Posted by & filed under Transcripts.

At Marketing Festival, last year, Martin Roettgerding Head SEM at Bloofusion, discussed about Google Shopping Campaigns and how to go about creating an effective bid strategy.

You can view the video here

Transcript

Thank you. Good morning. It’s really really great; it’s really great honour, great pleasure to be here and to be able to talk to you about Taking Google Shopping to The Next Level. I come from a German agency, where we have a lot of shopping, a lot of retailers among our clients and so, we do a lot of Google Shopping. And for starters, I brought to you a very little case study from one of these clients, who has been managing their Google Shopping operations for, since Google Shopping came out, like 3 years ago in Germany.

And this is the revenue curve for Google Shopping for this client. It’s at the beginning of last year and as you can see, they start rather slow in the year and this client always does okay over the summer, but you know, as a retailer, Christmas is always the best. So at Christmas, the curve of course went up. And then, you know, year again started slow, and around Easter, that was when we took things to the next level. And what we did and how you can do as well is what I like to tell you in the next thirty or so minutes.

Before I do this, let me quickly introduce the topic. “What is Google Shopping? How does it work? What are we talking about?” We are talking about these things, or if you live here, they look more like this. So we are talking about product listing ads mainly on the Google search results, but also on the shopping section of Google as well! And the way that works is, it always starts with a retailer who has a lot of products. And they have a lot of product data as well, so there is a lot to know about these products. For example, every product has a title and a price and a brand and a color, for example and some other things.

And to advertise these products on Google Shopping, they have to provide some of this data in a certain form to Google and they do so by basically uploading the huge table with the product data, in most cases, and they upload this to Google Merchant Center. And Google Merchant Center lets the service, where we can manage these product data feeds, we can see if there is a problem, make some settings and so on.

And then we connect this to Google Adwords, and in Google Adwords, we have Shopping Campaigns. Now at the heart of Shopping Campaigns, there are product groups and that is what we do in Google Adwords. So basically, we always start with a group or products and we can leave it at that, then we can run Google Shopping, but it is certainly not very sophisticated. So we can go ahead and sub-divide such product groups.

For example, we could sub-divide this one into different brands. We could say, “Okay, we have Brand No.1 and Brand No.2,” and then Google will always automatically add a group for everything else. Now we can further sub-divide this. For example, we could say, “In Brand No.1, we have Product No.1 and Product No.2, and again everything else, and the reason we are doing this is because we want to bid on those things.” Now for example, we might want to say, “Whenever someone clicks on the ad for Product No.1, we’re willing to pay 20 cents for that click.” Now it’s important that we don’t bid on keywords here. We bid on products. Google will take care of keywords for us, because whenever someone is looking for anything where Google says, “Okay, that’s relevant in connection to this product,” Google may show the ad.

Now for those two up here, we are bidding on single product. We don’t have to do this; we can, the other one, we can also bid on groups of multiple products. And above all, we are completely free how to build our product structures, our product group structure, I am sorry. So you can do this however you like, however suits your needs. You don’t have to bid on each product individually, but you can do if you want and so, this is a very crude example that you, as I said, you are completely free.

The other thing we can do in Google Adwords is negative keywords, so we can look what did people type into Google, to see our shopping ads and we can see if we like those and for example, we may see something like, “Cheap Guitar” and we may not like that query because well, we don’t have any cheap guitars or anything or may be it just doesn’t work, so we can go ahead and use negative keywords to just block this. And that is it in a nutshell. That is how Google Shopping works.

So we upload some product data to Merchant Centre that gets linked to Google Adwords, there we build our product group structure and may be use some negative keywords. And there are two basic areas where we can optimize Google Shopping – the first one is Feed Optimization. So we can look at our product data and make sure that Titles are in order or that we have sorted every product into the right Google Product Category and so on. The other thing is we can do Bid Management. So we can structure our campaigns, bid on these products in any way we like and that is something we can do basically all day.

And well, these are the two areas we can optimize and if this was, in the other conference, we would probably be talking about one of these two, but, you know, when I got the speaker guidelines for this conference, they had only said, “We are not a conference for losers,” so we are not going to talk about one of those. Instead, I’d like to show you the third one and it’s something like a new layer that you probably didn’t know about before and your competitors don’t know about and hopefully it will help you take it to the next level as well!

Okay. Now, before I do this, I actually bought you something. Because I am, you know, this is all digital marketing, it’s all very…well, theoretical, it’s all digital and ladies and gentlemen, may I present you, this is an actual product. So what I brought you, ‘Running Shoes’! And not just some “Running shoes’; I brought you some “ASICS running shoes.” And if you look very closely, this is…these are the “ASICS Gel 1170 running shoes.” So I am not sure if you’ve written this down already, but I did, so no worries!

Now we have these three ways to describe this product. [“running shoes”, “asics running shoes”, “asics gel 1170 running shoes”] And that is not usually our perspective as online marketers, but this is what people might type into Google and Google might show the ad for this product. Now that is not our perspective. Our perspective is more like this, the other set of data and well, if we look in Google Adwords, we don’t even have this complete data set there or we may only have the id of this product, so we only see product 1, 2, 3. Not very great!

And as marketers, we are bidding on this product; so for example, we may bid 40 cents on this product. But that also means we are bidding 40 cents on the term “Running Shoes” and it means we are bidding 40 cents on the term “ASICS Running Shoes” and of course on the term “ASICS Gel 1170 Running Shoes.” Now, if this was a keyword campaign, this would be a strange set-up, because you probably wouldn’t bid the same on each of those very different queries.

So well, in this case, we can’t avoid this yet because, in Google Shopping, we are bidding on products, not on search queries. But actually we wanted to understand our shopping queries a little better and so we did, well, some digging and took a closer look at those search queries. And this was all the analysis we did, and it was rather simple, because we just exported every data, every search query we had, put them into a spreadsheet and just counted the number of words. And then we wanted to know how valuable different kinds of queries were. And when I say valuable, I mean one specific statistic and that is the revenue per click.

It is not conversion rate, it’s not average order value, it’s not total revenue, it’s revenue per click and for the simple reason, because we are in the business of buying clicks. So if we buy clicks and we bid on clicks, then of course we want to know how much do we get per click. And in this case, the picture just looked like this. So basically the longer the query, the more valuable it is, the more money we made per click. So one word query, in this case was only, was half as much as a 6 word query, in this particular case, for this particular client.

And of course we also looked at the other side. Again, the same kind of analysis, but we also wanted to know how much do we pay per click, so we looked at our CPC’s and the picture was the opposite. So the shorter queries were actually the most expensive ones, and the longer queries were a little cheaper, and if you combine those two, it looks a little strange, because one of those goes up, revenue per click, the longer the better and the other one goes down, because the longer, the cheaper.

So this is kind of a contradiction. But if we go back to our example product, it might make sense, because right here we have a very short query ‘Running Shoes’; it’s also a very general query. So whoever types, you know, ‘Running Shoes’ and will see the ad for this product, probably doesn’t mean this product and the probability that they are actually going to buy this product is rather low. At the same time, there are so many running shoes out there and so many other retailers who are bidding on running shoes and basically bidding on this term, that means there is a lot of competition for the term ‘Running Shoes’ and a lot of competition, of course, means high CPCs.

On the other hand, we have this rather long query, the “ASICS Gel 1170 Running Shoes” and that is so specific that basically only fits this shoe, so whoever is typing in such a long and specific query is looking for specific shoe and the probability that they are going to buy this shoe is of course much higher. At the same time, not everyone has this product, so there is less competition and the click is cheaper.

Now, at first, we will set aside at finding this path. Actually, there were so many exceptions to this rule, meaning we found some queries that were rather short, but still very valuable. And others were rather long, but rather worthless, so we were looking for better patterns. And one pattern we found was the brand preference in the query. So what I mean is this. Someone’s looking for “ASICS Running Shoes” or the “ASICS Gel 1170 Running Shoes”; they already decided on the brand they want, and we found that queries like that were more valuable.

This is the first client we ever did this for, and on average, they make 4 euros and 8 cents per click on their shopping ads. Now if we divide those queries into queries that do contain a brand preference and queries that do not, the picture looks like this. Queries that do contain a brand preference are worth high above average, while the rest is far below average. And in this case, it’s 3 to 1. Queries that do contain a brand preference were 3 times as valuable in this case. Now we found this across our entire client base. It’s not always 3 to 1; they may be closer together or further apart for some clients, but this works for everyone.

Okay, again, this explains a lot, but there was still one other pattern that we found and that was the product reference. The reference, in this case, for the “Gel 1170″; whoever is looking for this is looking for exactly this shoe. So again, we did the same analysis for this client and we saw that those queries were again much higher above average and as you can see, the rest is below average, but only slightly and that gives you a clue that simply there aren’t that many search queries out there that do contain reference to a specific product. But still, if they are there, they are worth much more.

Now the complete picture is this. We have our queries that do contain a product reference, very valuable; then we have queries that do contain a brand preference, but no product reference; still considerably significantly above average, and we have all of the other queries that are far below average. And looking at this, wouldn’t it be nice if we were able to bid differently on these different kinds of queries? Wouldn’t it be nice if we were able to bid high on high value terms to just get those and still try to get the lower value queries, but bid appropriately? So use a low bid for those queries and at the same time, wouldn’t it be nice if our competitors didn’t have this ability and would still bid according to the average value per click, meaning the 4 euros and 8 cents here? So they would basically overbid on the low value terms and underbid on the other terms.

And actually, that is possible, and I am going to show you how, and I am going to show you how to construct shopping keywords. The goal basically is to have some queries go into one campaign and to have other queries go into another campaign and then we want to use some bids for some queries and other bids for other queries. That is what we are trying to accomplish. And we can do so, because there is one cool thing that came with shopping campaigns, and that is ‘campaign priorities”; they work like that. Basically, someone is looking for something, a product. You have this product, but you have it not only in Campaign No.1, but also in Campaign No.2 and Campaign No.2 has twice the bid, so we have a conflict here. But from text campaigns or keyword campaigns, you probably know that the higher bids usually win, so Campaign No.2’s bid would be used most of the time.

Not so in shopping campaigns, because in shopping campaigns, we have priorities and that means we can just say, “Campaign No.1 has a higher priority than Campaign No.2 and so Campaign No.1 gets to use it’s bid for the ad auction.” It doesn’t mean that the product will always show, but we will use this bid and then we will show, depending on what everyone else is bidding and the search query, how relevant they think it is and everything. And we can use this to separate our search queries.

So let us start with a non-brand query. Again, we have a query and in this case, it is ‘Blue Running Shoes”; there is not brand preference in there. Again we have the same product in 2 campaigns and we have one with a high and one with a low priority. Again we have this conflict and again Campaign No.1 wins. So basically here, Campaign No.1 will always win, so Campaign No.1 gets all those search queries. Now when we have a branded search query, it works like this. Again, we have the search query, “ASICS Running Shoes”; again we have those 2 campaigns. Now everything else goes to Campaign No.1, so we have to make sure this query goes to Campaign No.2, despite the lower priority.

And this is really really simple, because we can just use the negative keywords to block this query from Campaign No.1. We just exclude the keyword “ASICS” and we make sure that all of those search queries go to Campaign No.1 exclusively. So there is no conflict and Campaign No.2 gets to use the higher bid, just for this kind of query. Okay, now we can use this to build a complete strategy around this. Of course, this is not just some weird trick that works for little campaigns; of course you can just do this on the large scale as well!

Again, the goal is to have all of our branded queries go to a brands campaign, and to have all of our product-specific queries go to the products campaign. And of course, everything else goes to the, I call that the “rest campaign.” Now product-specific queries for a product’s campaign, this is kind of optional because not everyone has product-specific search queries. For example, if you sell “Summer Dresses”; no one’s looking for the “Red Summer Dress Model No.123X5.” That is just not how it works, so in many cases, there are no product-specific keywords, so you will just have to do without a products campaign, but you can still divide into brand and non-brand queries.

To do this, you will need to gather your keywords. So first, your brand keywords. That is pretty easy because you have those in your Product Data Feed anyway. Now you need to provide the brand for each product to Google and you can just go, look at your Feed, copy these and you are done. Well, actually you are not quite done, because you should go over these brands and basically clean them up. Cleaning up means you may have a brand like “Tommy Hilfiger”; now few people are looking for something like a “Tommy Hilfiger” shirt, but they are more likely to look for a “Hilfiger” shirt, so make sure you have the part people are actually looking for, and in this case, use “Hilfiger” as a negative keyword. Or you may have brands like “Sammies by Samsonite”; no one’s going to look for that. That’s some kind of suitcase, and people are more likely to either look for a “Sammy Suitcase,” so you need that or a “Samsonite” suitcase, so please use those two.

Then for your product-specific keywords, getting those can be a little more tricky, but if you are in luck, you can get them as well from your Product Data Feed because there is a field called the “Manufacturer Part Number” and if you are lucky, that is the same number, I think, a designation that people are actually typing into Google. So for example, if you are selling notebooks, laptops, then they have a cryptic, Model No. Like 123 / abc something like that and that is what people are actually looking for. Or if you are selling some parts for replacement, anything, often times there is a Model No. that people are typing into Google.

Now you don’t always have this and you don’t always have this in your MPN field, so it can be more tricky. So for example, you may have to go to your Title, look through your Titles and see if there is a certain pattern. For example, all of your laptops may have the Model No. at the end of the Title, so you could extract it from there or you may have to do it manually, but to be honest, if it is that complicated, rather don’t do it; just separate brands and the rest and start with that and may be look into this later on.

Okay, now let us look at the campaign structure. First we need a campaign for products and that gets a low priority. Then we need a campaign for the brands and that gets a medium priority. And again, a campaign for all of the rest, with a high priority. Now internally, these campaigns can look however you like. You can build your product groups and do your bidding inside those campaigns in any way you like, because that is a nice thing about this strategy. It works like a layer on top of everything. So those don’t get into the way of each other; just do it however you like.

But basically when you are bidding…I am sorry, first thing is negative keywords, of course. We need to exclude the brand keywords in the campaign for the rest and we need to exclude the product-specific keywords in the campaigns for brands and everything else. And then for bids, you would usually bid fairly low in the campaign for everything else, for the rest. You would bid high on brand terms and you would bid very high on product-specific terms. Now as for the budget for these campaigns, please, please, please use a shared budget.

Because if you don’t and one of these campaigns run out of budget, for example, the rest campaign has no budget any more, then there is a problem that all of your traffic for the rest campaign has to go to the brands campaign next, so the system breaks down. So just use a shared budget and you won’t have this problem. Now there is one more point to this, if you think that is theoretical. If you don’t want to change everything, just based on some presentation, you can of course just build up these campaigns from your old structure, from your old campaigns that you have, and just keep the old bids. And then you will start to see different results for these different campaigns, and then you can just adapt your bids to that. So that is kind of the way to play it safe, I guess.

Okay, let me sum up how this is supposed to work and how it actually does work. Let us say someone is looking for something product-specific, and this is supposed to go to this product’s campaign, right? Like “We have this product in this campaign”; but of course we have the same product in the other campaigns as well! But in this case, someone is looking for, again, the “ASICS GT 1170” and we have blocked this ASICS term in the other campaign, so it can’t go there; there is no conflict, the product campaign for this query is all there is. So this campaign gets to use it’s bid for this query.

Now if there is a branded query, like “ASICS Running Shoes”, this is supposed to go into the brands campaign. While it can’t go to the rest campaign, because it’s blocked there, but it is not blocked in the products campaign, so there is of course a conflict here between products and brands campaign. But the products, the brands campaign has the higher priority because it has medium priority compared to low priority and that means the brands campaign wins in this case and gets to use its own bid for this branded query.

And of course, there is the campaign for everything else, and we only have the query that does not contain either a brand preference or a reference to a specific product. This will go into this campaign, because while it is not blocked anywhere, sure, we have this conflict between all three of these campaigns, but since the rest campaign has the highest priority, it of course wins and gets to use the lowest bid and well, that is how it is supposed to work and so we have our queries all divided high value of queries into the products campaign, low value into the rest and in the middle, there is a brands campaign.

So this is what we did for this client, and that was what we did Easter this year and well, they had a very good summer! Basically, they tripled their revenues and well, the client was happy, we were happy and I hope you are too! Thank you.

 

Leave a Reply

Your email address will not be published. Required fields are marked *