You managed to set up your Google Shopping Campaigns and now you are closely monitoring their performance. Sales are coming in, so there is potential in this channel. But, maybe you are not happy with the returns. Or you want further improvements. This is where optimization comes into play and today we will dive deep into it.
As you have seen Google AdWords, and more specifically, Google Shopping campaigns have various settings that you can use to control every aspect of the campaigns. Given that the job of a Campaign Manager is to get the best Return On Ad Spend (ROAS) for their campaigns, one may as well look at these settings as optimization opportunities.
Let’s start with a very important principle:
“You cannot reach maximum performance on PPC if you do not have a solid campaigns structure.”
There are 2 main reasons (1) optimization granularity and (2) ease of reporting. A good campaign structure gives you the granularity necessary to do optimizations on lower-levels and thus get the advantage of the opportunities. You cannot optimize if you have an “All products” campaign - one-size-fits-all campaigns are simply doomed to fail. At the same time, though, you do not have to overdo the granularity part because it is easy for the whole account to become unmanageable (unless you use a Google Shopping campaigns management software). So there is a fine line on how to do this. You will find the optimum for your account with experimentation.
Ease of reporting, on the other hand, is equally important because if you do not have it, monitoring and analyzing performance becomes way too tiresome for you, which is something we see far too often not only on PPC but also with all Analytics, including Web Analytics. If reporting is made easy and practical, you have more to benefit from. In any other case, trying to figure out what to do with all the numbers you find is a real challenge. This is why we are having the revolution of self-service reporting.
“Low hanging fruits”
If you have one or more campaigns with a limited number of ad groups, breaking these further will give you significant benefits. It is quite easy actually. First, break your targeting either by Brand or by Product Category. Then, choose the one that will give you 20-30 options and start with this. If you feel up to a bit more work, break them onto both Brand and Category.
Now that you have your campaigns broken down by either Brand or Category, the next step is to break them down even further, by:
* Price groups
* Products on sale
* Top-performing SKUs
* Margin groups
Note: It does not make sense to break things down if you do not use this granularity for optimizations (eg. on bidding, ad-scheduling, negative keywords, etc.)
Provided that you have selected the best KPI for your campaigns (most possibly this is ROAS), the next step is to consider the North Star Metric, which, in simple terms, is the value you deliver to your clients.[a][b][c] One of the primary actions you can do to reach your North Star is bidding.
“Low hanging fruits”
Machine Learning (ML) and Artificial Intelligence (AI) are the hot topics these days across all industries, including online advertising. Google is investing tons of resources to apply these models to improve all their products. Of course, AdWords, their main revenue-driver, could not escape this. The last post on AdWords blog called “Grow your business faster with machine learning“ is no coincidence.
In regards to Google Shopping Campaigns and bidding, Google’s efforts have given us the “Target ROAS” bidding method. You can read more about how to setup Target ROAS, but there is also a summary to get the idea. When using this bidding method, you do not set a CPC bid rather than a target performance in terms of ROAS, meaning how many $ you get for every $1 invested in the ads. Keep in mind that this will not be the actual performance; it is just a target. In other words, this is not a method of “affiliate advertising” with a fixed performance model. It still remains a CPC model under the hood.
To launch the Target ROAS bidding method, you need to have conversion tracking and conversions recording enabled. Give the model some time to learn (minimum of 2 weeks), and then the performance will start to grow.
We know for a fact that many advertisers use the bidding method mentioned above. So, if you want to outperform your competitors, more advanced activities should be implemented. Going into “manual bidding”, always with specific bidding strategies based on your business and on your goals, is key.
Manual bidding will not make any significant difference if your campaign structure is fairly basic. For more sophisticated campaign structures (i.e. you have followed the “advanced optimisations” on the “Campaign Structure” section above), though, you have a targeting granularity that allows you to bid smarter.
However, “Target ROAS” would probably not work as it needs at least 200 clicks for each Product Group to get enough data to perform.In this case, we recommend doing a periodic optimisation based on ROAS performance using a decision tree diagram. In regards to frequency, you can start with weekly bidding, but if your budget is lower than $2k, you could go with bi-weekly bidding. To do the bidding, build a decision tree diagram and then follow it for each of your Product Groups.
When you find the best decision tree that works for you, you can automate this process using AdWords Scripts or even the AdWords API.
Another thing you can do to outperform your competitors is to include other data sources or points into the logic of your bidding strategy. The most common data point is gross margins, although you could also do inventory or price levels compared with the competition. Gross margin will help you avoid under-optimizing your campaigns. How? Lets us assume that you have 2 products with the same price (eg $100), but not equally valuable for you. Because the first product is an mp3 player, which makes you $20 from each sale while the second one is a laptop case that earns you $50 from each sale. See how looking solely at ROAS under-optimises your campaigns? The most granular way to get over this issue is by replacing the actual conversion value with gross margin. The easiest method is to label products based on groups of similar gross margins and then apply a bid modifier on these.
A very important but frequently under-looked dimension of performance is the location. Say, you are advertising to get sales from the USA market, yet, the performance in each state is different. Potential reasons on why you are not getting the same performance across all locations are:
* Different locations are inhabited by people with different income levels and this is reflected into the average conversion value on your campaigns.
* Different locations are inhabited by people with different familiarity levels with your products/brand or even the generic category and this is reflected into the conversion rate of your campaigns.
* Different locations are targeted by a different number of advertisers (your competitors) and this is reflected into the actual CPCs.
As you can understand there are many factors that play a crucial role on the consistency of your performance across various locations. The good news is that Google has, once again, given us the way to get round such issues. For starters, you can look at the performance by going into the “Predefined Reports” (see image below) or by creating your own reports on “Reports”.
When you get the data on the performance on different locations, use “Location bid modifiers” to adjust your bids on these locations and maximize your performance.
“Low hanging fruits”
The first step would be to set bid modifiers for your top-3 locations. If you are in the USA, these would be your top-3 states, whereas if you target other countries, such as the UK you should go for top-3 cities. Check the performance of these areas on the reports, look at the ROAS performance of these areas in comparison to your average ROAS performance, and apply a positive or negative bid modifier. Apply these modifiers at the campaign level on all your Shopping Campaigns and you are done!
If you are willing to invest more energy and time, you will be able to get even better performance. To achieve that:
* Set bid modifiers for your top-10 locations (if you have enough data) or
* Apply different location modifiers into different campaigns advertising different products.
The second method could work if you have broken down your products by category/brand rather than by price, which would make things more difficult for you. We rely everything on the fact that people in different locations buy different things while also taking into account the existence of various physical retail stores in different areas, which affect people’s buying decisions greatly.
The “era of mobile” has begun. In most countries, mobile queries have surpassed desktop queries. The same behavior is observed in consumers’ purchasing habits. Given the diversity of mobile and the varying buying behaviors, it is paramount to find ways to adjust your campaigns to these highly-valuable devices.
Before, there was no way we could control the devices we could target. Recently, though, Google has given us back the control for all device types (desktop, mobile, tablet) and we should take the opportunity to adjust our bids based on performance. To do this, there are two options (1) monitor performance using the “Predefined reports” or (2) monitor performance using the reporting breakdown by device on all standard reports (image below). Then, all is left would be to apply bid modifiers to the Campaign or AdGroup level.
“Low hanging fruits”
Look at the overall performance of your Shopping campaigns and apply bids at the Campaign level for all your Shopping Campaigns. If, for example, the ROAS of your “mobile” clicks is 50% lower than “desktop”, reduce the bid by 50%. See how this goes and adjust. In an ideal world, all device types have the same ROAS - this will mean you are doing the ideal device budget allocation.
Up to this point, you have already started with Campaign-level bid modifiers and have seen significant improvements on the ROAS. Now, it is time to amplify results by applying:
* Different campaign modifiers onto different campaigns.
* Device bid modifiers at the AdGroup level for high-volume AdGroups.
Remarketing lists for search ads
We wholeheartedly love RLSA!
“Remarketing lists for search ads” or RLSA is a feature Google has given us to be able to target, bid and get reports from different audience segments for our search advertising. That way, we can bid $1 to people that have never visited our website and $2 on people that have visited our store, added products to cart but have yet not converted. Why go down that road? Because these people are doing what we call comparison shopping and we want our store to be top of mind now that they are closer to actually buy. You can see more about this concept at our presentation on Google Shopping optimization.
“Low hanging fruits”
If you want to start or do not have advanced conversion tracking set up on your store, you can create only 2 lists, namely “Visitors” and “Converters”, and apply a +30% and +100% bid modifiers respectively. To do this, you need to have conversion tracking enabled (to avoid bigger problems). After you begin to get data from your campaigns, monitor the performance of these two lists and adjust the bids accordingly.
Create more lists with their own respective bids. Potential types of lists are:
* Standard lists of Google Analytics enhanced e-commerce (i.e. product viewers, cart abandoners, etc.).
* CRM lists - These include lists of past customers that you have uploaded to Google AdWords. Go further by applying customer segmentation on your past customers.
* Algorithmic lists - Build Google Analytics lists based on your top products or top categories that get updated as your sales data change (need custom engineering for this).
As previously mentioned, not all locations (also days and times) have the same performance, considering that people behave/shop/spend their time differently on weekdays vs the weekends, and in the morning hours versus the evening. As with locations, you can look at your current performance using “Predefined Reports” and then apply bid modifiers.
“Low hanging fruits”
The first thing you can do is apply bid modifiers on different days of the week (eg Monday, Saturday, etc). Depending on the nature of the products of your store, chances are the performance per day of the week will be different. Check the performance from all your Shopping Campaigns and apply the modifiers that are based on the performance of each day. If you have not launched Google Shopping yet or do not have much data, you can look at other Search Campaigns (even your Search Brand campaigns) as these usually follow the same trends.
Tip: Although you might see really big differences on the ROAS of different days (especially if you do not have much data), resist the temptation to apply big bid modifiers - this will most likely decrease your ROAS. It is very important to have a fixed daily budget and having significantly different bid levels will not work out well for you.
But there are more ideas on how you can use data into the different time dimensions to apply more granular bids. Examples are:
* Hour groups - Make 2-3 hour groups and apply different modifiers on these. An example could be 8-17, 18-24, 1-7 but look at your own data.
* Combination of hour/day - This is more advanced and you need to have a bigger account with more data to get statistically significant differences on these groups. But, if you have, the returns would really make sense!
* Per campaign - As with location, you can have different modifiers for different campaigns.
* Dynamic budget scheduling - The idea here is to not just adjust bid levels but also the actual daily budget on different days.
Although search is mainly “keyword-based” advertising, this is not the case with Google Shopping Campaigns. On Google Shopping, we use our products as targets and Google promote them on relevant queries. This means that although we do not have “positive keywords” that we can use to funnel relevant traffic, we do have “negative keywords” that we can use to make an advanced campaign structure.
Having said that, “negative keywords” are also used to eliminate wasted ad spend on queries that are of no value to us. By removing our ads from bad queries we are able to reduce our overall cost (and thus increase ROAS) and also potentially increase CTR, which will consequently allow our campaigns to perform better into the ad auction and give us lower CPCs.
“Low hanging fruits”
Start by reviewing the “Search Terms” report for your Shopping campaigns and then negate the queries that (1) do not make business sense to you (i.e. competitor-related queries that you do not want to advertise on) or (2) get a big number of clicks but no conversions.
You can do the optimization mentioned before every 2 weeks. After the first 3-4 optimizations, most bad queries will be blocked. Plus, you will most likely have got your account into a good shape. But, do not stop there; take it a step further.
More advanced activities you can do using negative keywords are:
* Use negative keywords to make a “query intent” campaign structure targeting different generic categories and product terms.
* Create a shared negative list with keywords that you want to apply to all your Shopping Campaigns. This would also be really helpful for all your new campaigns.
What are you waiting for?
Now you have a really good overview of the different optimization actions, you put the tips in the “low hanging fruits” sections into action and start reaping the benefits of your labor even today. You can literally make significant optimizations in the first 2 hours (device, location, weekday modifiers & negating search terms) that will give your account a big boost. Do it now or book some time on your calendar in the next 2 days.