Our DNA is machine learning and complex software development. We have
embraced Google Ads management after an “aha" moment that occurred while
helping a client improve their Google Ads campaign.
Chris Bergstrom graduated with a computer engineering degree from Simon Fraser University in Canada. He founded Bergstrom Marketing in 2005. Over 6 years, he grew the company to having 40 employees in India and six employees in the USA. All was good until Chris noticed he was getting tired more and more often. He was diagnosed with leukaemia, and spent two years getting chemo.
With the help of his sister Susie who is a graphic designer, Chris managed to keep the company running while undergoing chemo. Once his strength returned, he decided to focus on his first love – math and computer programming. He turned this fascination of math and computer programming towards solving the fundamental challenge of using artificial intelligence and machine learning to set keyword bids and bid adjustments in Adwords campaigns.
We manually downloaded the data from a clients’ Google Ads campaign, and used our spreadsheets to calculate new keywords bids.The results of our first trials we mixed. It turns out that calculating keyword bids and bid adjustments is deceptively complex. The heart of the challenge is to determine how much to trust limited data.
Our first major success came two months into the project. At that time, we were able to achieve a Cost Per Action (CPA) of $18-22 when setting the bids manually. When we used the spreadsheets, our CPA was $15. At this point, we realized we were on to something. Over the next two years, we made steady improvements to our algorithms.
Our last major leap forward was when we were able to incorporate machine learning systems such as Tensor Flow into our system. TensorFlow is a machine learning engine designed by Google that powers Google’s own search results.
The result is that after we started using TensorFlow, the average CPA dropped to $6.30. We often compare our system to Google Ads intelligent bidding systems. When we used Google’s Target CPA bidding strategy, the CPA increased from to $10.30 and the number of leads dried up.
So, why is our system better then what Google offers?
Google has a conflict of interest. Machine learning is a series of algorithms that “minimize loss". When applied to Google Ads, “loss" is “spending too much on a visitor" or “giving Google too much money". Here is where Google has a conflict of interest. Google benefits when people spend more money than they should to target a specific visitor. If Google helps you eliminate “loss" in your Google Ads campaign, then Google will make less money.
Google’s automated bidding strategies appear to be designed to help make people feel comfortable spending money on Google Ads. They do a fairly good job, but they’re not designed to be perfect. They’re designed to give people confidence to spend money on Google Ads. Our artificial intelligence is designed with you, the client in mind. We will analyze any Adwords campaign and provide suggestions that are better than any human can.We also focus on complex web development, app development and we do SEO for a select group of clients.
So, that’s our story. It’s been a wild 15 years, but we’re enjoying living in the mountains, writing software and helping people save money on their Google Ads campaigns.