Marketing attribution has always been imperfect, but as consumer behaviours continue to evolve, and the landscape shifts, perhaps the impending demise of the cookie is an opportunity to reassess and pivot.
Attribution has always been a messy affair, but the objective is simple; to assess which marketing channels and creative executions are influencing consumers and forming part of their “customer journey”; this enables marketers to understand which channels are most effective, so they can optimise delivery and cut wastage.
The Birth, and Problems, of Last-Click
In the beginning, there was “last-click attribution”, which isn’t really an attribution model, it’s the lack of one. Last-click was a basic and naïve way to track conversions and events online, by effectively saying that whatever customer click came last, was awarded full attribution of the sale or conversion. There were many problems with this approach, even in the beginning, such as the fact that digital marketing doesn’t operate in an isolated bubble (although some like to think it does); someone who clicks a search result and then buys may have been influenced by any number of things, such as a TV ad, a display ad, or a double-spread in a local newspaper, so it’s rarely fair to assume the last point of contact (in this instance, a search result) was the one thing that drove a customer action.
We know that customer journeys are long and varied, it’s rare that someone spontaneously thinks of something, decides to search for it, and purchases from the first result they click. But this is the false assumption that last-click forces marketers to make. The truth is that someone is much more likely to have been exposed to advertising multiple times in various places over a long period of time before they decide that they want something. This is exactly why the metrics of “reach” and “frequency” are so important in advertising!
Despite the imperfections of last-click attribution, it was the default model used across many early advertising platforms, including Google’s own search ads because it was easy and simple, and worked in Google’s favour (initially). Google didn’t need other channels, such as TV, receiving credit for driving conversions, they were happy with the fact that last-click artificially inflated the credit for search ads (because Search is obviously lower-funnel activity that appears towards the end of a customer journey).
As the digital landscape shifted and expanded, marketers decided they wanted to measure more of the customer journey to see what channels really influenced consumers, and where was truly best to invest marketing budgets. This is where attribution became desirable to marketers, and there was a flood of new tools and products to help facilitate this.
Starting with Google’s acquisition of DoubleClick in 2008, and the subsequent diversification of their advertising offerings, they too started to look at baking in other methods of attribution across their platforms, and as of 2021, Google are heavily pushing attribution and smart bidding solutions across search.
The message was clear, measuring multiple touch points in the customer journey and finding a way to credit them more fairly was much more effective on the whole, than focusing on individual channels and relying on biased data from “walled garden” platforms to determine effectiveness.
Attribution wasn’t a silver bullet, it allowed marketers to apportion credit across channels in a templated way or for them to make their own attribution rules, but people don’t always fit into templated rules, as they are possess free-will and are unpredictable, and no one customer journey is really the same. Some products claimed to take attribution even further and use AI or machine learning to build attribution, but even these have had limited effectiveness.
Regardless, in the absence of a potentially-unobtainable perfect solution, attribution seemed to tick most of the boxes and get marketers closer to a position of understanding what their average customer journey looked like.
Cookie Reliance of Attribution
The problem was that attribution mostly relied heavily on cookies. In the noughties, this wasn’t so much of an issue. If you had asked a stranger on the street, you’d be surprised if anyone knew what a browser cookie was. Cookies were seen as harmless blocks of data left by websites on a user’s device to remember settings, and record session data. Nevertheless, apart from the fact that Cookies only worked for tracking online journeys (leaving marketers and attribution platforms to plug the gaps for tracking offline channels), Cookies were also flawed in other ways.
“Not everything that can be counted counts and not everything that counts can be counted.”William Bruce Cameron
The problems presented by cookie-based attribution grew once consumer behaviours changed. Once the iPhone was born, it sparked a smartphone revolution, which in turn lead to an explosion of smart and internet-connected devices in the home. It meant that people no longer used one device for all of their online consuming, and their usage was split across many channels, and many devices. Cookies are device specific, so once again, this meant that marketers and attribution platforms had to find other ways of connecting these journeys (or some just didn’t).
In 2012, the EU introduced new laws that required consent for websites to drop Cookies, and a plethora of website Cookie-consent solutions (largely in the form of “Cookie bars”) entered the market. Eventually, the implementation and adherence to these laws were watered down, with many websites adopting the concept of “implied consent”, which in turn was endorsed by the UK’s ICO, and the issue appeared to fade away for a while. However, the intention of the EU was made clear, and repeated privacy issues that came to light in the subsequent years directly influenced the introduction of tougher new laws in the form of GDPR (the EU’s recent General Data Protection Regulation act).
The Demise of the Cookie
“Keeping the internet open and accessible for everyone requires all of us to do more to protect privacy — and that means an end to not only third-party cookies, but also any technology used for tracking individual people as they browse the web.”Google
With the position of Cookies already weakened severely by GDPR, and with the impending demise of third-party Cookie support in most major web browsers, marketers and platform developers are left with finding alternative ways of doing everything they’ve taken for granted as being able to do for the past 20 years. It impacts almost every area of digital marketing, from profiling users, targeting audiences, serving ad inventory, measuring ad performance, and more. It would be conservative for me to say that there has been a scramble for marketers to plug the gaps left by the end of Cookies.
One area that has fallen victim, is of course, attribution. Marketers are now looking for what the post-Cookie future of attribution looks like. How can we balance the rights of consumers to privacy, with the noble desire for marketers to know what marketing channels are effective to invest in?
Some have come up with sensible solutions, such as AI, or leveraging the “authenticated web” (people who are logged in to websites) to help model attribution methodologies, but many agree that these methods alone are flawed. Only 5% of the web is authenticated, and trying to find our way back to find easy off-the-shelf templated solutions might not be the best route forwards.
Attribution is important, and knowing where to spend marketing budgets and optimise delivery is critical to what we as marketers do, but like most things, you could look at this as either a glass half-empty, or a glass half-full situation. Whilst the end of Cookie-based solutions is a frustrating obstacle for marketers, it’s also an opportunity to diversify and go back to the drawing board. Cookie-based attribution was never perfect anyway, it was just a convenient way of reporting, so we shouldn’t be mourning the loss of out-of-the-box solutions that were just handed to us. We should embrace the fact that we need to be smarter, understand the data we have, connect the dots, and challenge the insights. This is what many are calling “analog measurement”, it means getting your hands dirty with data, sticking your head into spreadsheets, and actually looking at the data yourself.
Instead of automating attribution and insight, we now need to rely once again on the human experience, expertise, and interrogation of client data. This places power back into the hands of experienced marketers, away from blackbox platforms, and will push testing and innovation back to the forefront.
It presents an exciting new chapter in our industry, one we should immerse ourselves into, and one that emphasises the importance (and employment opportunities) of broad marketers (“generalists” and “T-shaped” marketers), and on Data Analysts, Data Scientists, and on anyone who just loves marketing and reporting. This also means that clients and businesses benefit, as their data will be analysed and reviewed to find their own effective journeys and solutions, and continued testing strategies should yield learnings and incremental advances in performance.
Attribution isn’t dead, it has diversified, and it’s been long-needed. Platforms and solutions will continue to rise and fall, and there are already alternatives to Cookie-based platforms coming up in the market, such as API-driven attribution from within the browser, or the evolution of Google Analytics attribution modelling, but we should use these as tools as part of a wider toolkit, and not as an absolute approach to attribution.