This past year a small group within our PPC department here at Natural Intelligence undertook an innovative automation side project that has amazingly resulted in an 8% boost to our gross profit. This is no small feat for a company with yearly revenues of more than $300 million.
UPDATE: We were selected as finalists for the 2019 Search Engine Land Awards for the Best Overall SEM Initiative!
Here’s how we did it:
As one of the 50 biggest paid search advertisers in the world, Natural Intelligence’s profits are tied intimately to the effectiveness of our PPC campaigns.
We are active in dozens of different verticals, and therefore we’ve had to take on dozens of PPC marketing analysts to bid for top positions on search engines.
One way of making PPC campaigns more successful is to know how different geographical areas perform and to adjust your keyword bidding in those geos accordingly.
Doing this manually requires an unreasonable amount of labor hours, and the more people you have carrying out your PPC bidding, the more you struggle with consistency and with maintaining a certain standard of excellence.
This was what we had in mind at Natural Intelligence when we made a decision last year to divert some of our resources from our day-to-day business to focus on an ambitious PPC automation project.
Two developers and 2 PPC data analysts worked on the project intermittently over a period of about 9 months. This was a side project that they carried out in addition to their regular duties.
Taking 4 of our most talented employees away from their daily business activities, even on a part-time basis, is not a decision to be taken lightly. However, we saw an opportunity for exponential growth—even if it meant stepping back from our immediate business goals—and we jumped on it.
This commitment to innovation paid off almost immediately.
How We Created an Automatic Geo-Targeting PPC Campaigns
Connecting to a Google API, we were able to create an automation that could cluster geographic areas that behave similarly, and automatically make adjustments to fit geo modifiers according to performance.
This allowed us to change bids in certain locations—be it countries, US states, or even to the granular level of zip code—and see how they perform.
For example, we could see if ads were converting x% better in New York, and increase our spend there automatically, in real-time, and decrease the spend in other areas.
When you work on global campaigns that target many countries, behavior changes are even more extreme than those between states, so such automatic adjustment is all the more important.
The advantage of automatic, super-granular geo-bidding proved itself in the numbers almost immediately.
We took some time to A/B test it because we were proposing major changes to campaigns with millions of dollars at stake. However, after some minor tweaks to early iterations, we saw the automation succeeding at a tremendous rate.
We started working on it last year, and already today it’s active on 85% of the company’s traffic and has significantly raised our efficiency and gross profit.
So what’s next?
In light of the consistent and sustained success of our geo-bidding project, we have expanded our PPC automation team to 10 people.
Since we’ve had such success automating by geo, we have turned the team’s efforts to automating by other demographic data, such as age, gender, days of the week, and hours of the day.
We are taking data from Google and third parties and clustering IPs by these demographic factors to see how they perform and how bidding can be optimized. Because we work in so many different areas, this requires analysis of a lot of historical data, going up to a year back.
This is an ongoing project that we are excited about and from which we expect to see a big performance boost once it goes through the same process of testing and perfecting that the geo-bidding automation project went through.
Conclusion
A small but dedicated team of people, with the support of a management team that values innovation, were able to make a major impact on our Google and Bing search campaigns and positively impact the company’s bottom line while outperforming off-the-shelf solutions like Kenshu and DoubleClick. The success of this project served as proof of the effectiveness of PPC automation and has led us to double our efforts to automate by demographic factors, in addition to geos.
Capture High-Intent Leads
