The IAB Tech Lab, the global digital advertising technical standards body, has made the mapping from Content Taxonomy 1.0 to Content Taxonomy 2.0 available for public comment as of December 11, 2024.
Originally developed by the IAB in 2011 and officially deprecated by the IAB Tech Lab in 2020, it’s time for the industry to move on from Content Taxonomy 1.0.
To date, a major barrier to moving on has been the heavy lift required. Upgrading to a new taxonomy involves many moving parts, such as relabeling content already labeled in the previous taxonomy, upgrading UIs and retooling existing workflows and operational strategies.
The conversations about the impact of this process, and whether, how, and where to start, are difficult ones. A solid mapping between the current taxonomy and a more recent version provides a tangible way to ground those conversations because it can highlight what’s at stake. Yet, creating such a mapping can pose a barrier in and of itself as not all organizations have capacity or capability to develop one.
This mapping aims to remove that barrier, providing a straightforward path to get organizations still using Content Taxonomy 1.0 to label the “aboutness” of content transitioned to Content Taxonomy 2.0. This year, the IAB Tech Lab also released a mapping from Content 1.0 into Ad Product 2.0, making this the second mapping from Content 1.0 into another taxonomy now available.
Because of the importance of this mapping to the industry, Subtextive wanted to accelerate the IAB Tech Lab Taxonomy & Mapping Working Group’s efforts. With the help of our soon-to-be-released AI-based taxonomy mapping toolkit our taxonomy team developed an initial mapping, which we were thrilled to donate to the Tech Lab.
The Taxonomy & Mapping Working Group, comprised of 237 people across 112 companies, then followed their standard processes and due diligence in reviewing, discussing, revising and finalizing the mapping for release for public comment, which is where we are today.
Who should use this mapping?
This mapping is designed for anyone using Content Taxonomy 1.0 to describe the “aboutness” of content that is not ads/creatives themselves and that is not an audience.
Publishers and DSPs who have been using Content 1.0 to describe their sell-side content inventory, ad verification vendors and brand safety/suitability providers who have been using Content 1.0 to describe digital content and anyone with similar use cases and digital content should use this mapping.
On the other hand, advertisers and DSPs who have been using Content 1.0 to categorize ads should move to Ad Product Taxonomy 2.0 which was created to describe the product categorization of an advertisement. There is now s a supporting mapping from Content 1.0 to Ad Product 2.0 and vice versa.
More information about migrating from Content Taxonomy 1.0 and the taxonomies that individuals should migrate to based on use case is available in this resource about migrating from Content Taxonomy 1.0.
Setting the stage for public comment
Here are some general fundamentals about taxonomy mapping that we hope will be helpful as you formulate comments.
What is a taxonomy mapping?
A taxonomy mapping associates categories in one taxonomy to categories in another taxonomy based on how similar categories are to each other in terms of meaning, scope, and usage, etc.
It is important to note that this taxonomy mapping is for released taxonomies that are already in production. Neither taxonomy can be changed at this point to make the mapping more complete or “better”. We can’t add categories to or change the structure of either taxonomy. We have to work with the puzzle pieces available.
What makes a mapping “good”?
Taxonomies capture the purpose of the taxonomy and the use cases it supports. They also reflect the zeitgeist at the time that they are created, which includes cultural norms, societal influences, business, economics, consumer products, etc. There will always be differences between taxonomies.
As a result, mappings between taxonomies are never truly one-to-one, exact matches or “perfect”. Instead, mappings are always a best effort that take into account a number of factors, such as how each taxonomy has been – or will be – used, and by whom, the scope and meaning of each category and, importantly, the impact of the mapping.
So, what makes a mapping “good”? To start with, a good mapping is a defensible one. This requires using a principled approach that 1.) considers the meaning, interpretation and usage of each category in both taxonomies; and, 2.) attempts to resolve differences between taxonomies according to a consistent methodology wherever possible.
Ultimately, what makes a mapping good is how effectively it aligns one taxonomy with another semantically and how it handles discrepancies between taxonomies.
This is where a careful consideration of such decisions and principled approach come in. For a mapping to be “good”, It’s crucial that differences between taxonomies be handled as responsibly and effectively as possible.
About this mapping
In formulating specific comments about this mapping, here are some highlights that might be helpful.
The overall package
As mentioned earlier, taxonomy mappings are always a best effort but not perfect, exact matches.
This mapping is no exception. Yet, it stands out as an A+ effort because of the thoughtfulness of the mapping coupled with its easy to consume presentation and “packaging”.
The mapping itself is provided as a .tsv file in the Taxonomy Mappings directory of The Tech Lab’s GitHub repo. The .tsv uses clearly labeled, dedicated columns for Content Taxonomy 1.0 category ids and names along with corresponding Content Taxonomy 2.0 category ids, names and supporting tree structure. This supports human readability and interpretation of the mapping while enabling automated processing.
To support operationalization, practical and actionable guidance to accompany the mapping, is accessible here. This is a living document to be expanded and refined by the Working Group in response to questions that come in from the community, a testament to their thought-leadership and dedication to a usable upgrade path.
The documentation as it exists today provides insight into the mapping itself, its guiding principles, choices that the implementer should be aware of and guidance for handling the handful of cases that spotlight key differences in taxonomy versions.
Details: The good, the bad and the ugly
The good
Almost all categories in Content Taxonomy 1.0 have a defensible counterpart in Content Taxonomy 2.0.
The thoughtfulness and consideration put into the mapping by the Working Group is clear. Decades of taxonomy experience from practitioners in the advertising industry were brought to bear on this mapping. These are folks well-versed in the challenges of applying these taxonomies to content and using them in the programmatic supply chain. It shows.
The bad
There are a small handful of categories in Content Taxonomy 1.0 that don’t have a great counterpart in Content Taxonomy 2.0. That said, a defensible mapping has been provided.
In some cases, these are categories that could be mapped to more than one Content Taxonomy 2.0 category. So, folks can choose from the available options to create a mapping, keeping in mind that any one of the choices will change the semantics of the category in certain ways. Even so, a default or “suggested” mapping of a single category has been provided.
In other cases, there wasn’t an analogous category in Content 1.0. In these cases, where a suitable parent category existed, the category was mapped to that parent category. You can access the guidance for those categories here.
The ugly
There are fourteen categories where there wasn’t a defensible Content Taxonomy 2.0 category that the Working Group could find.
For reasons that are unclear to the current Working Group, equivalents for these are absent in Content Taxonomy 2.0. Yes, absent.
Nine of these are categories that today we think of as brand safety and suitability categories, like “Profane Content”.
We can’t change the taxonomies themselves when we’re doing a mapping. We have to use the puzzle pieces that we have. This means leaving them unmapped in this mapping.
While we realize the importance of these categories, it’s up to publishers to self-label and up to the buy side to determine what is appropriate for their brand use case and the accuracy of those signals.
There is a path for those who want to support these categories.
These nine categories are covered in Content Taxonomy v2.2, which is identical to Content Taxonomy 2.0 but adds seventeen categories to cover “Sensitive Topics” and “Brand Safety and Suitability”.
Our recommendation is to map those nine categories to their Content Taxonomy v2.2 counterparts. By doing that, you’re going for extra credit and will now be on Content Taxonomy v2.2, which is just one release away from Content Taxonomy 3.0.
What about the other five categories, which are a mixed bag of categories to cover things like “Uncategorized” and “Incentivized”?
If you use these categories on either the sell side or the buy side, our recommendation is that you retain them.
Importantly, keep in mind that the mapping itself isn’t flawed because there are gaps in Content Taxonomy 2.0. Simply be aware that these gaps exist and be ready to handle them as appropriate for your needs.
What we’re asking of you
This is your opportunity to provide public comment!
Take it for a test drive. Or, at the very least, think through how this mapping would work for you. The Working Group wants to hear your thoughts and questions about this mapping.
By sharing your perspective within the 30 day public comment period, we’ll be able to understand what to do to release this mapping. Your input will allow us to tune the mapping and revise the Implementation Guidance to support the industry as best we can.
This mapping and Implementation Guidance is open for public comment as of December 11, 2024. Please leave any comments that you have on this issue in GitHub. We need your input now. Once this mapping is fully released, we’ll be following up with another ask, which is to upgrade. Stay tuned!
Temese Szalai
CEO
Subtextive