Google’s machine learning abilities recently made headlines when a now suspended Google Engineer claimed the AI chatbot named LaMDA was sentient. Whether you believe the sentience aspect of the claim or not, the reality is that LaMDA was trained to talk in a human-like manner by digesting trillions of words scraped from the internet.
On a less philosophical note, Google’s growing reliance on machine-learning has become a driving force that paid search marketers need to train to effectively and efficiently leverage the channel. Think of it like training a dog to fetch.
There are two basic steps to training your “Google dog” to fetch:
- First, your dog has to be taught to value and “hold” the object to be fetched. Translating this to search marketing: it’s not just about the keywords we use in our campaigns, it’s about having conversion goals that place value on fetching the right keywords.
- The next step is to systematically teach your dog to fetch in small increments of distance. Translating this to search marketing: it’s important to train the Google machine in a consistent and graduated manner, so it is able to learn from the data being fed to it. Too many variable changes (keywords, ads, budgets, bid strategy, etc.) at the same time will throw the machine back into learning mode and the dog will be running in circles until they learn to assess the data feed to start fetching again.
Each of the above steps when training your dog to fetch includes rewarding the right behaviors. You can reward Google with a well-performing landing page that helps drive conversions and, of course, by increasing your budget when the machine learning has been optimized. (But only do so in 20% increments to avoid triggering learning mode again.)
And now, not to be a mean girl, it’s time to make “fetch” happen.