How Machine Learning will save marketers in a cookie-less world
Note: A version of this story ran in Episode 3 of my Marketing & The Machine podcast.
Just a few years back, I don’t know if I knew how good I had it as a marketer. Over various campaigns I’d lean heavily on the all-mighty cookie logging and tracking what would happen after a user clicked on my ads. Reporting was streamlined and I could easily see what channels converted best, make modifications to fix the lagers, and get an excellent view of my customer’s journey.
Just thinking about it makes me want to pour a glass of Country Time Lemonade and spend the afternoon in a rocking chair.
That’s not the case anymore. Today, marketers are up against cookie restrictions built into browsers like Firefox and Safari (and soon to be Chrome), users flipping between devices and browsers, global privacy regulations, and other blind spots. It’s challenging marketers to be more forward-thinking and privacy-focused.
And as the cookie has crumbled, we’re now led to a very complex landscape with gaps in digital media measurement and conversions becoming more and more difficult to track. Like it or not, we’re in the midst of a measurement evolution!
So, how can machine learning help in this new cookie-starved world?
Let me introduce you to conversion modeling. In layman's terms, it just means using machine learning to fill in the gaps when some of your conversions can’t be measured. It saves your butt and helps you show the real impact of your marketing efforts when you’re unable to attribute conversions to a specific channel or medium.
With conversion modeling, any known data you can collect (such as device, date, time, conversion type) is fed to an algorithm. It’s paired up with historical trends to confidently validate and inform measurement.
The large benefit here is that conversion modeling gets you an accurate measurement while only reporting on aggregated and anonymized data. You get a privacy-centric and deep picture of your customer behavior, while not giving up the accurate reporting of the good ole’ days.
That’s a huge impact on the overall health of your business. Being able to measure accurately is such an essential foundation upon which your ongoing learnings, decisions, and optimizations are built.
The key to conversion modeling though is making sure to implement solutions that can help increase the amount of observable data for your campaigns. The more known (and accurate) data you can feed the machine learning model, the stronger your foundation, and in the end, the better the model’s results.
The inspiration of this post was an article recently posted in Think with Google. They’re of course, pushing tools like Google Tag Manager or global site tags on your website to ensure you’re set up for measurement success, but alternative options like Adobe Analytics would also do the trick.
