Tag Archives: Casandra Murray

Media Mix Modeling, the Next Frontier

By now, most marketers understand the benefits of a multi-channel view of their marketing efforts. It provides a much needed strategic picture of where and how efficiently their media dollars are being spent (read more about it here and here).

But what does a marketer do with the wealth of information in front of him? How does he monetize this new found knowledge? The easy answer here is to use this acquired data to optimize.

At this strategic level, the marketer is not optimizing intra-channel efficiency (the profitability of an individual channel), but rather the mix of media dollars across all channels.  Media Mix Modeling can be a very useful tool, helping the marketer make decisions which are backed up by historical performance and statistical significance. It can be used to create a strategic, media dollar, potential efficiency curve.

The curve is created by stacking the individual efficiency potentials of each channel together and plotting them on the same graph. This provides a comprehensive view of the relative profitability of each channel at any particular budget. It’s easy to see at what level of media spend the diminishing returns set in for each individual channel and the entire media mix as whole.

For instance, SEO and paid search brand channels are extremely efficient at very low levels of spend. However, these channels quickly become maxed out and lose their efficiency. Using a chart like this, the marketer can quickly see each channel’s potential and distribute the media dollars accordingly.

In order to take full advantage of these optimization techniques, the marketer must have control of budget making decisions at the strategic level. The usefulness of Media Mix Modeling quickly dissipates if each, or even a significant minority of channels, is “siloed” into preset budget regales of their relative inter-channel efficiencies (read more here).

The historical performance of aggregate marketing efforts can be seen by plotting several Media Mix Modeling curves on the same graph. Assuming that overall media efforts are improving, the curves will move up and to the right, showing the increased efficiency and scalability of all inter-channel media.

What Google Phrase and Exact Match Close Variants Means for Marketers

Google is releasing changes to paid search that allow the option to run phrase or exact keywords on closely related terms. Variations of keywords that include singular versus plural, misspellings, acronyms, abbreviations, accents, and the consideration for alternate suffixes can all be found using this new Google element. By using this feature, Google’s matching technology can identify any variation of these keywords and apply them to your search advertising.

In light of this change, marketers will want to examine their keyword strategy for Google campaigns:

  • Advertisers are automatically opted into the new Google feature.  They will be entered into more auctions, and more ads will show for these variants if they do not opt out of using the variants of phrase and exact keywords. This could potentially affect the competition on keywords by inadvertently increasing the CPC for advertisers
  • However, Google has indicated that the Quality Score will still be determined by exact keywords and not by close variants and that quality score and first page bid estimate should not be affected
  • Initial adoption is best suited for advertisers who are limited in their campaign management capabilities rather than advertisers who closely manage their campaigns. Advertisers who closely monitor their accounts should use this new keyword match type as an additional lever for keyword discovery similar to current strategies applied against broad match terms for both positive and negative contributions

Google has shared initial results that show an overall increase in click volume at comparable CPCs and noted that this may vary from advertiser to advertiser. This additional traffic is likely to be less relevant and less efficient than current traffic. That being said, the new matching behavior is a good feature for marketers who do not currently have extensive positive and negative keyword assets. This may also work for marketers whose main goal is enhancing traffic. CPC for any close variant matched keyword will probably be lower than the CPCs of the actual term, due to lower demand and competition.

For marketers who do have extensive keyword sets built out, this may have a negative effect on performance. These advertisers should monitor keywords closely since increased CPCs may make these terms (which in the past were cheap due to low CPCs and low traffic) inefficient moving forward. Additionally, since the quality score will still be based on exact match keywords and not close variants, it is important to make sure any misspellings and plurals are also included on both exact and phrase match as standard best practice.