Earlier this year, Google rolled out a new feature to let advertisers set different bid strategies for different devices (desktop, mobile, tablet). These device modifiers are one of the most important performance levers, but all too often they’re neglected by the campaign manager.
There are a number of different strategies for dealing with this new found device freedom, and we’d like to share the one that’s worked for us and our clients. Although we primarily look at Google Shopping, you can also apply most of the ideas to Search campaigns too.
What are Device Modifiers?
In June 2016, Google rolled out a long-awaited feature giving advertisers the ability to set bid modifiers for Desktop and Tablet in addition to Mobile.
That means that bids at the product group level can now be used as device-independent baseline bids. By setting modifiers, you can create an effective bid that takes device specific performance into account.
Device modifiers exist at both the campaign and ad group levels, allowing you to adapt the product group bid for a specific device from -100% (don’t serve ads on that device) to a +900%. If a modifier has been set for both the campaign and the ad group, the ad group modifier will trump the campaign. Which makes sense since the more specific modifiers at the ad group level should override their more general counterparts at the campaign level.
So, how do you decide how much to make your Device Modifier?
There is, however, one noticeable exception: If you set the device modifier to be a -100% at campaign level for a specific device, it won’t be served, regardless of the modifier on ad group level. This unleashes a whole new set of options – the ability to run device-specific campaigns for example.
Calculating your Device Modifier
Before you could separate Desktop and Tablet, you would calculate your product group (keyword) level bids based on both Desktop and Tablet conversion data (excluding Mobile). Then, you would compare their performance to Mobile and derive a bid modifier at the ad group or campaign level.
Everything was just fine with that approach until traffic started shifting towards mobile. At that point, advertisers wanted a way to use mobile conversions for calculating better product group bids. That way, you could use more data when making bid decisions at the product group level.
Here is an example:
By just looking at Desktop and Tablet performance, it’s unclear whether we hit a goldmine (with a ROAS of 49) or whether the performance was random – after all it’s only 47 clicks.
Should you bid up – and if so, how much – or leave the bid as is?
By including Mobile data, you can get a much better picture of conversion rates. The combined ROAS across all the devices is 12.5 – we can trust this number because it is derived from a lot more clicks. Having more data means we can be more confident when calculating a new bid.
Now we need to adjust the device modifiers to take into account individual device performance. To do so, let’s take look at ad group performance.
The simplest way is to divide the device ROAS by the average ROAS:
Based on this data, which device modifiers should you set?
- Desktop = 7.96 / 6.71 = 119%
- Mobile: 6.14 / 6.71 = 92%
Of course, there are other factors to take into account. For example, if your Impression share is already high, it may not make sense to bid up further to respect the Shopping S-Curve.
But what about Tablet?
Sometimes there just isn’t enough data
Even when you use mobile data to pad out your calculations, you can still run into the issue of not enough data. Depending on how fine grained your ad groups are, you may not have enough conversions on this level to make a good decision.
As a good rule of thumb, you shouldn’t make any decisions based on less than ten conversions. So in our first example (where Tablet only has 7 clicks), we simply do not have enough data to set a Device Modifier.
Let a tool help you out
As you can see, it’s pretty easy to start adjusting your Device Modifiers. However, it’s also easy to set your Device Modifiers once and then forget to adjust them based on new data. Automation tools like camato can provide SEA managers with a lot more comfort and sophistication when it comes to calculating and uploading device modifiers on a regular basis.