces sites usually have a series of steps leading to a booking/checkout and purchase funnel. Finance, real estate and job boards typically have a series of steps leading to submitting a lead/form and applying for a job. Media, education and entertainment sites typically have a free-mium model for customers to move from free to a paid subscription model. |
| Engagement | Engagement predictive scores model the likelihood of each individual customer to come back to the site/apps within a time window, typically tracked as part of the DAUs/MAUs (Daily Active Users/Monthly Active Users) | Customers repeatedly coming back to the site/apps/store is a strong indicator of customer engagement with a brand. Modeling and predicting that behavior allows you to take proactive steps to improve engagement and reduce inactive customer base. |
| Churn | Churn predictive scores model the likelihood of each individual customer to cancel or downgrade a subscription. | Subscription based ecommerce models. |
| Retention | Retention predictive scores model the likelihood of each individual customer to repeat a specific action/goal. While this correlates highly with engagement, it's more precise. It aims to compute repeat purchase or transaction likelihood towards a specific goal event, rather than just a repeat visit. | Repeat goal actions within a time window are strong indicators of strong retention and high LTV (Lifetime Value). |
Data required for predictive score calculation
The following data must be available in Blueshift for predictive score calculation.
| Data type | Description | Example |
| User Attributes | User attributes such as location, device type, category or brand or catalog attribute preferences, and so on. |
favorite author |
| Events | Aggregates on past behaviors. |
lifetime_orders |
| Goals | Custom goals. |
submitted lead |
While setting up a predictive score, it is important to ensure that the historical conversion rate (number of conversions in prediction window/size of the modeled audience * 100), is at least 1% and at least 1,000+ customers have converted in the past prediction window. The AI model learns to make predictions based on historical conversions. Too few conversions or very low conversion rates may lead to low performance or stats resulting in an insignificant model. Once data requirements are met, an AI model is created.
Comments
0 comments