Last Updated on: 6th February 2024, 11:26 am
Digital advertising has come a long way. From direct deals between brands and publishers to the extensive ecosystem that we operate in now. As did marketing campaign reporting.
In this article you will learn about:
- The current shape of campaign reporting
- Marketing campaign reporting in the cookieless future
- Tech giants’ privacy-preserving measurement approach
Table of Contents:
- How marketing campaign reporting works now
- What implications will 3rd-party cookie deprecation have on campaign reporting
- What Google proposes as the new approach in a cookieless future
- How Apple’s Private Click Measurement approach is similar to Google’s
- What is new about Mozilla’s and Meta’s approach
- What rules will dictate the future of reporting
How marketing campaign reporting works now
At first, only some general metrics, sent by the advertising partner in quarterly or monthly reports, were available. Today, in the programmatic ecosystem, data and insights are available almost instantly, and the number of available metrics is astonishing. How is it all possible?
Currently, every bid request, advertising impression, click, and, eventually, conversion can be individually analyzed. Information connected to each individual event is essential to assess:
- Is the bidding strategy correct?
- Do we reach enough users?
- Do the ads themselves or the products displayed convince users to click on them?
- Will they eventually buy advertised products?
These are just some of the aspects, and the number of used metrics is significantly larger. However, while answers to these questions help on their own, the most valuable is the connection between these separate events. Only such a combination helps to assess whether the bids are properly determined for users who then click on ads and buy products. Now it is possible thanks to tools such as third-party cookies or link decoration.
What implications will 3rd-party cookie deprecation have on campaign reporting
A 3rd-party cookie is a unique identifier, allowing to connect different data points to a single user. When a third-party cookie is no longer there – without replacement technology – it would be almost impossible to tie what happens on the publisher’s website, e.g. ad impression or click data, with events on the advertiser’s website, such as conversions.
With cookieless advertising, the general data, such as how many ad auctions a specific campaign won, how many clicks were attributed, will still be available through other techniques, like link decoration. But it won’t help answer the question: why was a specific auction won? Why did the user click on the ad? Cookieless advertising significantly limits the possibilities of campaign optimization, regardless if it’s manual or using machine learning algorithms. It directly leads to less personalized ads, more difficult ad frequency management, leading to less optimal budget spending. Therefore, browser vendors have put a lot of effort into creating new, more privacy-preserving reporting tools, useful for advertising in preparation for a cookieless future.
What Google proposes as the new approach in a cookieless future
Google decided to start by creating a tool for specific reporting use cases within marketing campaign reporting. Therefore, it announced Attribution Reporting API. This concept is still under development, and it’s receiving a lot of constructive feedback from some of the key open internet players, like Yahoo!. It has recently been updated together with the publication of the specification of FLEDGE for the first Origin Trials. This proposal that aims to tackle the issues surrounding cookieless advertising introduces on-device reporting, which includes two types of reports: event-level reports and aggregate reports
Event-level reports are aimed at: optimization, coarse reporting, and fraud detection use cases. They help answer the question: which click led to which conversion? To improve user privacy, conversion-related data will be strictly limited. This allows some inputs for training machine learning models, but won’t be enough for precise campaign reporting. To improve the privacy of this report, Google announced randomized responses, adding a small amount of noise to actual data in the form of conversion data for incomplete action or no data for some delivered conversions.
Aggregate reports offer more detailed conversion data, but without the possibility of distinguishing individual events. They aim to answer questions such as: what is the return on investment or what was the total ad spend?
Google’s approach looks promising for not only the traditional last click, but also view-through attribution-based reporting.
How Apple’s Private Click Measurement approach is similar to Google’s
Apple has already implemented their privacy-preserving measurement approach called Private Click Measurement. It allows linking individual events on the ad click (publisher) side with the conversion side (advertiser) with some limitations. These include:
- There will be no individual user ID included
- Only up to 256 parallel ad campaigns can be measured per website or app
- Only up to 16 different conversion event types can be distinguished
- Attribution reports will be sent with a random delay between 24 and 48 hours
The limits on both the publisher and advertiser sides are aimed at forbidding tracking through a method called fingerprinting, which could try to combine different data points to identify an individual. One example of such a technique is a time correlation attack. If the reports were coming instantly, the publisher would know exactly at what timestamp a specific user clicked an ad, and then the advertiser would know when they landed on the website, which could lead to identification. Private Click Measurement works both in web-to-web and app-to-web use cases, but only on Apple’s Safari and in the iOS ecosystem.
Apple has also introduced SKAdNetwork, which is a privacy-preserving tool for reporting the outcomes of app install campaigns. It is also based on a time delay attribution. According to the specification, it takes between 24-48 hours between an ad impression and an install-validation postback received by the ad network.
What is new about Mozilla’s and Meta’s approach
Late 2021, key privacy engineers from Mozilla and Meta released the Interoperable Private Attribution concept. Because of its design, Interoperable Private Attribution satisfies only aggregate reporting use cases. Within this area, it provides answers for cross-device and multi-touch attribution, which were not covered in other proposals.
Cross-device attribution could be possible thanks to the use of global Match Key providers. How? While browsing the internet, you must have come across pages where you can log in “using Facebook” or “using Google”. It is possible because these companies serve as identity providers, and can pass e.g. your email address to the website for login purposes, if this is your wish. This functionality works both in desktop and mobile web. In the Interoperable Private Attribution concept, these companies could hold the Match Keys, which could serve only for matching ad impressions with conversions, also across devices. Due to the privacy mechanisms introduced in this concept, Match Keys won’t help in decrypting the true identity of the user
What rules will dictate the future of reporting
It is unquestionable that the currently used cross-site individual-level identifiers following users around the internet will be significantly restricted in the future web. New tools will have to balance the limitations on the breadth of metrics and granularity of reports, so that users cannot be identifiable. This will definitely be challenging, as delivering efficient advertising requires machine learning optimization, trained on specific events, while on the other hand, brands require ad tech companies to deliver a broad range of metrics to evaluate the investment.
In some cases, time delay in reporting will also be problematic – 24-48 hours make a difference. On the other hand, delivering insights into multiple metrics on an aggregate level, without the possibility of checking individual users, should be satisfactory for the advertisers. There will definitely be less possibilities of creating custom reports, as granular data won’t be available. Moreover, there will be more gaps in some data sets, such as conversion insights, as no tool will cover 100% of the cases. In this case, modelling is the only way to bridge these gaps, but it requires adtech companies to earn the trust of their partners.
If you have any questions, comments or issues, or you’re interested in meeting with us, please get in touch.
