Nothing is Certain Except Data and Taxonomies
I find myself having conversations about data taxonomies enough that I thought it warranted a blog post.
My conversations about data typically start with a spirited discussion on use cases and platforms, and almost always end with me discussing how data structure is a key step to making sure that once all of the data is brought together (usually for use in a CDP), it’s usable. The data has to be useable in terms of data quality, but also in terms of context. GIGO is a very real thing (Garbage In, Garbage Out), but there are implications of having messy data. Extraneous, misaligned data can waste space and time.
I believe that data can be the most valuable commodity a person or entity has, and the real value lies in how it can be used. If the data is stored and structured in a meaningful way, it can be worth its weight in gold.
So, let’s start at the top…
What’s a Taxonomy, you ask?
According to the Oxford English Dictionary, a Taxonomy (/takˈsänəmē/) is “a system of classification.” It can be used in many different ways, including in Biology with regard to organisms, but it can also be used for inanimate objects and data, like attributes/qualities of an item.
I often talk about clarity of language in business, based on function and context. This is one of those times that this is especially important. Product hierarchy and classifications are part of a Taxonomy. Marketing parameters and campaign attributes are part of a Taxonomy. Yet these are two separate but related systems of classifications.
Why is this Actually Important?
Low quality data is a problem for so many reasons. The quality isn’t just contingent on the setup within the technology itself, like incorrect values or errors in processing; it can also be based on how the data is organized. If data points are correlated incorrectly or without consideration of the business need, not only can it lend to operational inefficiency, but it can also cause incorrect insights to be derived.
Channels, Hierarchies and Parameters, oh my!
As I mentioned above, business context is critical to these exercises. My examples below aren’t intended to serve as specific guidance for a business, but rather a way to consider the setup.
You can reach out to me directly if you want to discuss your business needs.
We’re uniquely positioned as an industry to be able to bring together 1P and 3P data in a more integrated way, through platforms like CDP’s and using various API’s and server side solutions. Historically, while we’ve had more access to raw, granular tracking data, privacy centricity and regulatory considerations have necessarily changed that approach. This makes operational efficiency more important than ever, and properly structuring data drives faster time to insight and more opportunities to activate on that data.
For the rare business that is starting fresh, identifying and then prioritizing the data sets to properly classify is a great exercise to do from the start. For the overwhelming majority of businesses that already have a significant amount of data, structured in a variety of ways, the exercise becomes a matter of reclassifying the metadata and identifying how best to marry data sets to derive the right insights. Prioritization is still key, and that’s based on what the most critical outcomes and use cases are at the time that the exercise is being performed.
Campaign Classifications & UTM Parameters
UTM parameters, or Urchin Tracking Module parameters, are values added to a URL to track campaign results. These parameters can be ingested in to Analytics systems like Adobe Analytics and Google Analytics, via processing rules that parse the values into meaningful classifications for reporting. Below is an example of the structure and data that can be appended to the URLs.
In order to effectively analyze the attributes, and results, from URL parameters, it’s important to know how the business rules translate those parameters into actionable insights. Finding an illustrative way to associate these values will save time for teams and individuals to understand how to choose the best KPI’s for campaigns, based on forecasts, historical trends, and results. Understanding how best to tie the funnel (Awareness, Interest, Consideration, etc) to channels (SEM, Display, Out of Home (OOH), Social, etc…) to sub-channels, tactics, and products, is an important way to link all of these aspects of a campaign together.
Product Taxonomies
For businesses that are selling products, having them properly associated with categories is another critical exercise. Fruit is a great example of this. Fruit is a top category, followed by type of fruit, then by subtype, color, variety, etc… If you want to sell more apples, understanding what attributes make that type of fruit the most desirable, by season, is achievable through how the metadata is associated with transactions, and ultimately to the people making the purchases. We’ll address audience building, based on these attributes and results, in another post.
Don’t Let Bad Data Happen to Good People
Structuring data isn’t the sexiest effort, but many campaigns certainly are. Properly bringing organized data together can ensure you have robust measurement to accurately plan and activate. While the campaigns themselves are important for engagement, awareness and revenue, if you can’t determine what aspects of the campaign resonated and most effectively drove the intended results, then there’s money being left on the table.
If you have questions, or just want to talk analytics, activation, and anything in-between, reach out to me at Katie@DataOnTrend.com.