How Prediction Learning Curves Can Improve Your Ad Effectiveness

In his fascinating new book “The Signal and the Noise,” New York Times political blogger Nate Silver discusses the concept of the “Prediction Learning Curve.”

The prediction learning curve maps the relationship between effort and prediction accuracy. Not surprisingly, the relationship strongly resembles the well-known Pareto effect—e.g. about 20% of the initial effort (time, money, effort) yields about 80% of the prediction accuracy.

How Prediction Learning Curves can Improve Your Digital Advertising

Or, as Silver says:

“…getting a few basic things right can go a long way….The first 20% often begins with having the right data, the right technology, and the right incentives.”

What does this have to do with digital advertising? Optimizing your digital advertising in-flight is, essentially, a prediction exercise. The goal is to take existing data, using the right technology and incentives, and make changes to your plans based on a predicted outcome that is better than current. And, as with the prediction learning curve, the first 20% of effort can have a huge impact on digital ad effectiveness.

Right Data, Right Technology, Right Incentives

Data – Real-time on-line ad effectiveness data is here and now. Brands can now measure brand recall, likeability, persuasion, etc. on a daily basis across display and on-line video, even for smaller campaigns.

Technology – Today’s  ad effectiveness technology holds out a small amount of ad inventory, and then serves up single-question polls after the ad (exposed), or to a statistically comparable group not exposed to the ad. This test/control design enables advertisers to understand the single variable impact of individual creative unit performance, site performance and frequency of exposure.

Incentives – Collaborative optimization tools enable agencies and digital publishers to work together to deliver a better result for their advertiser clients. Agencies measure ad performance by digital publisher. Digital publishers know that either ads perform well on their sites or their sites get dropped from the campaign. They are motivated to work with the agency to improve performance—or else.

So What’s the 20% to Improve my Digital Advertising?

Four basic factors can deliver a dramatic improvement in your digital ad effectiveness—with only ~20% of the effort:

1.  Creative Rotation – Most advertisers don’t copy test their digital advertising. They blindly run multiple creative units without any real understanding of the differences in ad effectiveness across creative units. The first opportunity is to identify your bottom performing 20% of creative, and reallocate this media weight to the top performing 80% of creative.

2.  Site Rotation – Similarly, ad performance differs across web sites. So, the second opportunity is to quickly assess ad performance by web site, identify the bottom performing 20% of sites, and rotate out of these sites and into your higher performing sites.

3.  Exposure Frequency – Unlike TV, digital advertisers have the ability to cap their exposure frequency—e.g. 2, 3, 4, etc. exposures per consumer. The question for advertisers is this: at what frequency should I cap? The opportunity is to quickly identify where frequency of exposure yields little or no incremental ad performance, and then cap your digital ad exposure at this frequency.

4.  Collaborative Optimization – The last opportunity is to get key players in the advertising eco-system to collaborate toward a common objective of improving advertising performance. Everyone has a motivation to do so.

Advertisers can now evaluate digital ad performance across agencies. Agencies have a powerful motivation to improve performance, as advertisers now have performance specific metrics across agencies. Agencies need to perform—or risk losing business.

Publishers also have good reason to collaborate with agencies, as they know that agencies will pull advertising from their sites if they underperform.

Importantly, there are now technology platforms that enable all three parties to collaborate in real time to optimize ad performance.

Moving Up the Prediction Learning Curve

These four factors represent the 20% effort in the Digital Advertising “prediction learning curve.” Let me say it differently: if you do these and only these four things, you can expect to achieve a significant improvement in your digital ad performance.

Or, as Nate Silver says:

“getting a few basic things right can go a long way.”

The choice is yours—to execute digital the way many advertisers do today, or to move up the prediction learning curve and deliver improved results for your brand.

Get free updates of Randall Beard’s Blog by RSS reader

Get free updates of Randall Beard’s Blog by e-mail

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

%d bloggers like this: