Tony is the Marketing Manager at Sales & Orders heading up our inbound marketing and advertising team. His latest adventure, though? Being a dad!
There are many schools of thought out there for how to successfully setup, structure, and manage Google Shopping campaigns. Some are incredibly simple in nature while others are substantially more complex.
Since the very beginning of Google Shopping, we have stood behind a single unified structure for the Google Shopping campaigns we create and manage for our retailers. That structure produces results, some astronomical. But for every good, great, and excellent campaign structure out there, there are also those structures that destroy performance and wreak havoc on conversions and returns. Today we are highlighting just that.
This Google Shopping campaign structure could be killing your returns, and it’s time to do something about it.
Many times, agencies and do-it-yourself store owners will group like products together by brand or category.
Already, this is where things can start to get messy. As an example, we have an instance where a retailer began using Custom Labels to further subdivide their top-level Brand Product Groups. They then subdivided again but by product type. This retailer also applied some form of custom naming to product groups, similar to how product SKUs and IDs are structured. As a result, their Shopping campaign winds up looking like this:
Within each product group (even when it got somewhat granular) there were multiple products housed. All of those products are then set at the same Max CPC bid as the group.
First of all, this has now become a bid and campaign management nightmare. The lack of true granularity winds up being very destructive to key performance indicators month-over-month. Be warned, the results you’re about to see are shocking:
In a month-over-month comparison for all Shopping campaigns this retailer manages for themselves, revenue, conversions, ROAS are all trending way down whereas costs continue to rise.
Unfortunately, this retailer’s campaign suffers from an all-too-common mistake of “Over-Structurization.” The overuse of multiple layers of subdivision coupled with a lack of full granularity and the complication of those subdivision categories has averaged them over a 20% decrease in Revenue (Conversion Value) and an unbelievable near 75% increase in Costs-To-Convert.
There’s Really No Fix
“Over-Structurization” is not only bad because of decreasing performance, but campaigns suffering from it are often lost causes. Due to the lack of granularity, there is really no way to determine which products are actually driving conversions and which are merely eating up spend without generating sales. This results in:
- Overbidding: The grouping of multiple products in a single bucket and being bid on at the same rate even though not all products will have the same performance.
- Underbidding: Almost just as bad as Overbidding, Underbidding can stop potentially strong Shopping ads from holding enough of an Impression Share to substantiate value.
- Loss of Control: Campaigns under these restrictions are often more than just cumbersome to manage. Rolled up data impedes the retailer/advertiser from controlling trends of positive and negative performance.
Shopping campaign structure is really all about getting it “right the first time.” If you are just getting started or are looking to test out a new structure, the key is to keep it simple.
You’ll want to ensure that you have the ability, at any given time, to spot the good, the bad, and the ugly. This is most easily accomplished by not over-subdividing your product groups. While you can continue to layer your structure such as first by Brand or Product Type, the secondary and tertiary levels of subdivision are what wind up causing issues.
Instead, consider using a rule of 2’s. If your first layer is by Brand then your second layer should be at full, ID-level granularity. The same goes for any first layer be they by Product Type, Category, or even Custom Label.
The simpler the better. The more granular the better as well.