“Here’s a song I wrote on a plane between Dallas and Austin…Goin’ to El Paso”…Waylon Jennings
I usually don’t take requests for doing blog posts but during a client pitch meeting, I spontaneously threw out a tactic that I’d implemented that really seemed to resonate with everyone in the room…so I’m going to share the process in detail here.
I currently have one client that uses Convertro for 3rd Party attribution…the client is a SAAS provider focusing primarily on the B to B marketplace with a Worldwide focus. Each client conversion has a lifetime value based upon the plan purchased combined with the location of the purchaser. My geo-bidding strategy had evolved to the point where for the US, I would Geo-bid to the MSA (Metro Area) level while for most other countries, I would Geo-bid to the State / Province level using Google’s definitions for Geography. However, I could previously only segment my Geo-bidding based upon the Paid Conversions without being able to consider either LTV or the proper attribution with the data available to me in Google.
When I first started using Convertro and focused on updating my Geo-bids, I defaulted to the same rules and strategies that I used in AdWords, knowing that since Convertro included attribution layered over the geo-specific data, utilizing this data helped me manage the account better. However, when I next revisited the Geo-bids and decided to dig deeper into the numbers, I discovered something that really caused me to take notice and change my Geo-bidding approach.
When I geo-bid across the entire US for any client, my default strategy is bidding to the MSA level while using State level bids (which corresponds to the State level “Unspecified” designation) as the “catchall” for visitors that don’t fall into one of the MSA’s. When using AdWords Conversion data, these conversion geo-segments dovetail nicely into a bidding strategy which can be implemented in a reasonably automatic fashion. In a typical conversion pattern, I see very wide CPA differences between MSA’s, though with few exceptions, I don’t make the bid modifiers larger than +30 or smaller than -30 since I want the geo-bids to act as a seasoning or a spice to the campaigns, but not take over the flavor of the dish (good or bad).
B to B SAAS customers don’t typically have the same geographic revenue distribution that one will see in ecommerce. For my client, within each Metro Area, there was an incredible amount of revenue clustering at the city level that could not properly be represented with an MSA Geo-bidding strategy. Therefore, I created a city bidding strategy with very divergent bid adjustments even though many of the cities were right next to each other and also, I felt very comfortable bidding to +- 60%…twice as high / low as before.
Without getting too deep into the demographics of my home state, I can tell you that each of the -60% adjustments represent either mostly residential suburbs or very industrial / blue collar towns that don’t map well to my client’s service offering. The +60% adjustments are places that have large white-collar technology footprints and ranking Portland at +35% is very high for a city its size, demonstrating that our being called the “Silicon Forest” isn’t just a title but is rooted in reality.
One issue to consider is that there isn’t a perfect match between Convertro’s Geo breakouts and Google’s. I believe Google has close to 12,000 US Cities, and even if I could get good data for all of them, Google won’t allow me to list all of them in one Campaign due to a predetermined location limit. However, I determined that I could map the biggest 4,000 cities or so with a 99%+ level of accuracy via Automatic matching (while easily fixing the outlier errors). Then, I could calculate a Geo-bid for the remaining cities and assigned it at the country level.
I was also able to bid Canada at the city level quite successfully. However, while Convertro provided excellent city level data for the entire world, I couldn’t find a clean way to map it to Google because I couldn’t find a way to mesh Converto’s data with the options Google gave me. Hopefully, that will change at some point.
It would be very difficult to tie account improvement to just this change since seasonality combined with many other account improvements are also impacting the client result. However, the number of non-brand conversions in Q3 will be 2x last year’s number, so we definitely are doing something right :.)