Big Village’s Audience Intelligence platform allows marketers to successfully reach high potential targets in the right place, at the right time through its fully integrated and wholistic planning and activation platform. By providing deep insights on media behaviors, values, and attitudinal trends across a wide range of traditional and emerging media, researchers and advertisers can plan and activate more efficiently, maximizing marketing ROI.
Proprietary segments from customer databases or custom research can easily be onboarded to unlock insights and reach segments specific to your business needs. So how does it work?
Audience Intelligence is built on over 5,000 segments that are curated directly from highly nuanced online signals. To complement these online signals, Big Village provides deep media insights on each segment through a nationally representative survey built directly on top of these audiences. Unlocking these insights and integrating custom audiences into Audience Intelligence is done through predictive modeling, machine learning as well as deep learning algorithms. By building a cross walk from an ad-hoc survey that identifies and validates the audience of interest and the passive digital data appended to the survey completes, Big Village is able to create rich attitudinal and media personas for each audience of interest.
Integration & Modeling Process
Custom segments ranging from proprietary brand personas to new product intenders are identified and validated through a custom survey. A digital data append complements the primary data collection and begins to create a digital profile of each respondent that is used in the modeling process to classify respondents into each audience of interest.
The following steps are used to create custom segments in the audience intelligence platform.
1. Segment Identification
Through collaboration with the client an audience of interest is identified. This can range from concept intenders, respondents with certain attitudes or consumption habits or proprietary clusters or segments from an ad hoc or existing segmentation. Once the audience of interest is identified a custom survey is designed to identify the audience and collect the online, digital signals used in modeling. The online “fingerprint” and signals is critical to the modeling, Audience Intelligence and Activation process.
2. Collecting Data Signals
As mentioned above the digital signals are essential to modeling and down the line activation. This data is collected passively by a 3rd party after each respondent consents to Big Village dropping a tracking pixel on their browser. This pixel allows Big Village to append online, digital behaviors from a given device with the respondent’s survey responses. Thus, enriching the profiling and insights on each audience.
3. Variable Selection
With thousands of variables appended and available for the modeling process we must select the most discriminating and predictive variables for the modeling process. Statistical methods such as chi-square analysis, mutual information and advanced feature engineering and simulations are applied to select the best features for the deep learning or machine learning models. These methods allow us to figure out which features explain specific dependent variable and which ones are independent from the dependent variable.
4. Modeling Process
Once variable selection is completed, Big Village runs state of the art and well-known models such as Logistic Regression, SVM, XgBoost, AdaBoost, Random Forest, Deep Learning with MCMC and Bayesian optimization to predict audience classification. These models (or sub-set of the models) are then reviewed and optimized to ensure robustness and the optimum precision and scale.
Optimization methods include advanced simulations, genetic algorithms, Bayesian simulations, graphs and hyperparameter search to improve model efficacy which includes model evaluation metrices such as precision, recall, f1-score and scale. We then perform face validity checks on demographic distributions and other key performance indicators. Any discrepancy is handled with model optimization and threshold adjustments.
5. Score Audience Intelligence
We then leverage the final models to score the Audience Intelligence platform to further analyze and create/enrich audience personas through the vast inventory of segments and data in Audience Intelligence. We gain insights into the segments’ media behaviors, interests, values and many other data points that enrich the overall profile of each segment and guide media planning and activation.
Audience enrichment and strategic insights for media and campaign planning can drive effective marketing and ensure your audience is activated in a manner to maximize ROI. With Big Village Audience Intelligence, you can not only learn more about who your audience is, what their attitudes are and what media platforms they are consuming but you can leverage these insights in your activation on digital platform with your agency of record. Contact us to get started.
Written by Mike Miller, VP, and Nirupam Sharma, Research Director, Insights at Big Village