Customised data collection app and machine learning stack
Theoretically grounded marketing knowledge
Innovative research design
An agile and robust research framework
After 3 years and over A$3.5million dollars spent on development, it has been applied to 44 separate groups of respondents, totaling more than 12,000 viewer sessions, recording over 100 full days of viewing, 70,000 test ads, in 3 countries, across 4 screen types and across 9 different media platforms.
Research design that uses the best of humans and the best of machines. We are continuously improving by retraining our models, considering relationships between variables and testing new platforms.
Considering both the relative importance of differentiating characteristics of competing media and creative content, capturing their role in driving advertising effectiveness.
The CAI Intelligent Research System has changed the way big brands brief their creative and spend their advertising dollars.
We conduct cross-platform, in-platform and creative testing across a range of media platforms and devices.
The number of variables we can test for is growing quickly. Here’s what’s on our current list:
Based on the variables being tested, experienced PhD researchers construct the most appropriate research questions with the client, designing the necessary test and control groups, creating appropriate exposure treatments, identifying the best way to measure variables and remove avoidable bias.
Our custom-built Data Collection App (android and IOS) does many things depending on the brief. It exposes the right viewer to the right viewing material, activates user-facing camera to passively collect facial footage, initiates a test advertisement tag, collects additional viewing environment variables (such as viewability metrics), intercepts ad load to replace with test ad, and redirects participant to the virtual store to make their shopping choices.
All data collected is pushed to the Data Analytics Framework, which incorporates a customised machine learning model to transpose it to a second-by-second measure of attention.
The framework combines the data at the individual respondent and ad view level. Output metrics are then delivered to researchers in CSV format.
Researchers analyse the outputs to determine answers to specific client questions. Some of the methods used for measuring advertising impact and attention include, Short Term Advertising Strength (STAS), partworth utilities and multivariate regression.
Results are written up, presented and delivered to the client, working closely on the most useful client application.