Machine Learning Times
Machine Learning Times
AI Success Depends On How You Choose This One Number
 Originally published in Forbes, March 25, 2024. To do...
Elon Musk Predicts Artificial General Intelligence In 2 Years. Here’s Why That’s Hype
 Originally published in Forbes, April 10, 2024 When OpenAI’s...
Survey: Machine Learning Projects Still Routinely Fail to Deploy
 Originally published in KDnuggets. Eric Siegel highlights the chronic...
Three Best Practices for Unilever’s Global Analytics Initiatives
    This article from Morgan Vawter, Global Vice...

10 years ago
The Right Data Can Sweeten Returns on Ad Investments


All the data in the world won’t increase the return on your advertising expenses by itself, but some clever ideas from a San Francisco analytics conference might.

In back-to-back presentations, attendees at Predictive Analytics World heard how to cut costs and increase output in today’s lightning fast online ad networks and how to gauge return on investment in a much older medium: television.

Building a Model

Mahesh Kumar of Tiger Analytics provided the insights on advertising networks that — within a half-second — help a publisher recognize an online user, contact an advertising exchange, relay the request to an ad network, choose among participating advertisers, select an appropriate ad and deliver that advertisement to the user.

In part because of the sheer difficulty, the click-thru rate on such ads are often as low as 1:3000, he said.

To remedy that, he suggested counting all the successful clicks and then comparing that to the unsuccessful attempts to create a model that will yield better results. However, he noted the computation time needed to crunch that much data would take too long, even with a supercomputer. Instead, he said you can use a small sample of the unsuccessful attempts, reducing the computing time by a factor of 100.

In a case study he shared, this helped reduce the cost of ad purchases by 20 percent while yielding a 56 percent lift in click-thrus. In his second act, Kumar explained how Tiger transforms the unstructured data of social networks into semi-structured data, and finally into structured data that can score social media content from Facebook, Twitter, blogs and DataSift. For example, a simple mention of “travel” in a tweet can be assigned a number which, when combined with other clues, may suggest the social media user is planning a trip.

The result, he said, is as much as a 25 percent reduction in cost per acquisition thanks to higher click-thru rates.

Measuring ROI for TV Ads

Thumbnail image for datarow1.jpg

Just before Kumar’s presentation, Daniel Kissin, the senior analytics manager at Expedia — one of the world’s largest travel companies — offered some ideas on how to measure ROI for TV ads. In today’s omnichannel world, where the results of other media can be measured in a flash, it  can be difficult to understand the impact of TV ads.

Expedia has quite a bit of experience in television. And Kissin said one key is to create connections between the TV ad and another action by urging listeners to call in the next 10 minutes,  respond to a particular website or use a coupon code. However it is done, the links can provide information comparable to click-thrus on the web by showing a particular ad drew a specific number of responses.

He also said it’s vital to inspect traffic in tiny increments of time, looking for sudden spikes at the times that commercials air. He acknowledged that with most programming, it is very difficult to discern a direct connection. However, he said with special events like the Super Bowl or the Oscars, a spike can be spotted easily. Expedia found that out in February 2013 when it advertised on the Super Bowl, he said.

Title image by Tiger Analytics.

By: Tom Murphy, General Manager & Editorial Director, cmswire
Originally published at cmswire

One thought on “The Right Data Can Sweeten Returns on Ad Investments

  1. Pingback: The Right Data Can Sweeten Returns on Ad Investments | Big Enterprise Data

Leave a Reply