For many years, mailers have been building data models using outside lists for direct mail acquisition. With the advancement of technology, these tried and true practices have moved online — using big data in a big way. Marketers are now building data models off of their current online customers to identify prospective audiences across the entire US Internet population (and abroad), and serving banner ads to them as they browse the web.
Steps to Build an Online Model
- Choose a partner: There are many different modelers on the market. Each has different reach across the web; sources, types, quality and quantity of data; and build, refresh and optimization practices. You'll want to work with a partner who uses 1st party online data and refreshes and optimizes their model in real time.
- Tag your website: The modeler will provide you with a tag (a 1x1 invisible pixel) to place on key upper and lower funnel pages of your website to analyze the data of visitors to your website and also those who convert. The goal is to obtain a statistically significant number of online buyers to collect data on and use as a basis to build a model off of. Each modeler has different thresholds for this. At RMI, we like to see a minimum of 800-1,000 conversions.
- Build the model: Once the minimum number of conversions has been reached, the model will be built to uncover prospects online that look just like your current online customers. For most modelers this process will take a couple days and is based on hundreds to thousands of data points.
- Submit your creative: For a real-time-bidding (RTB) campaign such as this, you will need to supply your banner creative in various standard ad size formats. This is so you always have a banner to fit the ad space up for bid, on the page your prospect is viewing. If you don't have the appropriate size you will miss an impression opportunity. Various file types can be used as well such as JPG, PNG, animated GIF, Video, and Flash.
- Go live, measure and optimize! Once the campaign goes live, your modeler should supply you with regular reports showing metrics such as the number of impressions, clicks, and conversions by type (click-through or view-through conversion). The campaign should also be optimized daily (if not in real time) to constantly refresh the target pool of prospects based on new online data.