Unstructured Data: Turning Chaos Into Performance
As advertisers, we’re bombarded daily with new ways to work with our data, many of which promise to streamline and maximize our efforts. Yet, while all of the tools and possibilities for digital advertising can lend significant potency to our marketing strategies, they can also result in frenetic chaos. Advertisers need to find ways to […]
As advertisers, we’re bombarded daily with new ways to work with our data, many of which promise to streamline and maximize our efforts. Yet, while all of the tools and possibilities for digital advertising can lend significant potency to our marketing strategies, they can also result in frenetic chaos.
Advertisers need to find ways to tame the chaos and deliver maximum performance on their advertising campaigns. Many see unstructured data as a chaotic addition to their advertising mix, but it doesn’t have to be that way. Here’s how to use the powerhouse of unstructured data to transform your chaos into unmatched performance.
Nix Pre-Packaged Segments & Embrace Unstructured Data
If you’re using a Demand Side Platform or other targeting platform, you are most likely targeting based on pre-packaged data segments. That’s the way most platforms work, and they’re rife with the same data that everyone else uses. This can significantly undermine your campaign performance. With segment-level data, performance is averaged across the highest and lowest performing elements of the audience segment, with limited ability to understand which elements are working and why.
When you run campaigns powered by unstructured data – something I’m a believer in – you’ll not only improve performance but also get a clear view into which data elements are successful and which aren’t.
Search marketing provides a good example of the benefits of targeting using unstructured data elements. Most would agree that search marketing would never have been as prosperous as it has been if advertisers were only enabled to upload lists of keywords, apply the same bid price across all keywords in the group, and were only shown delivery, CPC, and CPA performance for the keyword group as a whole. The ability to bid, report, and optimize at the individual keyword level is one of the key features that has made search marketing successful.
The same expectation should be applied outside of search. For example, the “search-like” capability to bid, report, and optimize at the keyword level in a display-based search retargeting campaign will enable improved performance in a campaign, whether the campaign is measured on a CTR, CPC, or CPA basis.
When pricing control and optimization occur at the element level, the door is opened for heightened visibility into campaigns and an understanding of what campaign elements will pack the most punch with your audiences.
Sometimes, it might just be a minor shift in keywords – swapping “mobile phone” in for “smart phone” – based on the ability to see that “smart phone” is driving a higher return on investment. This ability to see performance at the keyword level drives major improvements in campaign ROI.
In site retargeting campaigns, using data in its unstructured form yields similar positive results. Instead of targeting all users who have visited a site as a unified group, unstructured data targeting enables optimization down to the most micro of elements – like pages, categories or products viewed.
It gives you the ability to bid, report, and optimize based on very specific details of your audience’s behavior and the ability to improve the performance of site retargeting campaigns.
The power of unstructured data doesn’t end with keyword and site retargeting. The same concept applies to contextual targeting and CRM targeting, as well. With keyword contextual targeting, advertisers can leverage unstructured data to define custom contextual categories, and to optimize bidding based on the performance of individual keywords that are present on pages.
With CRM targeting, the value of unstructured data is also high. Through integrating your offline CRM data online without putting it into pre-packaged segments, you can target much more granular behavior… again driving improved performance and deeper insights.
Performance Means Going With What Works
One way of demonstrating the value of unstructured data is to observe the variability of performance of data elements within campaigns. The chart below shows examples of click-through rates (CTRs) of the best 5% performing keywords, and the worst 5% performing keywords for campaigns in the Auto, Retail, and CPG verticals.
In traditional, segment-based targeting, all of these elements would be treated the same; there would be no insight into which keywords were performing well and which ones were not.
However, by reporting performance at the keyword level, it is seen that the top performing keywords can perform five times or ten times better than the worst keywords. This data can be used to optimize campaigns around the best performing keywords, significantly improving campaign performance.
For example, a well-known consumer packaged goods (CPG) company recently ran a campaign with the goal of attaining a .1 percent click-through rate (CTR). Through using unstructured data, the company was granted insight into the CTR for each keyword in campaign reports.
This enables automated optimization that allocated budget to the top performing keywords. The result? The company achieved an average CTR of.27 percent– almost three times higher than the initial goal of the campaign.
Unstructured data might seem complex to use; but, it actually is far simpler. This is because it eliminates the time required to model, analyze, and create the pre-packaged audience segments prior to campaign launch. Instead, unstructured data is applied to a campaign, and the optimum audience is automatically built based on actual campaign response.
In a nutshell: while unstructured data may seem like just another chaotic addition to the world of online advertising, it actually is a great tool for driving performance.
Opinions expressed in this article are those of the guest author and not necessarily Search Engine Land. Staff authors are listed here.