I’m obsessed with Google. I love the data, the control, and the never ending changes. Imagine: An account has a search campaigns that is three years old and performing as desired. You have not made any changes outside of small data-based optimizations. The client decides they want to pivot strategy. Should you make changes to the existing campaign or launch a brand new one? In some situations, it may be more advantageous to start a new campaign but it’s important to consider the overall goal. Training an algorithm, especially in the context of machine learning, involves a series of steps and consideration. Below are some of the factors I evaluate when deciding what is the best move.
Should you refresh or start fresh?
Choosing when to launch a new campaign is like deciding whether to renovate your old house or build a brand new one. Before making any changes, consider how the proposed changes will impact the campaigns in context to the desired performance. What are you trying to accomplish? This helps lead you in determining strategy. Are you looking to tweak it for better performance or overhaul the whole thing?
How much data have you acquired?
Consider the amount of data you need to retrain. One key factor to think about is the amount of data you have. If your dataset’s on the slim side, it might be smarter to start from scratch. How much data do you have and how much do you need to retrain your campaign effectively is ultimately up for you to determine?
How much history do you have? How relevant is the data?
This determines how long you may/may not have to enter a retraining period. If the new strategy, conversion actions, audiences and other settings are similar – then retraining could be a real option. Think about how long it will take to retrain your campaign to achieve what you want. Sometimes, it’s more time-efficient to start fresh rather than trying to redirect the campaign on a completely different strategy.
Leverage the data you have to determine the best approach
Data is like fuel for the retraining engine. You should have enough of it, and it should be relevant to your new goals for it to work in optimizing to the new goals. The decision to start new or recycle campaigns should be based on the data, the state of the existing structure, and your vision for the future. The specific steps and tools used may vary depending on the type of algorithm and the problem you’re trying to solve. The journey of training algorithms is a collaboration between human ingenuity and computational power, with humans defining the tasks, providing data, and ensuring ethical considerations are met. Adapt the process to suit your particular use case and available resources.
If you’re ready to take your Google Ads to the next level or have questions about campaign performance, reach out to us at Hive Digital. One of our PPC experts are happy to perform an audit on your account and come up with a custom marketing plan designed for your business goals.