Automation driven by improvements in artificial intelligence and machine learning will bring significant changes to how we manage pay-per-click (PPC) in 2018 and beyond.
There’s still a role for humans to play, one of which is to use what we’ve learned from years of experience to bring an account as close to perfection as possible before letting the machines take it from there.
Shopping ad performance may appear to be largely dependent on how well the automated algorithms work, with little room for account managers to optimize things. This is because the targeting is done automatically by Google, based on whatever products are in the merchant feed and how those relate to shoppers’ queries.
But there are still ways for account managers to tweak things to move the needle in the right direction.
In today’s post, I’ll explain seven ways to improve the performance of shopping ads.
Eliminate ambiguous and expensive queries
I remember several years ago, one of our customers wrote about how his shopping ads were showing for the letter “M.” That’s right, just the letter “M!”
That was the pinnacle of poor relevance. But since then, the same innovations have brought us to the point where computers can manage aspects of accounts as well as humans. Those same innovations have dramatically improved the accuracy of Google’s automated targeting that selects which products from a merchant feed are relevant to a user’s query.
When I look through queries we use, they seem to be mostly relevant. In the handful of cases where they aren’t, the issue is usually an ambiguous query rather than a complete fail in targeting.
For example, in this test account, we see queries for “vegas gold” and “royal blue” leading to ads for apparel in those colors.
While it’s possible that Google is using the context of the user’s previous queries to determine that they are in fact shopping for clothing, we see no conversions from these queries, so they make for good negative exact match keywords.
This simple optimization can instantly save wasted money, but it should be reserved for queries with a very low chance of ever converting or driving awareness for your store in the early shopping stages.
Improve query matching
Over time, as advertisers have gotten more advanced with their shopping campaign structures and Google has gotten better at matching products to relevant queries, a more frequent issue we see is that the query is relevant to several products in a feed, but the ad serving system doesn’t pick the best one to show in the ad.
Here’s an example: Say we’re selling t-shirts for a variety of brands. When a user searches an upper- or mid-funnel keyword (one that is relevant but not specific), like “yellow shirts,” many brands sold by the advertiser could be a good match, and Google gets to pick which brand t-shirt to show.
Unfortunately, as you can see from my screen shot, they don’t always pick the one with the best performance.
There are two ad groups with traffic for the query “yellow long sleeve shirt,” but the click-through rate (CTR), conversions and return on advertising spending (ROAS) are all better when the ad for a yellow t-shirt from the brand Gildan is shown rather than one for Anvil.
What we need to do is filter rather than eliminate the query. When excluding, we’re saying there is no value whatsoever in that query (If that is true, then exclude).
When filtering, we are identifying where different queries are in terms of estimated value and setting unique bids on them based upon our perception of that value. The recommended optimization, in this case, is to add a negative keyword in the lower-performing ad group to help steer more impressions to the product that drives more profit.
An important point to note is that this technique requires the use of a lot of ad groups, since that is the lowest structural level in AdWords at which we can add negative keywords.
I advocate for as much granularity as possible, including putting every item in a separate product group so that bidding can be made more flexible and optimization techniques like the one described here become possible. This is called a GRoup of Individual Products or a GRIP structure:
Keep your AdWords synced to your feed
Once advertisers deploy more advanced account structures, with more levels in their product group splits, they may find the need arises to spend more time keeping their AdWords structure synced to the data in the feed.
For example, if you group products by brand, and then start to carry a new brand in your feed, AdWords won’t simply add the necessary new product group or ad group. This has to be done manually by the advertiser.
Advertisers avoid having to stay on top of the need for these changes by including “everything else” catch-all product groups in their structure. While this approach ensures that new products are eligible to trigger ads, it doesn’t give advertisers the level of control they should have to manage proper bids and negative keywords. We recommend spending some time with the boring but crucial task of keeping the account structure in sync with what has changed in the merchant feed.
There’s no ad to write, but you can still optimize
There are fundamentally two ways to optimize: the lazy way and the right way.
The reality is that many advertisers choose the lazy way because they have to make tradeoffs, and doing the right thing can be very time-consuming without the proper tools, scripts and automation in place.
For example, when looking for shopping queries that aren’t performing well, you could find several that are relevant to what the advertiser sells, but for whatever reason aren’t converting. The quick optimization technique is to cut the wasted spend by adding negative keywords.
But what if it were possible to improve the performance of those relevant queries so you could get the extra sales? For that, we have to work a bit harder, and we could optimize the data in the feed. What follows are three feed-related optimizations.
1. Titles and images are your ads — optimize them
The first feed optimization is to fix titles and images. Perhaps the image doesn’t look great, it’s too similar to those from other advertisers so it doesn’t stand out, or it doesn’t put your product in the best light or simply confuses the user.
Here’s an example of an unusual image for a query, “buy a Chromebook.”
The ASUS Chromebook is shown with a visual representation to indicate it can go from laptop to tablet. But a user who’s unfamiliar with the concept of a “convertible laptop” may be confused by this. On the other hand, the unusual image could draw more clicks. It’s a huge element of the ad that is worth testing.
When it comes to the title, in addition to including all important attributes, the optimization can focus on the order in which various attributes are mentioned.
As Google recommends:
2. Your price is a key driver of the ad’s performance
Another reason a relevant query may be underperforming is that the price is not competitive. This may manifest itself in getting lots of queries but a low CTR when users are turned off by the price.
In a previous article, I suggested how it may be better to change a product’s price than the product group’s bid to achieve profitability. Online shoppers are price sensitive and even being slightly higher priced for the same product could have a big impact on conversions.
Here is a webinar my company did with Search Engine Land columnist Andreas Reiffen recently on reverse engineering Google Shopping.
One of the best ways to deploy this optimization technique is to use Google’s Product suggestions report which will tell you if your price is not competitive.
3. Your ad is not complete without extensions
Just like you wouldn’t dream of calling an optimization complete without trying all available ad extensions for search ads, the same goes for shopping ads. It’s just that the available extensions are different.
To fully optimize a shopping ad, be sure to try all of these features and programs:
- A special offer submitted through the merchant promotions feed.
- Product ratings.
- Google customer reviews.
- Local inventory ads.
Optimize your site
Finally, if a lot of people are making it to your site but not buying, don’t just eliminate traffic with negative keywords — fix whatever ails the site.
You can use Google Analytics to find where customers are dropping out of the conversion process. The Goal Flow report in Google Analytics can be particularly telling about where users are having issues. This report can even be segmented, making it easy to see if the same problem you have with paid traffic exists with organic referrals.
Also, make sure your site is optimized for mobile. In late March, Google officially started rolling out mobile-first indexing, another indicator that mobile is where the game is at these days.
Jon Diorio, group product manager for Google AdWords, hinted at the recent Mobile World Congress that mobile page speed would become a bigger ranking and Quality Score factor because usability studies show that slow sites are detrimental to the user experience and cause a decrease in conversion rate. Jon said:
Conclusion
Shopping ads may seem pretty simple and automated if you have mostly managed keyword campaigns before. But despite a lack of keywords and ads to manage in shopping campaigns, there are many ways humans can help Google’s automation go in the right direction.
There are plenty of ways to boost your shopping ads — from managing negatives to choosing the right structure; from sculpting queries to optimizing the feed; and finally, optimizing the last steps of the consumer’s journey as they visit your site on a variety of devices.
Contributing authors are invited to create content for Search Engine Land and are chosen for their expertise and contribution to the search community. Our contributors work under the oversight of the editorial staff and contributions are checked for quality and relevance to our readers. The opinions they express are their own.
