I know what some of you beer aficionados are thinking...
not THAT kind of IPA!
This is about the rising privacy standards, the limits of existing tracking and attribution models, and the proposal for a new model, Interoperable Private Attribution (IPA) to improve both privacy security and signal for marketers.
Tracking and privacy have been at the forefront of marketing conversation since the dawn of cookies and early attribution model formats. As much as we would love to say that every conversion came immediately from the first interaction with your business, that simply isn't the case.
So with that in mind, we turn to the attribution model discussion as well. Specifically, the section we are going to be talking about today is a new model that is entering the arena, Interoperable Private Attribution, or IPA.
IPA isn't currently available, but it is a proposal for the PATCG to consider. It was developed by a team of Google, Apple, Mozilla, Meta, and Microsoft employees. The reason we are discussing it today is due to its importance in the future of privacy and marketing.
The reason for IPA’s creation surrounds a main focus on user privacy, marketing utility, and AI server based technologies. Before we jump into each of those points and exactly how it functions, we are going to lay out a more in depth description of attribution and the attribution model concept as a whole.
What is Attribution in Marketing?
Attribution asks how do we credit each ad touchpoint in the path to a customer conversion. Let's say there are 6 ad interactions before a customer converts, how to we attribute credit and to which ad? As you can imagine, there are multiple ways that credit for a single conversion can be distributed across multiple marketing channels.
What are Marketing Attribution Models?
There are a number of attribution models that assign credit for conversions along the conversion journey that brands and marketing teamms can apply. Here is a brief list of the most common attribution models and what they mean.
- First Interaction (First Click or First-Touch Attribution): 100% customer journey credit to the first touchpoint in a sales funnel
- Last Interaction (Last-Touch or Last Click Attribution): 100% customer journey credit to the last touchpoint in a sales funnel
- Linear: Linear attribution entails when customer journey equal credit is distributed across all touchpoints in a sales funnel
- Time-Decay: Credit is distributed across all touchpoints, but with more credit being given as the touchpoint is closer to the end of the sales funnel
- Position-Based: Also known as U-Shaped attribution, this model gives 40% credit to the first and last touchpoints, and spreads the remaining 20% credit across all touchpoints in between the first and last
- One of the better explanations is an "oldie but goodie" from Avinash Kaushik here
Generally speaking, there is not a best attribution model because of the variety of sales funnels out there for different product types as well as ad network use cases, such as how you decide to attribute a conversion from a display ad versus direct traffic or organic search traffic.
Personally, I find myself using Position-Based attribution though. That’s due to the fact that in digital marketing, a full funnel strategy is vital. Along with that, ensuring that your top-of-funnel as well as bottom-of-funnel efforts are rewarded as the most valuable sections of your funnel is key to scaling growth.
It should be noted that there are data-driven attribution models that can be found on specific first-party data tracking platforms like Google Ads. Facebook Ads is not quite at that point yet, however their Facebook Pixel has recently began using statistical prediction algorithms in order to fill attribution gaps in marketing data. It is also noted that Meta (Facebook)'s Erik Taubeneck contributed to this Introduction to IPA (Interoperable Private Attribution).
How Tracking Works Now
Advertisers had a number of options to measure and track their marketing campaigns and ensure that credit (attribution) was given to the right marketing channels.
However, with the introduction of new privacy focused systems like iOS 14 in early 2021, marketers faced significant challenges in the way of tracking. The online ecosystem is moving more towards privacy focus in every sense, and less personal data transfers allowed while browsing the web is the future of user-data.
Here are some of the most common tracking methods out there to help marketers solve the question attributing appropriate credit to their efforts:
- Global ID Tracking (Pixels): Users are assigned a tracking tag ID that then follows users across the web and sending data back to a central server, allowing ad-tech companies to identify specific user actions on all online events
- Cookies: Retaining user information on a browser instance, allowing ad-tech companies to reference that information versus browsing patterns on other networks/websites to track behavior
While relied on for 20+ years, both these methods require users to compromise increasing privacy standards.
As much as we enjoy ads for products that we actually want and benefit from relevancy, privacy and online anonymity is central to the growth of future web adoption and technologies.
Web3 is a great example of that demand for higher transparency and privacy as outlined in Web3.org's privacy principles.
So how do users and advertisers provide ad tech companies with accurate reporting, enable relevancy, and retain privacy sharing less data?
With the introduction of the Interoperable Private Attribution Model, we see a possible early solution that also keeps marketers in business.
The Goal of IPA
Before breaking down the more technical side of how it’s done, a quick note on the goals of the development team of IPA is good to keep in mind.
- Privacy: “Our privacy goal is to limit the total amount of information IPA releases about an individual over a given period of time.
- Utility: “Our utility goal is to support all the major aggregate conversion measurement use-cases.”
- Competition: “Our competition goal is to ensure equal function for all existing and new ad-tech players.”
The Technical Make-Up of the Interoperable Private Attribution Model
So how can your marketing team utilize IPA in order to ensure that your ad spend is properly tracked within an interaction model? The marketing touchpoints need to become something that can still be differentiated while maintaining privacy via machine learning and server-side computation.
Here's where things get technical in the explanation of how IPA handles conversion credit and user tracking, buckle up!
Match Keys: "A secure identifier that can be set by apps and websites people commonly log-in to across devices. Since only the browser/OS can read the match key, and the actual value is never revealed to anyone, it cannot be used for tracking or profiling."
Multiparty Computation: "Matching of ad impressions and conversions happens server-side, within a Secure Multiparty Computation (MPC). The actual values of the match-keys are hidden from the MPC itself. This approach eliminates an entire category of privacy risks approaches like SKAN face. It also enables cross-device conversion attribution."
Blinding & Double Encryption: "In this system, when the server receives the encrypted data, they first apply a ‘blinding factor’, changing the encrypted numbers. Now they decrypt the data - but it has already been changed. So even once decrypted, the server can't see the original match key. Instead of having one trusted server to decrypt the data, we now have two. Before data leaves the user’s device, it is encrypted towards both helper servers. Metaphorically, this is like locking it with two padlocks, one from each server. The first server removes one layer of encryption, then applies its 'blinding factor' to change the numbers before sending them along to the second server. The second server removes the second layer of encryption and applies its own 'blinding factor'. Now the data has been changed twice. Neither server knows both “blinding factors” and neither server was ever able to see the original match key."
Randomness to Deter Gaming the System: "The IPA system will add or subtract a small amount from the correct answer at random. If the correct answer was ‘6 ad conversions’, the system might feed back any number from 4 to 8, with different results each time. This makes it impossible for the ad-tech vendor to identify the behavior of a single individual by running multiple queries. We can make it impossible for ad-tech vendors to game the system this way by introducing a ‘privacy budget’. This means that ad-tech vendors can decide how many requests they want to make, but the more requests, the more noise is added. The more requests they want to submit, the more random noise will be added to the results."
Tracking Conversion Value: "When impression and conversion reports are generated, we can include additional metadata within the encrypted report, such as the conversion value. After the matching stage, the metadata from matched conversions can be aggregated to produce an output report, like the sum of the conversion values."
The Future of the Attribution Model
It's crazy how complex marketing efforts have had to become when it comes to multi-touch attribution model tracking In just a few short years. The types of attribution models will undoubtedly continue to evolve with concepts like IPA.
The click attribution model that you decide to go with seems to hold more and more of a place among the most vital aspects of your marketing strategy when deciding on how to go about executing on growth.
Now that you have the information on where attribution and performance marketing tracking is heading, you are properly equipped to jump into the discussion yourself. If you feel that you are ready to take the leap and jump into marketing at the pro level, check out how we can help with your growth. Between our PPC agency, affiliate marketing team, and conversion rate optimization service, I'm confident that we can assist you in carving out your piece of the pie. Let's get to work!