October 22, 2015 by Gemma
Google Shopping is becoming an important channel for client acquisition and as an increasing amount of retailers turn to this method of advertising, it is more important than ever to have campaigns performing optimally.
Welcome to this presentation. I’m Gaël De Vos and I work with Twenga Solutions. We are a French company but we are acting on a global level, and basically what we do is provide solutions for advertisers to better manage their advertising budgets.
First a word about Twenga: We have 150 employees, mostly in our Paris office but we’ve also got offices in the UK, France, Germany, Spain, and Italy. Originally, we were a comparison shopping engine, so we’re very much used to integrating merchant catalogues. We currently index 400 million products in our databases.
We have clients in 15 countries, and in terms of services, we offer traffic acquisition solutions for Smart LEADS and Smart SEM, which is what I’ll be talking about this morning. Smart SEM is a suite of technologies aimed at helping advertisers on AdWords and Google Shopping.
This is the agenda for the presentation:
▪Challenges for advertisers
▪How is search marketing changing?
▪How we see going beyond traditional optimisations on Google Shopping
Plus a couple of case studies and of course, you’ll be free to ask questions at the end.
What challenges are search advertisers facing today?
In terms of challenges for advertisers today, this is probably stating the obvious, but these days it’s becoming more and more complex for retailers. More complex because there are lots of platforms and we can expect new platforms to appear over the next few months and years. There are also more formats, multiple touch points, multiple devices, and of course, multiple solutions providers. So, all of this is becoming really complicated to manage from marketers’ point of view.
Another challenge is what we can call the ‘scissor effect’. Most advertisers in the retail industry are seeing their margins decreasing because of the competition, and acquisition costs are also on the rise: because of this competition and because of this complexity.
In a nutshell, we believe that the traditional recipes for acquisition can no longer work and a sophisticated approach is almost mandatory for bigger retailers but also for smaller retailers, which is a challenge in itself.
How is Search Marketing changing?
Now, where does Search Marketing stand? As you know, Search Marketing spending is still growing, but this growth is slowing.
Mostly in Europe, it’s still quite healthy as you can see, but this spend increase is quite slow. This is the big picture, but if we zoom into some areas of search marketing, we can see that this is not quite the truth. The first dimension to consider is that retail is still growing quite healthily. The other industries are either flat or even going slightly down. You could say that they are in slight trouble.
In terms of market share, Google of course is the biggest actor ahead of other platforms such as Bing, Yahoo! and others. Google is the place to be for search advertisers.
Once again I’m kind of stating the obvious, but its good to have these figures in mind.
Now, inside Search Marketing, Google shopping is soaring. Over the last year we have seen that the available advertising space on Google has doubled. A key figure: The share of keywords for which Google shows Google Shopping results has grown from 7% to 16%, which is really impressive and explains the year on year growth of Google Shopping.
In a way, your traditional AdWords is stabilising, perhaps even going slightly down, and Google Shopping is seeing a high rise at the same time. This is of course a decision from Google, so Google is making more ad space, and more traffic available on Google shopping than it used to, and this is of course, taking some share from Google AdWords.
This can be shown in a standard search results page. More and more keywords are open to Shopping and what is very important, is that after talking with retailers, the general perception we have is that the quality of traffic on AdWords is slightly going down. So, this means the best quality traffic seems to be saved for Google Shopping.
This is shown in these metrics…these metrics are very impressive. They show a comparison between Google Shopping and Google AdWords. In terms of bounce rate, engagement, visit duration, page views, etc. All the metrics show that users are more engaged and more likely to convert through Google Shopping than from Google AdWords.
The future of Google Shopping
Now just a quick word about what’s next on Google. As you all know, Google is releasing new formats and new platforms almost continuously.
Firstly, Local Inventory Ads are going to be a very hot topic over the next year. In the US, they’re already prioritised before traditional ads, and of course, this will be a further challenge for advertisers, mostly in the battle between pure players / click & mortar and brick & mortar stores.
Then there’s a new feature, the ‘buy now’ button, which is coming very quickly on Google, firstly on mobile. In Google’s view, this will be a way to essentially own the whole shopping experience. So, this will also be a massive topic for advertisers this year.
In terms of competition this is important because it means that the available real estate is changing. With your traditional AdWords, plus Google Shopping, plus organic results, you may have enough space for 30 people to compete, whereas on these new formats, and as we’re moving fast towards mobile, the competition is increased and being on the top spot is even more important.
This explains why we decided to build a new platform for Google Shopping and of course, for Google AdWords too.
Going beyond traditional optimisations on Google Shopping
We believe that Google Shopping was quite simple a couple of years ago. All you had to do was upload your product feed on your Google Merchant Center, and then place your bids on the Google AdWords account to operate campaigns… and violà! At first ROI was pretty good and CPCs weren’t that high, so the competition wasn’t very hard.
All of this has changed. Now, in order to be successful on Google shopping, you need to focus on what we believe are key areas. Firstly make the best use of your products. Secondly, leverage all the data that you’re able to put into the system to the max. Of course, there’s also some Google Shopping know-how that is quite specific.
Making the best use of products
In terms of products: as probably all the people who run Google Shopping campaigns in this room know, Google Shopping allows you to bid at a very granular level. It means that you can literally place a bid for every product in your catalogue that’s uploaded on Google Shopping.
Even if you have 10,000 products or 100,000 products in your catalogue, theoretically you are able to bid at a product level, yet most advertisers bid at a higher level. What you see in lots of Google Shopping campaigns is the bid for a category, for instance, or perhaps for a category filtered by brand.
Perhaps you will have 50 very similar products, in the same category from the same brand that will have the exact same bid. This is because, for advertisers it’s a nightmare to fully leverage the possibility of bidding at a product level. You might say “OK I have these 5,000 products, what is the accurate bid of each and every one of these products? How do I compute this figure?”
Most advertisers just give up, and place bids at a higher level, but of course, common sense is that among a product category, product performance is very diverse. If you’re selling, I don’t know, trousers or shirts, you know that even among very similar products, one will be a top seller and another will not sell at all. We feel that there should really be a difference between all of these products’ bids.
This is an example from a campaign we are running currently for a French advertiser. They sell sports gear for people who run: stuff like shoes and running electronics.
The graph here I will explain later because it’s a bit too scientific, but the key point is shown here at the bottom of the screen. When we took over the campaign, these running t-shirts all had the same bid…because of the reasons I just explained.
After a couple of weeks, here are the bids we can see in Euros,(but of course it would work just the same in Pounds). You can see that between items that look very much alike, the bids we’re placing in the campaign are very different.
This can be seen at the category level, but also at a finer level, for instance for categories such as ‘The North Face running t-shirts’. They have perhaps 20 products that fall under this description and between these 20 products, we’re able to really understand that there are huge differences in terms of bidding.
This is very important because whenever a user types the keyword in Google, Google will choose amongst lots of products that can match the keyword. And by bidding at the product level, we’re in a way helping Google show products that are the most likely to sell. This is really a very strong factor in performance.
Leverage all data
In terms of data, how do we get to this very granular, or very deep knowledge of the catalogue and how can we compute this in campaigns? First we score products in the catalogue. This means that we monitor all that is happening on the website and we look at all the events. Instead of just focusing on sales, we, for instance, look at user behaviour on the product page.
This is a bit technical, but it explains that by gathering lots of data, we’re able to understand – okay, I have this category page: which products are the most visited? – By doing all of this, we can calculate a score for each product in the catalogue. Then we combine this with an approach that is made possible by Google, which is segmenting the audience.
Those in the room who are advertisers on Google probably already use or know about RLSAs– It’s an important part of Google. A couple of, well almost a year ago, Google launched this feature and made it possible for advertisers to switch from a model where you were bidding on keywords or products for Google Shopping to a model where you can combine bidding on keywords and bidding on audiences.
What is bidding on audiences? Basically you have the first approach to bidding on audiences. The basic principal is simple. For every visitor that comes to the website, you can understand things about the purchase intent of this particular visitor, and then you can put this user in a particular segment so that further on, whenever this user looks for another keyword on Google, you will be able to use this previous knowledge in order to make a better bidding decision.
For instance, a very simple user case is people who made it to the basket page. So they put an item in the basket, but they did not purchase. These guys are probably the sweet spot for retargeters for instance, and if one of these guys is looking for a product on Google, I can bid very high to get this user back on my website.
This is the standard user case for our listing on Google. But we believe we can go further. We can literally split the audience of the website in as many segments as relevant. In order to improve performance, so we can make use of an advertiser’s CRM data.
The simplest way to think about it is, for instance, prospects versus clients. Why not build segments of prospects and segments of clients? This means that later in your Google campaigns, you will be able, if you want, to generate new clients by overbidding on your prospects segments. And if you want really to lift your client base, to overbid, on the other hand, on your client segments. I’m sure you will have questions about this section after the presentation.
And we’re going even further. What we are doing now is what we call calculated segments. Basically that’s an algorithm – it’s a machine learning process – that takes the user, when the user lands on the website and the algorithm process says “OK, This user is very similar to this little bucket of users that I already have, so I will put the user in this segment”. And this way, we can build lots of segments. Why is it important to build lots of segments? It’s because each segment has a different conversion profile. We can have segments where we put users that have a very low intention to purchase. For these users, literally, it’s a waste of money to try to attract them in further campaigns on Google.
Other users are very ‘hot’ in terms of purchase intent. These users, of course, will be very interesting, and our goal will be to place the advertiser’s ad in the top position for these users.
All of this is summed up by this cycle. We connect as much data as possible onsite; we combine this with what we call dynamic user profiling, so to build these small bits of audience that can be operated separately. And all of this goes into our bidding algorithms so that the campaigns really can benefit from what we call ‘Smart Actionable Data’. ‘Actionable’ is because, really, we can have something you can incorporate in terms of marketing.
Who has the Google Shopping know-how?
What does it mean in terms of landscape on the market? With Google AdWords, things were simpler in a way. You had agencies (some of them very good), and you had technology providers (also, some of them very good) and basically, an advertiser would pick between or even manage the budgets in house.
Because of the complexity of Google Shopping, and this new approach that we believe is required to succeed on Google Shopping, there are different areas that need to be mastered to be successful. First, collect data, analytics and perhaps attribution issues. Second, bidding optimisation. Then there’s a third key area, which is feed management.
In terms of the market, the interesting thing is that lots of companies that help advertisers are good at feed management. Some of them have very good bidding tools and of course, there are a number of data collection or analytics tools on the market. But the question is how do you make all of this work together?
Should you have separate providers? How do you manage this complexity? This involves some key challenges. The biggest challenge, of course, is that to have good performance, a unified vision of all of this is really key.
This is what we believe we bring to the market. We are able to master these three key areas in order to deliver very good performance on Google Shopping to advertisers.
This is a case study from the merchant I mentioned earlier. They are selling premium running gear and equipment and when we took the account, the goal they assigned us was to boost sales, because they were quite happy with their cost of sales.
They had a cost of sales around 30%, so the ROI was in line with what they expected, but they thought, “hey, my campaign is still small and after a month it seems that I can’t go further.” We revamped the campaign entirely, and the sales almost tripled. With a cost of cales that remained at first stable, and then even went down.
So we have a nice testimonial. What does this gentleman say?
“We were looking to increase our sales on Google Shopping without compromising our CoS. Since activating Smart Feed – Smart Feed is the name of our Google Shopping management solution – we have seen a decrease in our costs and an increase in revenue.”
Now if we go deeper into the campaign, what did we do? We needed to do a lot of optimisation. When we took over the accounts, there were product rules all across the capital and the same max CPC (that’s basically the bid you place on Google) across all of the product groups.
We did what I explained earlier: we redid the structure so that we could really bid at the product level. In terms of audience, they were making use of RLSAs, but basically they had four or five lists (or segments), and they managed this – this is something that only Google AdWords professionals will understand – but they were using bid modifiers in order to change the bids. As a result, they were not really combining bidding on the product and bidding on the audience. So it was either one or the other.
What we did was really build a full structure with one segment per adgroup. Bids were decided at the most granular level and in the most accurate way.
Here are some of our clients, over a lot of demographics. This is our statement: What we believe is that leveraging all available data on a real-time basis to combine product-level bidding with hyper audience segmentation is key to success [on Google Shopping].
For those of you that are interested, we have a very nice white paper that explains all of this, and it’s available on our website.
Are there any questions in the room?
Thank you very much for your attention.
Learn more about the platform in our ultimate guide to Google Shopping
Cutting-edge Search Marketing and Product Ads solutions