As I wrote this, I debated whether to gate the content by requiring some sort of email address or information exchange. I think this is immensely valuable (almost 5,000 words!!), as it is much of what I’ve learned in years of managing Shopping Ads.
However, I decided to go ahead and let the content stand on its own. I would ask you one favor, if you find value in this would you be so kind as to share it with others? Email it, tweet it, however you want to! Thank you, and enjoy the show…
Shopping Ads is my favorite part of Paid Search.
For anyone who works in Ecommerce paid search, having a working knowledge of Shopping Ads is essential to keeping up with the competition.
This article is for the purpose of giving you that working knowledge, and we’re going to do that by focusing on these 3 components to Shopping Ads:
- Feed Setup & Management
- Campaign Setup
- Ongoing Optimizations
Feed Setup & Management
In order for you to manage Shopping Ads well, you need to be aware of how they work. Shopping Ads are automated, in the sense that Google or Bing are not directly matching a search query to keywords. They are taking the data you have submitted in your merchant feed, and they are matching that to search queries.
Because of this, understanding shopping feeds is crucial to success in Shopping Ads. If you struggle to understand the concept of the feed, think of it in terms of a restaurant.
Your (1) server (the data feed solution) submits your order
to the (2) chef (the Merchant Center for Google or Bing), who then finalizes it and sends it
to the (3) runner (AdWords) who makes sure it gets presented
to the (4) proper table (SERP).
For feeds, you need to have all aspects of the process working correctly in order for your feed to be sent.
Data Feed Solution
The Data Feed Solution is how you take your product data and submit it to Google Merchant Center. There are options here, so be aware of the best fit for your particular needs.
Small Company / Low Number of SKUs – if you fall into the category of having a handful of SKUs, you can actually just use Google Sheets to integrate directly with Google Merchant Center. This has the benefit of being simple to start and keep track of, however it’s probably not a great option if you have a lot of products (there are size limitations to this method), or if you have constant feed changes.
Large Company – if you are a good-sized company with developers to spare, then you may want to consider creating and sending your own XML file directly into Google Merchant Center. If you do go down this route, make sure you aware of all the complexities of requirements to get your products approved, as well as ongoing changes approved.
Anyone – thankfully if you don’t fall into either of those camps, there are options out there. 3rd Party Feed Providers range in size and pricing and it all depends on what you need to be done. Companies like Feedonomics are a full-service solution that can be a big benefit if you want their assistance in helping optimize, as well as send your feed. Companies like GoDataFeed tend to be a little more affordable, but also tend to have less versatility and customer assistance based on that pricing. Companies like ChannelAdvisor or CommerceHub tend to have the benefit of offering other services as well (marketplaces, for instance).
Whatever you choose, be aware that you will still need to work with the feed provider to get your data to them in some format.
The Data Feed Itself
Now that you know how to get the Data Feed up, what are you actually sending to Google Merchant Center? The important thing to note about the Data Feed is that it’s important to get your product information accurate, but get as much as possible into the feed. There are required and recommended fields, and it’s tempting to only add in the required fields.
However, keep in mind that Google & Bing utilize your data feed to match to search queries, so the more (accurate) data you get, the more you can potentially match up to.
One helpful thing to be aware of when considering Bing Shopping, is that Bing accepts the Google feed directly in their Merchant Center. Importing this can save you time rather than creating a completely separate feed… unless you want to curate products uniquely for each channel.
Once your feed method has been selected, you will need to send it to Google or Bing Merchant Center. These are free accounts that you can create (and verify). Ensure you fill out the requirements in business settings, tax, and shipping… otherwise they’re not difficult to get set up! Create a feed and pull from your feed provider and you’re on your way to getting Shopping Ads on AdWords or Bing Ads!
Please note, Google Merchant Center is a unique entity, but Bing Merchant Center is currently within the Bing UI only in the “Tools” tab.
As a tip, when you first push up a feed, it can take 24-72 hours in both Google and Bing for the feed to be approved, so be patient and don’t wait until you needed the products up yesterday to do this.
The final step, is to link your Google Merchant Center account with AdWords. From my perspective, Bing makes more sense here since Bing Merchant Center is already integrated with your Bing Ads account. However with Google, make sure you link GMC and AdWords so that your products are eligible to display.
You will need to link your AdWords account in the Google Merchant Center settings here:
You can view this link in AdWords in the settings section:
Once you got the feed connected and products sent (make sure to fix any product disapprovals!), you need to set up your Shopping Campaigns in AdWords.
We could spend hours and hours of discussion on this, but we don’t have time for that so I’m going to have to do less discussing, and more detailing. Here is what I have found to be key principles to be aware of for setting up a successful Shopping Ads campaign strategy.
Principle 1 – Campaigns by Category
First, don’t just stick with a single campaign in your account, trusting the search engines to match everything up perfectly. This limits you in many ways, but primarily in bidding. Since you set bids based upon product groups, and however you have segmented those product groups out, you are in essence setting one bid for scores of queries of varying levels of intent. We’ll speak more on this in the next point.
Initially, I would recommend segmenting out your catalogue based upon natural categories based on how your customers search, or on other factors such as product cost or success. It’s crucial to grasp the idea of controlling how you can bid as you consider setting up your Shopping campaigns.
For example, in a client with an apparel focus and high volume of brand loyalty, it may make sense to separate your campaigns by Top Brands, or at the very least by ad group so you can control bids individually.
The thing to remember about segmenting categories or brands by Campaign or Ad Group, is that you can better control negative keywords and thus push queries in between campaigns or ad groups.
Example (over-simplified): let’s say that the query [blue athletic shoes] converts well in your Adidas ad group, but not in your Nike ad group. Because you have these segmented by ad group, you can exclude that query in Nike. This will prevent it from wasting spend, and push those impressions to the Adidas ad group. The one thing I’ll note with this to keep in mind, however, is to remember that be aware that you can have multiple ads showing for one Google Shopping SERP, so be careful not to over-exclude… you may be keeping competitors from showing and that may be in your favor! Be smart about excluding, but be willing to try it.
There are lots of other ways to consider segmenting campaigns, you can segment all Top Products into a single campaign with a High Priority setting, and then put the rest of your products in the lower Tiers to ensure your Top Selling products always get first shot at impressions, regardless of bids.
Whatever you do, figure out what works best for your account and get building out beyond one campaign and one ad group.
Principle 2 – Segment by Queries
I have written extensively on this elsewhere, so I won’t go too far down the rabbit hole here. The basic idea is that keywords are a form of communication, and demonstrate buying intent. This means in Search we can assign value (a CPC bid) based upon what a person is communicating through their Search Query.
You would likely be willing to bid higher on “I want to buy an awesome Force FX lightsaber with free shipping” then you would “saber” because there is buying intent in the first query.
However, in Shopping Ads, you bid on a product basis. This means that the $0.50 bid you set for that lightsaber product, is bidding $0.50 on “I want to buy an awesome Force FX lightsaber with free shipping” as well as $0.50 on “saber”. Starting to see the problem?
We need to be able to bid separately on query buying intent, and thankfully there is a way to do that. Using the Priority settings structure, as well as negative queries, you can pull unique query groupings into unique campaigns so you can bid separately on them.
If you are interested in learning more about this, please go read my How-To post for Search Engine Land where I spent 4000+ words telling you exactly how to do this!
Principle 3 – Segment Devices
This is a newer way of segmentation I’ve been experiencing with, and one I’m increasingly excited about. It’s a commonly known fact in PPC that unique devices convert uniquely. You may find drastically different Conversion Rates for mobile phone users than you do for desktop users. In the past, you could only set a bid modifier at a campaign or ad group level for devices. This has a few problems.
- This means that your bid for every product in that campaign or ad group is adjusted for that device modifier, regardless of individual performance.
- This means that you are limited to bulk modifiers unless you pull unique products into unique ad groups and utilize bid modifiers at an ad group level, but then your bidding and management is beginning to get overly complicated.
- This means you can’t easily see and set unique bids for individual product groups on individual devices.
- This means you can’t easily compare multiple attribution models at a device level.
I have found increasing success in clients by keeping our primary campaign structures less complicated, and then integrating Device segmentation on at a campaign level.
This looks something like this at a campaign level:
- Shopping – Top 50 Products – Desktop – US
- Shopping – Top 50 Products – Tablet – US
- Shopping – Top 50 Products – Mobile – US
- Shopping – All Products – Desktop – US
- Shopping – All Products – Tablet – US
- Shopping – All Products – Mobile – US
- Shopping – RLSA – Desktop – US
- Shopping – RLSA – Tablet – US
- Shopping – RLSA – Mobile – US
This allows you to tightly control and respond with bidding to what the data is saying, rather than just setting general bid modifiers.
A Note of Caution: One important thing to be aware of, is the idea of cross device attribution. With Shopping, you need to understand how users journey to purchase across multiple devices before being too quick to kill bidding across any single device. You may discover many people first enter on a Mobile device and then purchase later.
You can analyze this in Google Analytics by keeping track of how your Shopping campaigns do across multiple Attribution models. You may find mobile does better on a first click basis, which suggests you should be cautious in making bid decisions that pull back on mobile when they may be starting a lot of your customer journeys.
Another final (free) recommendation, would be to consider on your landing page how to make the connection for users between mobile and desktop. Perhaps you become more aggressive in suggesting wishlists or “save for later” login options so you can better understand at least some of your customers and their journeys across multiple devices.
Phew, you made it this far. Great job! You’re basically a Shopping Ads pro now. Your feeds are set up, your campaigns are live, now it’s time to manage things.
The first thing you need to be aware of with Shopping Ads, is that you do need to upkeep them. While there is a level of automation to them, there are also many, many ways they can go wrong. Here are the top things you should focus on to keep your Shopping campaigns humming and profitable.
Just as in Search Ads, query mining in Shopping Ads campaigns is essential. What is query mining? I am referring to the process of going through your Search Terms report and analyzing it for keywords.
The thing with Shopping, is that you can find a lot of benefit in finding both negative, as well as positive keywords.
What you will find is that Google and Bing can play fast and loose with how closely they want to match your products up to user queries. Because of this, you will also see queries of varying quality flex in and out of your account.
As an example, just this past week I discovered the exact match term of [amazon] in an account that has been running for a decade (and I have been running personally for nearly two years now). I wasn’t selling amazon action figures or stuffed animals of creatures found in the Amazon basin. It was just a terrible match.
I recommend getting into the account regularly to mine through your search terms. How regularly? It will depend on the size of the client, but I think you should be going through your Shopping campaign queries at least once per week… and that’s if you have a good structure and are already on top of things!
Did you know you can use Shopping as a keyword research tool? Because Google and Bing are matching up your products with searches their algorithms believe to be right, they will likely find keywords you hadn’t discovered in your research phase. Keep an eye on your queries in Shopping for this reason as well, then find those top terms and built them into Search ads so you can control the SERP in text and Shopping!
Product Curation is another option you have for ongoing optimization, but I would advise caution here.
By Product Curation, I simply mean determining to only send certain products in your feed. There are a few reasons you might consider this. You may have more products than you have budget. If you have $5K per month to spend on Shopping Ads, but 200K products… you probably need to curate what you send. Or, you may have been running Shopping Ads for awhile now and have some conversion history you want to utilize in making a decision to pare back on products not selling.
The benefit to curating, is that you can focus your data, optimization efforts, and bidding efforts on the few products that do drive the most overall sales. If you have account history, you may consider starting to curate your feed by eliminating product IDs that have never accrued any impressions, after which you could try eliminating product IDs that have not accrued clicks. On the other hand, what if those are good products that just need title or description tweaks? The plot thickens, and proceed cautiously.
Another reason for caution here, is because Shopping Ads tend to be of higher importance towards the Top or Middle of the funnel. The last thing you want to do, is to make a decision on products with little obviously tracked revenue, that ended up being great products at starting the customer journey. You’ll strangle revenue flowing in from other sources (Brand campaigns, Organic, Direct, Social, etc) and never know that it was related to those products… so caution is needed.
A significant aspect of Shopping Ads is the product feed, so it stands to reason that making smart adjustments to that feed would be an important part of optimization. There are several things that can be done to the feed, but when the rubber meets the road, we are busy marketers and I think there are a few easy wins to focus your energies on to get the biggest bang for your buck.
The primary thing to note in feed adjustments, is that testing is notoriously difficult with Shopping Ads since you can’t run the test simultaneously. What you will have to do, is analyze what impact normal seasonal trends (from previous years) have on the timeframes you’re running your test, and then look at KPIs such as impression share or CTR rather than volume only. I would also recommend limiting bid adjustments during your test period so you minimize variables.
Titles & Descriptions
When it comes to titles and descriptions, we must reset our brains a little and think back to old-school SEO. While this doesn’t technically mean keyword stuffing, it does kind of mean keyword stuffing.
Identify the terms your searches use to purchase your products and make sure those terms get into titles and descriptions. Here is where you can do some testing. Try running titles for a certain product brand with primary descriptive keywords at the front of the title, and then run a test for another month.
For example, you might try these variations of titles in testing:
- Green Running Shoes Nike Free RN 2018 – Mens Size 10
- Nike Free RN 2018 Green Running Shoes – Mens Size 10
- Nike Free RN Mens Size 10 Green Running Shoes
Keep in mind for title testing, that you can utilize Google Feed Rules to test without having to change your feed. This can be extremely helpful in controlling your own tests as the PPC Manager without having to make a big production of getting the dev team or feed provider to assist in your test until you are ready to move forward on a larger scale.
Research Recommendation: How do you know what keywords to include in your title? From my experience, it’s easy to do some basic experiments with Keyword Planner. This may tell you to test putting “lego” before or after “millennium falcon” in your product (example below) based on what is more commonly searched for in terms of search volume.
Price is a crucial aspect to Shopping Ads and one that is increasingly being revealed as having a significant effect on revenue growth, as well as potentially on the algorithm itself (see this slideshare of a presentation by Andreas Reiffen, beginning at Slide 11).
This makes sense of course, since people are often drawn to the lowest price when given multiple options (all else being equal). This can be quite problematic for you if your store tends to be consistently higher priced (without a significant other value you provide) on Google Shopping. Worse yet, you may discover that you still get curiosity clicks followed by a bounce as that visitor heads back to the SERP and buys from your lower-priced competitor. So you ended up losing the sale AND still paying for the click.
The issue here, is that PPCers rarely directly control the product pricing. I say rarely sarcastically, because the real word is more likely “never”. There are many, many factors that factor into product pricing and you can’t control all of them… but you can have a say.
- I think the action for pricing is two-fold:
Be aware of this as a major impact on profitability if rapid changes changes happen with little adjustments made in management. In other words, if a key product suddenly becomes less profitable or drops in sales, yet nothing major has changed strategically, that is a great time for a competitor pricing analysis. More often than not, you will find your competition has dropped prices on you which impacted your sales. Keep your radar on at all times.
- Be communicative with your management team regarding the crucial role Pricing plays in Shopping. Unwilling to lower pricing? Maybe it’s time for a shift in strategy since Shopping is so crucially about price, and maybe you can be the one to investigate and identify the need for that. Are there ways your business needs to think about pricing better in 2018 marketing and how it relates to Shopping Ads? It is you who will need to be aware of, and communicate the need for Shopping Ads to have a place at the pricing discussion table.
The final thing to note in terms of quick-easy wins in feed optimization is image testing. Since the image is the crucial visual part of a Shopping Ad, getting this right could benefit your traffic and sales.
For various image testing suggestions and tips, I would send you to the presentation at SMX West that Purna Virji (of Bing Ads) gave: Thats Hot! 3 Shopping Campaign Tips That Would Make Paris Hilton Proud
The key tip to bring out for images in Shopping Ads, is to make them stand out. There are many ways you can do this (though make sure they are in line with Shopping Ad image policies), but the number one most common would be testing both product images and model images.
In the screenshot below, the Hello Molly Fashion store may find themselves getting traffic and interest not simply because model images are always better, but because they are different than all the others.
Marketing is often about distraction and catching attention, and there is a strong case to be made for testing images that disrupt your current SERPs.
Finally, we get to the last section in this mega-long article. The final ongoing optimization in Shopping Ads that we’ll note is the topic of bid adjustments.
When I speak of making smart bid adjustments, I am thinking of two different categories. Thinking through adjustments on a micro, or product group, level and thinking about bid adjustments on a macro, or account wide, level. Both are important and I’ll detail them below.
In thinking through how bid adjustments should be made on a micro basis, we turn to the product group level. The product group level is broken up as specific or broad as you want it to be, however it is on the product group level where we make the actual bid decisions that decides what will happen in the SERP!
In thinking through micro-bid adjustments, it’s important to note that bid automation is an increasingly valuable thing in 2018 (time of writing this post), yet there are still not a ton of affordable options that work for small to mid tier Shopping clients.
AdWords does offer several, eCPC, Target CPA, as well as Target ROAS. I have utilized Target ROAS before but I found it struggled to keep up with dramatic changes in the market (such as Black Friday). In my opinion, it’s still coming to age. Target CPA is less valuable for ecommerce marketers since ROAS tends to be more important to us than how much we spend per sale. eCPC may have the most benefit of assisting us in bid efficiencies with Shopping Ads so I definitely recommend testing it as long as you have a good number of conversions. In my opinion, 30 in 30 days isn’t enough for accuracy, and AdWords themselves states that the more segmented your account, the harder it is for automated bidding to work… so be aware of that!
Because of these things, I’m going to focus less on automated bidding strategies in this section, and more on creating manual bid rules based on automated strategies. If you think about it, good automated rules are algorithmic. That means if we can think of some smart, repeatable ways to think smartly about bidding, then we can call those bid rules and feel proud of ourselves.
What I do in these instances, is to utilize Saved Filters in the AdWords UI (note, you can’t do this in Bing UI at the product group level, but Bing Ads have noted this will be coming soon).
What you want to do is to think through the ways you normally make decisions on how to determine whether a product group is or is not profitable. Then save that filter and voila, you have an easy bidding rule you can check once a week or whatever preset to your rules.
Two rule examples you would create a product group filters in the AdWords UI:
Profitable Products with more market share to be gained.
- Past 7 Days (give enough time to make a good decision)
- All product groups spending over $100 (ignore product groups without enough data to make good decisions)
- Search Impression Share lower than 50% (there is more market share for them to be gained)
- Tracked ROAS above your target (let’s say 300%)
- Over 1 Sale (you want to make sure this is more than just catching products with accidental, big one-time sale anomalies)
Poor performing Products that are bid too high.
- Past 30 Days
- All product groups spending over $200
- Search Impression Share higher than 90% (they control impressions for much of the market)
- Avg CPCs higher than average account (these are products you’re likely over-bidding)
- ROAS under your target amount
Each of those is one example of how you can create “automated” bidding rules that you have saved as filters that takes multiple, smart things into account in making an informed decision.
If you actually do want to consider automating something like this affordably, my recommendation would be to investigate Optmyzr. We utilize them to assist us in account management at ZATO, and they have a rule function where you can build rules like the above and tell them to automate them to run on a schedule.
Ok, so you thought through micro-bidding… but how do you think through bidding on an account-wide basis? Here we are leaving hard data (somewhat), and journeying into the unknown.
It’s essential that you be aware of the idea of Attribution. If you are unfamiliar with that notion, read these two articles I wrote on it first:
In attribution, we determine which channel gets the credit for the sale. Of course, the user journey is so much more complex than this, but there also needs to be some way of getting directional ideas from our attribution tracking.
The reason Shopping Ads gets messy fast with attribution, is because Shopping Ads target all stages of the funnel. This is in part, why it is essential to segment your marketing campaigns according to different stages of the funnel based on query intent (see above). However, once you’ve done that, I’d like to encourage you to go one step further.
That is, as you analyze and make micro-bidding adjustments, that you do so keeping the funnel in mind. My recommendation is something like this:
- Pull generic, non-brand queries into a “top of funnel” campaign, and then utilize a first click attribution model to identify impact on overall sales.
- Pull brand, or mid-journey queries into a “middle of funnel” campaign, and utilize a time decay or position based model to analyze success.
- Pull SKU, or unique buying intent queries into a “bottom of funnel” campaign, and utilize a last click model to determine success and set micro-bid adjustments.
As an aside, this is also why I prefer pulling RLSA audiences into unique, RLSA campaigns. Since these people are nearly always mid to bottom funnel, we see issues with lumping them into our top of funnel campaigns for bidding purposes (they’ll always be undervalued in first click).
If we don’t think better about making bid decisions according to where the Google Shopping query is in the funnel, then we will be massively under-valuing Google Shopping which will only lend itself to hurting overall sales.
You made it! Congratulations. I hope you had as much fun as I did, and I hope you are better prepared to attack your Shopping campaigns!
On the other hand, if you find that you are struggling with Shopping Ads and the above sounded nice and complicated, then please feel free to reach out to as here at ZATO as we specialize in Shopping Ads and are interested in finding great, long-term partnerships we can win with!
Peace out, and if any questions on this, please connect with me and ask on Twitter or LinkedIn: