Welcome to the world of attribution beyond the third-party cookie.
What are third-party cookies, and how do they work?
For clarity, let’s outline what third-party cookies are and how they work.
Cookies are small files stored in our browsers when we visit a website. Advertisers can use these cookies to learn more about users and their preferences and interests. As a result, brands can target users with more relevant ads.
Third-party cookies allow websites to track their users all over the internet. For marketing teams, this is particularly useful. If they advertise over various channels (i.e., Facebook, Google Search, and other publications), they can see the steps a user takes from awareness to making a purchase. This can help them understand which campaigns are the most effective.
Why are third-party cookies disappearing?
The simple answer is that user privacy is becoming a major concern for internet users. Tracking users all over the internet was a great way to know more about their interests and the various touch points they interacted with before settling on your product.
By understanding the steps it took to generate a sale, marketing teams could measure which channels — and what messages — generated leads, sales, and revenues.
Increased privacy means that the source of information will soon be gone. Safari and Mozilla Firefox have already stopped cookies. Google Chrome is set to do the same; however, they keep pushing back the deadline and recently suggested that they will remove third-party cookies in 2024.
How does attribution work?
Attribution is a fairly simple concept. The modern buying process is complicated. It involves many different touchpoints to make a sale. Some of these are:
- Pay-per-click (PPC) advertisement
- Search engine marketing
- Social media marketing
- Content marketing
- Affiliate links
If you make a sale, it’s essential to know what caused the buyer to act. Was it your PPC ad, social media posts, or great SEO? Marketing teams need to know which channels are driving results — and which aren’t.
Attribution helps brands apportion credit for a sale to different channels.
What are the current attribution models?
There are a few popular attribution models that are used by many marketing teams. Each works on a similar principle but attributes credit for sales at different points of the buyer’s journey.
First interaction: Gives 100% of the credit to the first touchpoint with your brand.
Last interaction: Gives 100% of the credit to the last touch point with your brand.
Last non-direct click: If a buyer directly types in your website and makes a purchase, the touchpoints that led to that action go unattributed. This model considers the last non-direct action and credits it with the sale.
Linear: The linear models share attribution equally across all the touchpoints.
Time-decay: Similar to the Linear model, it gives weight to touchpoints closer to the sale.
Position-based: Gives 40% of the credit to the first and last touchpoints and splits the remaining 20% for everything.
As you can see, all these models have merit. However, they rely on accurate third-party cookies to function.
As third-party cookies are gradually removed, it’s dawned on the marketing community that we’ve been spoiled over the last few years. Having all this third-party data made measuring and fine-tuning campaigns very easy.
How attribution can evolve
So, this new world will require an adjustment if we want to measure marketing campaigns.
One of the most important requirements is trustworthy and reliable engagement data. However, that may be more challenging than it appears.
One problem here is that attribution itself is reasonably flawed. The buyer cycle is complex and human behavior and psychology doesn’t always fit neatly into most models.
It’s hard to say how, for example, Facebook can contribute to a conversion. If you have several touchpoints and convert to multiple channels, attribution can be full of inaccuracies and false positives.
Another big issue is when that same inaccurate data is used to power your remarketing campaigns. An old computer science phrase goes, “Garbage in, garbage out.” While the data we’re talking about isn’t useless, we all need to be concerned if we are hitching marketing budgets to marketing information’s integrity.
Tackling over-attribution
Digital marketing is all about optimization now. However, if it is to continue to outperform human users, solid data is necessary.
The last non-direct click, as described above, is still one of the best models for measuring the effect of a channel. Yes, it doesn’t consider the other touchpoints that have led to a conversion. By design, it doesn’t over-attribute to channels without an effect and, therefore, can be considered reliable, if not exhaustive, data.
The importance of first-party data
With the upcoming deficit of third-party data, many brands will turn to first-party data sources. While this could be a positive development, the reliance on user consent could limit its effectiveness.
Collecting first-party data (from CRMs, surveys, emails, and social media interactions) will be essential if you want to use Google and Meta “lookalike” audience features. So the area has a lot of applications for audience targeting.
Alternative attribution models
Here are a few alternative models that could offer a solution to phasing out third-party cookies.
Conversion modeling: Conversion modeling uses machine learning to fill in the gaps where you can’t observe conversions.
Marketing mix modeling: MMM uses regression analysis to understand the impact of different sales channels.
Impact-driven effect modeling: Impact-driven effect modeling, which is what Amanda AI uses, measures a variety of permutations of your creatives and ad copy to see which has the greatest impact. Our machine makes up to 5 million daily adjustments, meaning you can track which ads had the greatest effect without relying on third-party cookies.
Conclusion
Third-party cookies meant that advertising teams had things too good for too long. However, as they are phased out, new attribution models are emerging.
Collecting accurate and reliable data is the biggest hurdle facing these new models. Full, all-knowing attribution is almost impossible due to the complexity of human decision-making. However, ensuring your data is the cleanest it can be and without misattribution will help power this new generation of attribution models.
