The Marketer Who Analyzes the Most Data Fastest Wins

Maximizing campaign ROI and profit from advertising is among every marketer’s top priorities. Because of this, yield optimization – increasing the spread between the cost of advertising and the revenue derived from it – is a critical focus for digital marketers. Fundamental technology breakthroughs in big data platforms now enable new approaches to advertising yield optimization that promise to dramatically enhance profitability for digital marketers who take advantage of them.

Different buyers often value the same advertising inventory differently. Underpriced inventory leaves profit on the table, and overpriced inventory left unsold does the same. Yield optimization techniques focus on finding the perfect price in order to maximize profit by ensuring that each ad unit is sold to the right buyer at the right time for the right price.

From an analytical perspective, yield optimization typically requires the construction of a price/volume curve that plots the expected impression volume available for a range of ad prices, or bids, in a targeted market. Once constructed, this curve helps answer questions like:

• How to bid when entering a market
• How to adjust bids when objectives change
• How to determine the most profitable bid

Here is an example of how one might use a price/volume curve to find the optimal bid for an advertisement in a specified market:

Importantly, the analytical complexity involved in constructing price/volume curves properly is amplified by several factors that bring anything other than modern big data platforms to their knees.

• Complex targeting. Ad inventory is often segmented in multiple, overlapping ways, depending on the characteristics of both the audience and the site. For example, the “18-24 year old” and “20-39 year old” market segments overlap. The ad inventory in these overlapping segments can potentially be sold at higher prices, but managing the total pool of available inventory becomes far more challenging.
• Frequency caps. Advertisers avoid oversaturating their target audience by limiting the number of times a unique viewer can see their advertisements within a particular time period. Supporting this capability introduces unique big data analysis challenges.
• Large amounts of data. Hundreds of terabytes of data may require rapid analysis, and as long as the ability to perform arbitrary price/volume analysis is required, this data cannot be summarized in order to reduce its volume.
• Time pressures. Marketers require real-time “what if” analysis of their campaign strategies. They can’t afford to wait hours, or even days, to make campaign decisions.

Attribution analysis is a necessary building block for yield optimization, as is near-real-time analysis. In addition, yield optimization solutions must be highly automated in order to apply complex analytical calculations to large data sets that are being updated in near real time. In fact, one leading online advertising network that reaches 86 percent of all U.S. Internet users estimates that its ad servers generate up to 10 terabytes of data per day, and they leverage up to 280 terabytes of data in constructing thousands of price/volume curves across multiple campaigns each day for yield optimization. This firm found that porting their existing yield optimization algorithms to a modern big data platform that enabled much quicker analysis of much larger data sets – reducing analytical execution time from over a day to just five minutes – led to dramatic increases in campaign performance (case study available here).

This is why the aforementioned big data technology breakthroughs are so important to marketers. Yield optimization is among a class of big data problems for which increased analytic performance translates directly into increased business value. In other words, when it comes to yield optimization, the marketer who analyzes the most data fastest wins.