• by Paul Pellman , August 10, 2011

Rarely is there an advertising conference one attends or an opinion piece one reads where you don't hear about "campaign optimization." But as advertisers and agencies know, online marketing campaign optimization remains an elusive holy grail in this new blended world of hybrid search, social and display advertising.

What many are beginning to realize is that online ad campaign optimization is complex and is only as good as the available data. Today, one campaign can include media placements across many different channels -- whether they're display ads bought on an ad exchange or a search term bought through Google.

Properly measuring how each of those media buys contributed to conversions is paramount to figuring out where to reallocate spend in the future to boost ad campaign performance.

The foundation of any good ad campaign optimization solution is the data it collects thought attribution analysis. Attribution is the science of measuring how each ad placement or media buy contributed to campaign lift or conversions. There are many flavors of attribution analysis and herein lies the problem. Not all attribution is created equal. Many times, the data collected by attribution products is incomplete, which in turn, leads to improper or misleading campaign optimization advice.

To keep it simple, here are three telltale signs your campaign optimization suite is not using proper attribution analysis.

• Optimization Based on "Last Click" -- Many attribution offerings still rely solely on last click (or last event) attribution. This means the entire weight of the credit for conversions is given solely to the ad that generated the click. The problem with this approach is it ignores or assigns little value to all other interactions and channels that influenced your campaign. This means optimization would be heavily skewed toward campaigns that generate clicks, like paid search (SEM), resulting in poor investment decisions in the future.
  

• Optimization Based on Arbitrary Data --Some have tried to devise ways to more evenly distribute credit for campaign contributions. One method divides up credit evenly among the different channels (last click, search, display, social). The other arbitrarily assigns different credit to each (last click 50%, search 20%, display, 10 %). Unfortunately, these methods are merely guessing at the effectiveness of each channel. Therefore, they do not provide the real insight marketers need to optimize.
  
  
• Optimization Based on Sampling -- sampling is a common method used by many attribution practitioners. Rather than processing an entire data set from an advertising campaign, these attribution analyses consider only a sample of the data. For example, some only examine the paths of converted visitors, ignoring all of the other marketing interactions that didn't result in a visit or conversion. Credit is, again, incorrectly weighted, leading to inaccurate campaign optimization recommendations.
  

The right way to handle attribution is by measuring the influence each and every display impression, click, social interaction, and e-mail campaign has on a campaign. This approach is called fractional attribution. It processes 100% of all campaign data and assigns credit to the impact each path had on delivering conversions so that complete patterns and trends can be identified. By basing optimization and forecasting algorithms on this data, online marketers can fine tune and improve campaign performance.  

When relying on campaign optimization to make decisions about future online advertising investments, it's important to take a close look at "if" and "how" attribution is measured. After all, without proper attribution, there is no real campaign optimization.