Analyzing Paid Search Ad and Keyword Effectiveness

The Challenge

For many large retail businesses, the last decade of digital transformation has created a tremendous sea of data, but organizations continue to struggle to turn raw data into actionable insight and measurable business value. One of our clients, a multi-billion-dollar national consumer packaged goods (CPG) company, struggled to quantitatively define the effectiveness of their digital marketing efforts. In other words, they lacked the ability to track digital marketing performance and thus struggled to optimize return on investment (ROI) on hundreds of millions of dollars of annual paid advertising dollars. Elder Research’s team–made up of data strategists, data scientists, and data engineers–were tasked with analyzing the current distribution of paid product ads to determine the relative value of various Google paid search keywords. The result would give our CPG client the ability to compare digital spend in an “apples to apples” fashion and optimize spend based on ad performance metrics.

The Solution

Elder Research’s team utilized our proven Agile Data Science process to develop baseline  KPIs, including start and completion metrics for video ads, as well as generate an easy-to-understand “Video Exit Score” that allowed our client to evaluate digital marketing spend and ROI on different market segments throughout a video advertisement. To determine keyword value, our team analyzed data from the client’s Google Analytics API and AdWords reports. Two keyword pricing models were built to define a market value for keywords based on observed cost for similar website activity and assigned dollar values to particular online interactions based on input from industry experts.


Our team identified under-performing ad characteristics, resulting in greater return on investment. For example, our analyses and modeling were able to identify an under-performing ad characteristic that corresponded to a 55% decrease in click-through-rate (CTR).

Video Exit Score Distribution by Density

Additionally, our keyword pricing models uncovered overspending in 25% of our clients’ paid search budget and discovered that four of the top ten clicked keywords were overpriced. To quickly visualize the most and least effective keywords thus making the models more easily accessible and action-oriented, our team designed a dashboard that gave brand teams instant access to real-time data to optimize their paid search budget. As a result, our client now has the capability to more rapidly access data on ad performance and take optimization action upon that ad performance, resulting in greater performance on less ad spend.

Dollars Spent on Keywords