When analyzing campaign performance by geography, it isn’t uncommon to find that some locations are more competitive than others. Sometimes a simple bid modifier does the trick but certain situations call for more. Recently, we isolated high volume, high CPA geographies into separate campaigns in an effort to improve lead volume and CPA across said geographies in addition to the remaining geos. The case study below details our results.

The Background

In this particular test, the client is an e-commerce platform with no physical locations. The client is unaffected by seasonality.

In looking for expansion opportunities, we found that the client had a lot of unrealized potential in Texas. The Texas campaigns were capped because the CPA was inefficient but we knew there was more volume to be had, if we could improve the CPA. When we took a look at the highest conversion-volume cities we, no surprise, found some of Texas’ largest metros: Dallas, Houston and Texas. Unfortunately, their CPAs were too high. We also found that many other cities were converting at a lower cost, albeit with lower conversion volume.

The Hypotheses

We decided to create separate campaigns for Dallas, Houston and Texas, hypothesizing that:

1. If we separated out the most expensive metros, we could open up budgets to get more sales through the less expensive regions.
2. If we separated out the most expensive metros, we could have better insight into which keywords were working in those cities and which weren’t. We could cut out low performers, and make the major metros more profitable to open up budgets there, as well.

The Action

With the context of the situation and our hypotheses in mind, we duplicated the previous campaigns to create separate campaigns for Dallas, Houston and Austin. We kept the original campaign, excluding the aforementioned locations.

As data accumulated in our new campaigns, we focused on weeding out underperforming keywords and adjusting time of day and mobile modifiers for each campaign. We adjusted bids across all four campaigns to find the sweet spot between position and return.

Analyzing The Data

We spent the better part of a calendar month restructuring the account, so for comparison we pulled conversion data from the calendar month prior and the calendar month following the restructure. Both months have the same number of days.

Our hypotheses were confirmed once we analyzed the data post-restructure.

Calendar Month Prior To The Restructure

Calendar Month Post-Restructure

All three major metros showed improvement in both conversion volume and CPA. Within these three cities, on average, conversions increased by 145 (52%) and CPA decreased by $17.20 (28%).

Moreover, conversions increased throughout the lower volume geographies by 68 (37%) and the average CPA decreased by $19.61 (30%).

Conclusion

In conclusion, our hypotheses were correct. We were able to increase conversion volume and efficiency by separating out high-volume, high-cost geographies into separate campaigns.

Since the restructure is still fairly fresh, we expect the campaigns to continue to improve as they build history and especially as we implement additional optimization strategies and tests, including but not limited to localized ad copy and sitelinks.

Have you had success in separating out campaigns by geography? We’d love to hear about your experiences in the comments!