Google: Audiences for Search Advertising
Although Google is incredibly complex in terms of a search platform, the ultimate goal is for it to be extremely simple for the user – and to listen to questions. Director of Performance at Google Matt Bush focuses on the three principles of search marketing:
Be personal: think about the context of that person and give them the information they really want
Be actionable: make sure that person can actually do something with that ad and can make a decision quickly
Be measured: connect those customer journeys on and offline.
This is my fifth year at Digital Journey's. Good to see my fellow co-founder of Digital Journey's, Dave Coplin, in the corner. He works for Microsoft. You'll hear from him later. Google and Microsoft can actually get on. We can be friends.
Okay, so who's been to Tokyo? Anyone been to Tokyo? What an amazing city. I was there last week for the first time ever. I was actually with Rob Pierre. I'm still a little bit jetlagged. I got back at the weekend. It might be because I drank a bit too much sake, but either way it was a fantastic trip. And the one thing that stood out for me, probably more than anything else, was the friendliness of the people and the way they treat their customers. So they treat their customers like royalty. Genuinely, I felt special every time I went into a store.
So for example, you go into Starbucks and you get a thank you on your cup. In the UK, you go to Starbucks...there's no one from Starbucks here, is there? You go in the UK to Starbucks and they write, spell your name wrongly on purpose, I'm convinced. "Miranda. Miranda's cup." It's like, "I'm not Miranda. I'm Matt. It's Matthew." You know, I'm sure they do it on purpose.
So this made me think, you know, what could we learn about the Japanese. So I started Googling it, which is an obvious place for me to start, and what do they think about their customers. And this is what I experienced. So you've heard of customer is key. In Japan, they've got three ways of saying customers are God. So they've taken it to a whole new level. And so I just did a bit more research and found this video, which kind of suggests what I'm talking about.
Turn the sound up a bit, please. So this is the Japanese Metro. If you'd been on it, you'd recognise it. The Tokyo Metro. Two Americans. I'll voice over, shall I? Two American trying to get a ticket and they can't work out what they're supposed to do because it's complicated and not like the New York Metro. So he presses the assistant button. And a man pops out from behind the machine. This is actually what it's like. And he does everything for him from the machine. Look at that.
So that's actually what it's like in Japan. If you ever go to Tokyo, you know, you'll see that kind of customer service. So it got me thinking, "Okay, what's the difference in the UK? What could we learn from the difference in the way that Tokyo treats its customers and the UK?" So I did a similar search for people thinking about customers in the UK and this is what I got. So, you know, a little bit different.
And I think that's just the way that, perhaps, we are brought up. I'm not entirely sure. But, you know, maybe you could give me some ideas as to why, perhaps, we think that customers are stupid and horrible. And I think, you know, essentially, I went...I started thinking about this event and how we should frame the presentation for this event. And the point of this event is about back to basics. And, you know, we talk a lot about digital.
We talk a lot about technology. We talk a lot about data. And ultimately, we're trying to do exactly the same thing, which is exactly what Dave was talking about. We're trying to speak to people. And so I thought about going back to basics on Google's core proposition. And the core proposition of Google is search. And it starts with a question. And, you know, again, just listening to Dave.
I've got three children and ever since they could speak they've been asking me questions. So one of my daughters, I've got three daughters, and she's always asking me, "What's your favourite song? What's your favourite movie? What was your favourite day?" Last week she asked me, "What's your favourite dictator?" Which was a new one. It's Kim Jong-il if you're interested. To be able to be that fucking mad and have movies made about you. I mean, it's quite impressive.
So it starts with a question. And before Google, who remembers pre-Google? Anyone remember pre-Google? Google is 18 years old this week, actually. So happy birthday, Rob. Happy birthday, Google. I think Rob's a little bit older than 18, but we'll leave that one for the champagne afterwards. Before Google, people used to ask questions of their libraries. And the New York Public Library recently launched or promoted a whole bunch of cards that they had from people actually asking them questions. Here's a bunch.
And so they'd phone in or they'd write in to the library to try and get a question answered. Here's a couple of them. This is a book on how to grow hair on your chest. I haven't Googled it. I don't know how many searches there are, but, you know, this is something that apparently was quite important to New Yorkers in the 1950s. Next one. Does the female human being belong to the mammal class? I think the answer is yes, just before I get told off.
But, you know, just gives an indication of the kind of thing people were asking. Finally, and this is almost poetic, this one. This is a phone question. Is this the place where I ask questions I can't get answers to? There's, like, some kind of weird stuff going on in there and, you know, this got so much traction that actually I've set up a hashtag, #letmelibrarianthatforyou. So if you go onto any social network or social media you'll find loads and loads of information on that.
And some of them are fabulous. And it's just fascinating to see what people were searching for all those years ago and how they actually searched for it. So fast forward to Google. We've now got over a billion people searching on Google every day. We see over two trillion searches each year, which is a big number. Sixteen percent of those are brand new every single day. So it's complex and it's getting ever more complex.
I know I'm supposed to be talking about back to basics, but we need to explore the complexity first. And it's based about...so every search is a word of intent. It's basically asking...you're seeing from someone exactly what it is they want at that specific moment in time. And if they're telling you what they want, we've got responsibility at Google to actually give them what they're asking for. So some examples of how we're doing that at the moment.
This is a search for "is [inaudible 00:06:24] still open?" A relatively standard search that we see on Google. And we could just have a bunch of blue links and we could have a link to the Google...the [inaudible 00:06:33] site and you could click on there and you could try and find information about [inaudible 00:06:36], but this is actually a colleague of mine searching. And what this is actually telling you is [inaudible 00:06:41] that are open near her right now, which is the Hohoben one.
How far is and when it's open. The Finchley Road one is next to where she lives. So Google knows that she spends a fair amount of time at home. She might be going home first before she wants to go to [inaudible 00:06:56]. The Camden High Street one is where her yoga is. Again, where she spends quite a lot of time. So we're using all of those signals to actually make that search much more effective and give that user the answer they need immediately.
So then they can go off and do something else. Best hiking shoes for me. So again, you know, we're seeing a huge increase in best and top queries on Google. We could just give you a whole link...a bunch of links, a bunch of reviews. What we've done is taken reviews you see on Google. We've looked at the ad quality, the feed quality, and so on and so on, and we've ranked them. So you can actually see really easily exactly what products are available. I'm told that the Merrell Moab Ventilator is a very, very solid shoe. So if you're looking to go hiking, that is perhaps the one.
Anyone seen Sausage Party? I heard it's very funny. So this is movie time for Sausage Party. Again, really, really popular search. And instead of just, again, just showing you a bunch of links of cinemas, we know where you are and so we'll give you a bunch of cinemas that are near you, give you show times, and if you click on any of those show times you go straight through to the booking page for that cinema, for that movie.
So just make it really, really seamless and easy for a consumer, for a person to actually make that connection. And we're currently seeing, in the U.S., on Android, one in five queries of voice searches. And we're now up to about 90% accuracy on voice searches. If I think about my wife, in particular, I can't remember the last time she used her phone and actually inputted text. Every search is voice search. Even if she gets it wrong three or four times, she'll persevere.
I want to test something, and there's every chance this doesn't work. So this is one of those moments when I'm going to ask you to make loads and loads of noise, because we want to know, or we want to understand your context and where you are to be able to give you the right results. And so sometimes people are in busy environments. They're in bars. They're at football matches, as I will be tonight, Spurs.
And so, you know, we want to make sure that we can actually understand the voice search query over and above the ambient noise. So what we're gonna do is, in a second, I want you to make as much noise as you possibly can. Let out all of that anger. You've been sitting here for an hour or so already. I'm sure you need to errr. Pretend you're a football crowd, or act like a tiger, like I just did. I'm not really not that fast. And I'm going to try and get voice search to work. Now, when I tested this last night with my kids, it didn't work brilliantly, but we can give it a go anyway. So, right. Is everyone ready? Yeah? Right. Please make some noise.
[audience making noise]
Keep going. Hearing experts in London. Right, so I Googled...oh, shit. It worked, but now it's asking me to sign in to the Wi-Fi. Right, so I Googled "hearing experts in London." I should have told you that before. Is that hearing experts in London? Yes. So over and above all of that ambient noise, we still got the voice query that we wanted. So we just try and understand what situation users are in and try to make sure they get the answer that they want.
And so this is kind of like a funnel of how Google evaluates its search products and how it's constantly evolving that search product. So in 2015, we had over 120,000 evaluations of our products where we seriously considered launching them. So 120,000 things were potentially launched in 2015. This led to 10,000 side-by-side experiments, which led to 7,000 live traffic or small amounts of traffic AB experiments. And ultimately, 665 launches.
So in 2015, we launched 665 new search products from 120,000, or actually, probably millions of things that we thought about originally. So there is a huge amount of complexity going on into the search product. This is actually the timeline, the 18 year timeline of Google. And a couple of things, like, for example, auto-complete. So that was launched in 2005. We take it for granted now.
We type something into Google, we expect Google to know what we're typing or know what we're looking for and it fills in that answer. And what else, you know, the time that's saved on each search is probably miniscule. It's microseconds that we're talking about. Two trillion searches over the last 11 years, a year, is a huge amount of time that we've actually saved. And Google Health, I don't know how much you know about Google Health, but we, if you think about your health, it's the most important thing for you.
If you want health information, you want it to be fast, you want it to be precise, you want it to be accurate. And so we worked with doctors to make sure that we were giving information on the home page. So as soon as somebody searches for the most likely medical conditions. And Google now is where we're actually surfacing information that we think you'll need. So when I was in Tokyo last week, I got directions to my hotel without asking for it.
I got a currency converter. I got a translator, which came in unbelievably handy because my Japanese is rubbish and, you know, there's not a huge amount of English spoken there. And so on and so on. So just thinking about, you know, how we can surface stuff before people are asking for it. And then finally, now on tap, which is increasingly becoming Google Assistant. So this is where you could be on any page.
So I could be having a conversation with Rob on text. We could be talking about a restaurant that we might potentially want to go to. I wouldn't have heard of it because he's got a lot more money than me, so it would be out of my league. And I could just tap the Google button and it would automatically read that test, give me information on that restaurant, and reviews, and so on and so on, and potentially give me the option to book it via Open Table. So all within one click without me doing anything at all.
And so this is what we're going to be seeing a lot more of in future on the Google Assistant side. So, you know, I'm supposed to be talking about back to basics and all I've talked about is complexity. And yeah, right, technically, Google is incredibly complex. But all we're trying to do, unless we're in in the background, all we're trying to do is make it really, really simple for the user.
And as Dave was kind of talking about before, you know, whilst we we've got so many things going on, the art of marketing hasn't really changed. What we're really trying to do is get the right message to the right person in the right place at the right time. And so a bunch of things that we're thinking about along those lines right now, three principles of kind of search marketing. Be personable.
Be personal, sorry. So give people ads that actually relate to them, that are actually thinking about them and their context. Be actionable. Make sure that somebody can act on that really, really quickly. And be measured. Understand that search doesn't work in a silo and actually it needs to work and operate within multiple other channels. So what do I mean by that? So, 45% say that relevancy of a company's message influenced their opinion of the brand.
So if you're giving information to someone that's relevant to them at that moment in time, rather than spamming them with loads of appetiser that is not relevant to them, then clearly they're gonna think more positively about your brand. And to kind of put this into context, here's three people, one query, which could be a [inaudible 00:14:04] that's yet to be made. We shall see. So the query is "Hilton London." So the intent is Hilton London.
Somebody wants to know something about Hilton London. But what about three different contexts that could be in? This guy, who's just got off the train, the Heathrow Express at Paddington. He's searching for Hilton London on his mobile phone. There's every chance he wants directions to the hotel. So let's give him that, if that's what he wants. Let's try and assume that's what he's looking for. This person is in Paris. They're on a desktop. It's late at night. They're probably looking to book something at Hilton London.
This person is actually in the Hilton London and they're a Hilton Rewards member. So, you know, maybe they're looking for the fitness club or maybe they're looking for restaurant bookings within the hotel. And we can understand all these signals and we can give them information that actually makes sense to them at that moment in time. So how do we do that? Well, we've got the intent.
We know that they want information about Hilton London. We've got the context. We know what time of day it is, what device they're using, where they are. Hopefully, ideally, we've got Hilton data so we know...or Hilton knows that this is a Hilton customer. They've seen them before. Maybe they're part of the Rewards program. Maybe they've got a booking. And if we don't know any of that, we've got Google data.
So have we seen...do we see similar people within Google's data that are Hilton customers? And if we haven't got any of that, then we know age, gender, other demographic information. So putting all that together builds a really, really strong picture of how we should be actually having a better...or giving that person a better ad experience than perhaps we have done previously. Here's a case study.
This is Roland. Apparently, they do high fashion and big in-store experiences and peach dresses, apparently, by the look of this picture. They wanted to connect much more closely with their existing customers and do something special and make them even more loyal than they were. So basically, they used a product called Google Customer Match along with their first-party data to offer...to give an offer to them for their autumn, winter sale.
So on the left, as you look, is the ad that ran to people who weren't already customers of the brand. On the right is the ad that ran for those that were, just offering them a 40% discount if they come to the sale. And as you can see, 56% lower cost per click, 87% less EPA, 82% revenue from search ads. So, you know, phenomenal success. And as a result, they're rethinking everything they do on marketing.
And because this, like, massive increase that we're seeing in audience and we genuinely believe that 25% of new case search will be audience targeted in 2016, in some verticals like retail, like travel we're seeing brands at 70%, 80%, even up to 100%, which is basically where you should be if you can know a little about your audience and give them a different message a bit differently for them, then why wouldn't you?
Be actionable. So Google is one of the few places where it's important or we want to try and get people away from Google onto what they're actually looking to do as quickly as possible. And one way you can do that is with search ad extensions. So search ad extensions are just tiny bits under the main search ad. We see huge amounts of companies spending a lot of time working on their search ad copy, but less time on ad extensions.
And a few that you can see on here, so there's site links just pointing out that you can click straight on that and go to the women's page. There's directions, so take you straight into the store. And there's click to call, so actually, you can call the store. And what we see is somewhere between a 10% to 20% improved click through rate just from making these small changes to your ad copy. And the beauty of this, actually, is that it can be automated.
So you just put this information into AdWords and if Google thinks that you'll get better conversion from those site links or those extensions, it will automatically run one of more, depending on what conversions it thinks it's gonna get. A few more examples. So in the travel sector, often, people want a lot more visual experience. They want something a little bit more engaging. They want information about where they're going.
And so we launched that creative on the left, just to make sure that all sits on the home page. Ads on maps have recently been launched. So again, trying to highlight stores that could be of interest to people. And then, on the right-hand side, is shopping ads. So this is where we can see product reviews. We can see pictures of the products. And we can see if it's in-store locally to us. So with local imagery ads, we can actually see if it's in-store. So just trying to make it really super useful for the consumer.
And then finally, be measured. So connecting customer journeys on and offline. As I said, search doesn't operate in a silo. It shouldn't operate in a silo. We know that users are using multiple screens to conduct sales over the internet. We also know that the majority of users, of people, are actually going into store to purchase even though they may have done loads of research online. So in the UK, it's about 85% of commerce actually still happens in-store. In Japan, it's still at 95%.
So huge opportunities for us and for any brands that's out there on the e-commerce side. But a lot of research is happening in advance. So, and I'm sure this is a relatively typical user journey for you, so people are jumping around from platform, from channel, from device, and it's actually pretty difficult to keep track of. And so we've launched a couple of products to help on this side. So cross-device is a way of tracking users across device, obviously. Across mobile, tablet, and desktop.
And essentially what this uses is Google sign-in data. So Google sign-in data across Gmail, across Android, across apps, across Google, and so on, to actually give you a single view of that user as they go from one device to another. It's all anonymized. It's all aggregated. But essentially, companies that are currently cross-device get about 16% more conversions than those that aren't.
So they're actually really starting to understand how that user is jumping from one device to another. And then in terms of actually driving people into stores. Store Visit is a product that we launched in the U.S. a couple years ago. We've now tracked over a billion people from search engine to store. The way that it works is that if you search on a...if you click on an ad on google.com, and then you visit a store within 30 days, we will be able to measure that visit.
Again, completely anonymized, completely aggregated. It's all about privacy, as I'm sure you can imagine. And the way we track that person is we use Google's mapping technology to map out the size of the store. We use Wi-Fi. We use cell tower date. We use GPS. So we can accurately see if that person's going into store. And if we're not confident of that accuracy, that we won't report it. And the numbers here are staggering.
I mean, we all know it's what happens, but even we were surprised by how successful this is. So four times more overall conversion, and on mobile in particular, which is increasing the bridge between the on and the offline world. And it's a device that you've always got. Ten time more mobile conversions. Not 10%. That's 10 times more mobile conversions. And an example of a company that's done that, used this incredibly well, is BMW.
So the auto industry, as I'm sure you're aware, is heavily research led, but heavily based on dealership purchasing. And so BMW putting a lot of information out on the web, but they weren't really sure if it was actually having an impact on people that were buying in-store. It's something like 90% of people will research a car purchase using Google initially, before they'll actually go and purchase. But only 26% were filling in an online form. So all of those people are looking at BMW information, no one's really saying, "I'm looking at BMW information."
Using Store Visits, we can actually see, once that person's clicked on the ad to look at BMW information, whether they actually went into store. And as a result, 3 and a half percent of paid clicks resulted in a visit to a BMW dealership, 146 dealerships. And then BMW took out...they were quite conservative on the value of this, so they're quite aggressive in the amount of value they placed on other media. But even so, they still saw 173:1 ROI on search marketing campaigns. So as you can imagine, this is now very much at the heart of everything they're doing on search marketing.
So just to kind of recap. Three principles of search marketing. Be personal. Think about the context of that person and give them the information they really want. Be actionable. Make sure that person can actually do something with that ad and they can actually make a decision really, really quickly. And be measured. Connect those customer journeys on and offline. Thank you very much.