April 14, 2015
There are many different ways to structure Google Shopping campaigns. When beginning, I prefer to break out campaigns and/or ad groups by product types. Unless there is a reason otherwise, I’ll generally set bids at the product group level instead of the individual SKU, allowing the data to dictate which products convert better for the more generic queries. For example, a query of “oval coffee tables” could potentially trigger any of my oval coffee tables.
Along with product type campaigns, I’ll begin with an all products campaign to act as a catch-all. This campaign gets a lower bid than all other campaign product groups and isn’t supposed to accrue as many impressions. Its purpose is to show ads for products that might not have exposure in other campaigns. It can also double serve in the PLA units with an existing product group ad. As an example, two of my oval coffee tables may show, one from the product type campaign and the other from the all products.
On the flip side, I generally don’t begin right away with a top products campaign. The notion of the top products campaign is to create a campaign specifically for certain products. These items may:
- Be top sellers
- Have higher margins
- Have higher average order values
- Show better conversion rates
I believe in the concept, but I like to accrue enough data before these products go into their own campaign.
In one of my accounts, performance was strong, however, there were certain areas that weren’t as cost effective as they could be. For one, I did not have a top products campaign even though I had enough data to warrant one. Additionally, there were many products throughout the individual Shopping campaigns that were performing poorly. Some of these products had lower bids and several queries had been excluded over time, yet the fact remained that a new structure could improve performance.
I ended up taking a two-pronged approach. Along with creating a top products campaign I created mirror product type campaigns that only targeted the poor performing products. In the original campaigns, the poor performers were excluded. For example, if product A was featured in the mirror campaign, then it was excluded in the original product type campaign. Here’s a visual representation of the structure.
The product bid was then set much lower than it was in the original campaign, say $0.50 from $1. Theoretically, I could have excluded the products and allowed them to show in the all products campaign at a much lower bid. I didn’t want to take this route because I still wanted the volume. Even though these products weren’t converting on the last click, I knew they were helping in the overall process (via further searches and remarketing). Thus, these products received their own budgets while the product type campaigns had the poor performers removed.
The top products campaign encompassed the ten items with the highest conversion rates and lowest cost per conversions that met a conversion threshold. These products received a higher bid than all other product groups. They were also excluded within the existing campaigns so they would have a better chance to show for more generic queries.
Here is what click data looked like with 10 days worth of results after the addition of the new campaigns vs. the previous 10 days.
The most noticeable difference was the increase in both clicks and impressions. With the poor performers going into their own campaigns, budget was freed up in the existing product type campaigns. Additionally, the poor performing products now had lower bids, but were still accruing a fair amount of traffic. Overall CPC was also lower by $0.03. Even though we were bidding higher for our top products, the lower bidding for the poor performers brought the overall CPC down.
When looking at conversion metrics, cost per conversion and conversion rate are similar during the two time periods, but overall conversions were higher in the latter 10 days.
The additional conversions were in part due to the top products campaign. The fact that higher bids were placed on these products (and excluded from other campaigns) led to more conversions than when they were part of the product type campaigns. Also, of the 11 product type campaigns, 6 saw more conversions during the 3/28/15 – 4/6/15 time period. Of those 6 campaigns, 5 saw the most conversion volume. In other words, these 6 campaigns that saw higher conversions during the latter time period encompassed the majority of the overall Shopping conversion volume.
Also of note, and not unexpected, is that the poor performers campaigns didn’t see many conversions. Again, I was OK with these lack of conversions because I was looking to keep similar traffic volume. The benefit was that these clicks were less expensive.
Finally, let’s take a look at revenue.
In line with the conversion increase, we also saw a nice revenue increase. Both our return on ad spend and average order value increased as well. I mentioned earlier about product type campaigns getting their own budgets with the addition of the new poor performers campaigns. Another benefit was that the product type campaigns allowed for more impressions on many products with the poor performers excluded. In our top converting product type campaign during the 3/28/15 – 4/6/15 time period, average order value was nearly $17 higher than the previous period.
Even though the campaign changes have helped, I do want to caution that there is only 10 days worth of data in each time period. The trends look positive, however, a longer time period will truly determine if these updates have helped. But judging from the results so far, I’m encouraged by what I’m seeing. As all Shopping campaigns continue to accrue data, my plan is to add more products to the poor performers and top products campaigns.
What tactics have you tried to improve Google Shopping results? Leave your comments below.