Fred, it’s so great to have you on the show.
It’s good to be here, Stephan.
You got a book that came out earlier this year, Digital Marketing in an AI World. I would love to start the conversation by talking about that. I’m really keen on AI—except if it takes over and we become its pets—but other than that, I would love to hear what the value proposition is for our listeners in terms of AI and what it can provide them today, not just in the future of, let’s say, 5–10 years from now. What is the opportunity now?
It’s more about thinking how you position yourself as Google and the engines start doing more of the work that you’ve done manually in the past and figuring out what your role is going to be in PPC management and digital marketing. That’s really what the book is about. The message is one of hope and good news. It’s not like the end of PPC careers, it’s just that we’re going to have to shift a little bit on what we’ll be working on, how we position ourselves to our clients in terms of the value proposition, and we’ll still continue to do PPC. I’ve been doing it for over 20 years. I think I got another good 20 years in me despite all this artificial intelligence.
That’s great. Let’s share a little bit with our listeners about your background because you were one of the original employees at Google—not one of the first few, but one in the first 500 employees which is a very young stage of Google Inc. How did all that come about? What was that journey like? You were there for a number of years.
Yeah. I wasn’t even sure about joining Google to be quite honest because they had this strange-looking homepage that only had a search box on it. This was back in the day of directories like Excite and Yahoo. You were expecting to go to a search page that was overloaded with sports, news, and all of that stuff. I just didn’t quite get Google at the time, but then, I started doing searches. I was like, “Oh, yeah, they actually do find better results than the average engine.”
The way that I ended up working at Google is because they had a pretty tight Stanford Network and I went to Stanford. Somebody reached out—and I had just lost my job in the dot-com bubble burst. I had been an engineer—it was like, “Hey, do you want to come over to Google and help us review all of these ads?” I was like, “Are you kidding me? I’m an engineer, I’m not going to sit there and review ads for you. That’s just beneath me.” I reject the offer at that point which probably pretty stupid and it cost me a lot of money.
But a couple of months later, they came back and they’re like, “Actually, we got something a little bit more interesting. Now we need someone who speaks Dutch because we’re expanding AdWords into its sixth language.” They’d already done French, Italian, Spanish, etc., and now, they were looking to go into Dutch. I’m originally from Belgium. So I spoke that language. I happen to be based in Silicon Valley, so I could go to the Mountain View headquarters and help them out right there. I ended up translating AdWords into Dutch. I did all of the customer support for the online self-service customers in the beginning up until it was about 5000 customers. That was how I got into Google.
That’s cool. How did your job at Google evolve then?
It was a pretty fun evolution. I joined when there were about 500 people and then I quit when there was about 60,000. That was just full-time employees. Obviously, a lot of growth and changes including for my own role. What was interesting is that I had dabbled in PPC before I went to Google. I was on GoTo for a number of years before that while I was still in college and I was using it to sell video cassettes that I bought from Blockbuster—for people who still remember the Blockbuster video chain. I would buy these cassettes that they sold before they were supposed to. Then I’d put them on eBay and I was looking to drive some traffic to these eBay listings.
The business or the economics of that was that I could basically pick up a cassette for $10 and sometimes sell it for $40 because Blockbuster was selling them too early and they were technically supposed to cost $100. That was my game but it wasn’t high volume. I wasn’t exactly making a lot of money on that. I dabbled with being an affiliate for Amazon. Again, didn’t make it very far.
But then, I came to Google and, all of a sudden, I realized that the biggest spenders on AdWords were mostly affiliates. Affiliates are interesting because they tend to be on the cutting edge of marketing, always pushing the envelope, and doing things that nobody else has thought about yet. I joined their ranks. I was like, “If they’re doing it and they’re spending that much money, they must be making money.” I started my own affiliate account again and I started running ads on Google. I was like, “Oh, my God. All of a sudden, I was making more money from affiliate marketing than from my job at Google.” It gave me a unique perspective because now I was using the product that I was supporting.It's not like it's the end of PPC careers. We're going to shift a little bit on what we'll be working on and how we position ourselves to our clients in terms of value proposition, but we'll continue to do PPC. Click To Tweet
I just had this unique perspective on it. A customer would call in with a problem and I’d actually be able to put myself in their shoes and say, “Yeah, I get it because I’ve had that same frustration.” What were these frustrations at the time? It was the fact that there were no search terms reports and there was no query data, so you only had your keyword. It could be a broad match keyword but Google wouldn’t tell you what the users actually searched for. There was no conversion tracking. There was this whole premise that advertising on Google worked better than traditional media.
That was true, but by today’s standards, it was so minimal in terms of the reporting and the oversight capabilities that we had because we literally just knew we spent $1000 on AdWords and we got $2000 worth of sales. So good, but what keywords and campaigns are driving this? We absolutely had no clue because conversion tracking didn’t exist. Because I had seen this frustration myself, I started building my own solutions and because I’m an engineer, I built my own search terms report, my own conversion tracking, and then the product team noticed and they were like, “Wow, that makes sense. We think all advertisers should have access to this.” Then I started consulting with them and got more involved in the product side.
Very cool. You probably see the end results many years later after much evolution of things that you started in the very early days to help advertisers on AdWords?
Absolutely. The most visible things that are still around that I started would be the AdWords editor. I was on the original team of five people that started that. Then I was also on the team that did the acquisition of a small company called Urchin.
I used to use Urchin before it became Google Analytics.
Nice. Were you on the online service or the hosted service?
Gosh, it was so long ago. It was the hosted service.
Alright. For those people who don’t know what Urchin was, that’s Google Analytics now. Urchin was cool because it was one of the first companies that shifted how they did analytics. Most companies give you a piece of software, you’d install on your own web servers, and then you’d get your analytics locally. But Urchin figured out how to do it at a scale where all of the data went to them and they would then process it and give you amazing dashboards and what we’re used to in GA today.
Yeah. Google was so good at picking the right companies. There were so many folks, for example, that they could have bought instead of YouTube. One of my clients back in those days was Go Fish. Go Fish was a competitor to YouTube. They could have potentially beaten YouTube if YouTube hadn’t got bought by Google. Then Urchin was my favorite of all the analytics packages out there at the time. In fact, there are still some legacy clues that Google Analytics was Urchin like UTM (Urchin).
Pretty cool. When did you start Optmyzr? How did that come about? Were you able to maybe have a skunkworks project on the side of your own while working at Google? Was it as part of your 20% time or you decided to play around with things? Tell us more about that.
So Optmyzr, we’re a tool vendor that helps a lot of agencies. So I took a sabbatical while I was at Google. I figured in that sabbatical two or three months, I’m going to start something up but it’s just really difficult when you have such a limited time frame and then doing something as a moonlighting job. I could have done some agency work but then it potentially becomes a conflict of interest. I stayed away from much of that.
I eventually turned into the only Google Ads evangelist—and back in the day was Google AdWords evangelist—and I basically made that role because some of the executives like Sheryl Sandberg, they would come to me and they would say, “Listen, people are asking us to go and speak at events.” They’re super comfortable presenting on the product, the vision, and the things executives do but then when it came down to answering questions about, “How do I use this specific feature and how does that interact with this thing?” That’s where they have added their depths a little bit.
I stepped in and started handling those types of events and created a role that I said was the AdWords evangelist. I got to do a lot of travel. I got to do a lot of education. I had a lot of fun with that. I worked with the marketing team and PR team. It was super exciting for me.
Then, as the company kept getting bigger and the management changed, they said, “Listen, we like what you’re doing but we got to do it at a bigger scale. How do we take your role but you sit in Mountain View and figure out you make the presentations and then, you teach other people how to give those presentations?” That shift was not something that was appealing to me. I had some resources built up that I could decide to leave. I left and funnily enough, the first day that I no longer worked as a Google employee, I was already on my way to a Google event as a Google contractor making more money to speak at that event then they would have paid me as a salaried employee.
That’s just the way.
Exactly. The nice thing was at that point, the conflict of interest was gone. Now I could actually start to pursue some side projects. Initially, I was an agency. That’s probably where a lot of people might still think of me as an agency, but I did an agency for about a year. The problem that I found was I would speak to people about what they should do. I’d get onto a lot of phone calls with customers and then at the end of the day, I’d realize I had no time left to do all of these things that I knew we should be doing that was the right strategy. Part of it was simply that as good of a job as I thought we had done inside of Google to build tools that really scale, there were still a lot of pain points for advanced advertisers.
One specific example, I worked with a shopping client. They had something a hundred thousand products, not even that many products, but to build out the structure for their shopping campaigns that I thought they should have, it literally took me a week. I said, “That’s unacceptable.” I started looking at technology to solve these problems. Luckily for me, at the time, AdWords scripts had just been introduced. Actually, they’d been introduced a little bit before then but nobody was using them. There was no momentum behind these things. I had a chat with the product manager. After I left and went back to campus with a friend, talked to him, he was like, “It’s so sad. We have this amazing Google AdWords scripts program but nobody’s using it.”
I went home that afternoon and I built a little script a couple of hours that helped me with one of these scalability and efficiency problems. I was like, “Oh, my God. This is gold.” I started speaking about it at conferences, I started writing about it on Search Engine Land. It resonated with people, a lot of PPC experts had these same problems. Like myself, they didn’t necessarily want to pay the high fees for some of the enterprise-class tools. Some of the lower end tools that were more affordable didn’t necessarily have the capabilities that the expert needed.
That was that sweet spot in the middle that I started to get involved in and eventually met my co-founders. We started Optmyzr to build that technology, make it available at an affordable price so that any agency or practitioner can use these tools to make their lives better and save time. That’s how Optmyzr came to be.
Let’s talk a little bit about what Optmyzr toolset provides. Why would somebody need to use that and not just use the Google Ads toolset that comes with your Google Ads account?
Here’s the problem with the Google Ads interface, they built it for the masses. When I was at Google, we actually had many interesting discussions where we said, “Should we have an AdWords for travel and an AdWords for retail?” It always came back to, “No, that’s going to complicate things. We need to keep one product. It needs to work for every single vertical but it also needs to work for the small mom-and-pop advertiser as well as it does for the Fortune 100 company.”
That’s challenging now because the Fortune 100 company that has a whole marketing department, potentially 10–15 people working on the ads account, they’ll probably going to want to do some more sophisticated things, they understand it better than that small mom-and-pop operation yet the interface had to be the same. That’s where you start to see the pain points. If somebody wants to deploy a certain methodology, like on the shopping side, we call it the GRIP Structure (GRoups of Individual Products) it’s kind of like SKAG on the search side or the Alpha Beta structure, these things that experts know about, they’re very painful to build. It’s a lot of tedious spreadsheet work, etc. It can take you multiple hours. We said, “That’s where we can actually offer a solution where we have a purpose-built specialty tool that helps you do this.”
Optmyzr breaks down into a couple of areas. We have data insights, helping you understand more quickly what’s wrong with the account. Let me give you an example of that. Probably the most painful thing that happens to an account manager is your client calls you up out of the blue and says, “Hey, Jimmy. Why are my conversions lower than they were last month?” You may not have been paying attention to that account today because you’ve got a big account load so now you’re on the phone and you quickly have to come up with a good answer. What you do, you go into Google AdWords and you open up the date comparison columns. Now, you have to do horizontal scrolling and you have to look where each column is. It’s confusing.
What we do is we have a Cause Chart. It’s basically a tree visualization. We put conversions at the top. It will say why your conversions are down 10%, why do you have conversions, because you get clicks, and conversion rates. Those two factors sit below that. We tell you, “Actually, it’s your conversion rate that’s down by 20%. Your clicks are fine.” Now in an instant, you can tell that client, “Oh, yeah, sorry. Clicks are down a bit. It looks like the conversion rate has gone down. I remember we did some landing page work for you. Let’s go and take a look at that and maybe fix up the landing page. We’ll get you back to where you need to be.” That’s a good example of insight.You should be able to play with the data, explore it as it leads to new insights. You should be able to ask a follow-up question without having to spend another half day doing a manual analysis of it. Click To Tweet
Quality Score is another one like that. Then, we have One-Click Optimization—this is about specialty optimizations. We know that your average account manager spends a lot of time looking at search terms, query data, and figuring out, “Should we be adding negatives based on this or new positive keywords?”
We have a tool that says, “Let’s take a look across all of your queries, but we’ll split the queries into individual words.” We do n-grams basically, unigrams in this case. Then, we count up the totals and we say, “Here’s a word that commonly appears across many queries that have a lot of costs but never any conversions. That might be a good idea for a negative broad match keyword. It will save you money.”
We have a tool that helps you do ad testing. We have a tool that helps you build RSAs—responsive search ads. These are great, these are something you should have, you should test them out, but they’re painful to set up. In fact, I have a video on YouTube where I show if you have a campaign even with just five ad groups in Google, it’ll take you in excess of three minutes to build responsive search ads for all of these ad groups. With Optmyzr, you can do it in 27 seconds.
That’s amazing. We’re going to have to include that video in the show notes of this episode, which, listener, it will be at marketingspeak.com/206. Please get me the link to that YouTube video and I’ll include it.
I’ll definitely get that to you. Again, these are specialty workflows. What’s interesting about Optmyzr is that, as a company, we’re willing to invest the time to build something that may, in a sense, be a throwaway tool. We do understand that most advertisers have transitioned to using RSAs. Now, the RSA builder is much less relevant, but we do think, given how many customers we have, that it’s worth it for us and for our customers to just put that tool out there even if it’s only going to be used for six months.
An advertiser may have an in-house team, but they also have to allocate time and resources to different projects, so they may never want to do this for you. They might just say, “Jimmy, sorry, but you got to go and do this manually. It sucks for you that you have to spend this much time, but it’s just not a priority.” That’s where Optmyzr can really help out.
Two more big areas that we have, we have Enhanced Scripts. This is where we’ve pre-built AdWords scripts for you, that you can just copy and paste into the account. We have a reporting tool so that you can put your report generation and emails on autopilot. Those are the big areas. Then, we have workflow tools as well. If you’re an agency and you need to layout your process in a tool and connect different tools to that process, that can be done through Optmyzr.
We have a brand new tool which is called a Campaign Automator. This goes from structured data through your template to having fully built-out campaigns. It’s super popular in the automotive space, in e-commerce where you have an ever-changing inventory. Every day, for example, we take a look at what is in your inventory. We figure out, “Oh, there’s a new vehicle that you’re selling that requires a new campaign because we didn’t have a campaign for that brand.” We spin it up and we know exactly what keywords you want in there or what ad extensions, expanded text ads, responsive search ads, and all of that is set up automatically. Then the next day, you sold that one car from that one manufacturer and we take the ad down because we know it’s out of inventory. We make sure that you’re always advertising just to stuff that you sell.
That’s awesome. That sounds like the next level past the wizard thing, let’s say, Windows or whatever. It’s just an easy configurator. This does it completely automatically for you, it sounds like.
Exactly and that’s what we’re realizing, too. We talked a bit about my book in the beginning. What do account managers do when automation is becoming so prevalent? Here is the funny thing about automation from Google. There’s almost no situation where you can just say, “Hey, I’m new to this whole thing, just do it for me.” It always needs a basis for something, so it needs a campaign to be built out. You need to select at least some keywords before the broad match and the close variants can start kicking in.
If you’re going to do automated bid management, oftentimes you need to build up at least 15 conversions in 30 days before it can start to automate. That means you’re still setting manual bids, you’re still building out the structure, and that’s where we see a big opportunity where we can really help advertisers. We don’t want to fully automate this. I mean, we think there’s a lot of value into humans making these decisions.
Yeah, for now, exactly. Things might change, but as a person, you have that template in mind. You know how you want to structure your campaign. If you can communicate to us what that template looks like, then we’ll figure out how to build it out for you through technology as opposed to you having to write a whole bunch of formulas in Excel and probably get it wrong a couple of times before finally getting to the right solution.
By the way, that’s another thing that’s really cool about Optmyzr because we have these data insights. We have seen presentations, I think it was Jeff Allen from Hennepin. He gave a presentation at a conference and it was a really great presentation about how he’d done some research to figure out something like the geo-performance of his campaigns. What he said was, “I went through this whole process. Then, after spending two hours doing the analysis, I realized I really should have split brand versus non-brand campaigns or I should have looked at mobile performance differently.”
The problem was, he had to go back to step number one of his two-hour process, re-pull the data using those segments, and then put it all back together in that same inn. That’s humongously painful. In Optmyzr, these types of analysis, you run it one way and then you click a button that says, “Oh, let’s run that same analysis but just for mobile.” Literally, click one button that says Mobile Go and then it’s done 30 seconds later. You have your new analysis, so the data becomes much more interactive.
This is like the Google Analytics model. You should be able to play with the data, explore it as it leads to new insights, you should be able to ask a follow-up question without having to spend another half day doing a manual analysis of it.
Interesting. Do you have or maybe in the works, some what-if calculator? If let’s say, the pricing changes of your product, or your conversion rate changes, or some other variable changes and you can actually say, “I want to change it and see what this impacts in terms of everything else further down the line.” That could be a really interesting tool or capability.
Exactly, like segmenting things out so you get deeper data. That is something we’re thinking about. Actually, that made me think of another product that we have which is called the Rule Engine. It’s not a what-if, but it’s if-this-then-that. It’s cool because now, you can connect this to not just ads data but your own business data. You could say, “If the weather in the city for this campaign has become rainy or the temperature has dipped below a certain point, then do something with it, like change the bids, pause it, or put a label on it,” so that I know how to put it, do something else with some other tool.”
You can start to automate these logical process flows using your own data, ads data, mixing and matching things in different ways, using different look-back windows. It’s pretty cool what you can do with that technology.
That’s awesome. How far have we come since the early days when the innovative thing was to do dayparting. I’m just thinking, maybe some of our listeners are not familiar with some of this lingo that we’re speaking like dayparting. They may not be familiar with that. They may not be Google Ads experts and just really interested in learning more about how the ad side of Google works and how maybe they could get started.
Let’s take a few minutes to define for our listeners some of these terms that we’ve talked about. Dayparting would be one example, another would be negative keywords, another would be ad extensions. You already defined RSA (responsive search ads ), so we don’t need to go into that one. A broad match is another term that you use that some folks probably are not familiar with. Quality score is another one. A lot of things. I’ll remind you of all those different terms to make sure we hit them all. Maybe in just a little bit of a lightning round, could you define all these?
All right. There’s a lot to unpack here, but at the highest level, AdWords now called Google Ads was really a product where the advertiser would pick keywords, put in a text ad, and put in a bid, usually a cost-per-click (CPC) bid. Then Google would start to show the ads based on an auction dynamic.
Now, the auction dynamic was looking at how much the advertiser for a specific keyword was willing to pay for that keyword, the CPC bid, but it also took relevance into account. In the beginning, it was just called relevance, it was called CTR (click-through rate). Click-through rate is simply the number of clicks you get divided by the number of impressions. A simple example, if you get 100 times your ad shows and 5 people click on it, 5 divided by 100 is a 5% click-through rate. If a different ad gets 10 people to click for the same 100 impressions, they have a 10% CTR. That eventually got turned into what Google calls a quality score. It’s an indicator of how relevant something is. The higher your quality score, the less you have to pay for that same click.
Now, I talked about keywords. The advertiser typically goes in and says, “Here’s my keyword.” It might be a keyword like flowers. Now, you can have a broad match keyword that says, “Google, it doesn’t have to be exactly someone searching for flowers but it has to be in the same category of things.” If they search for red roses, that’s a kind of flower so that might be eligible to show the ad. That’s your broad match keyword.
This is also why I’m talking about the search query and search terms report. The search term or the query, that’s what the user put into the search box. That, in some way, connects to the keyword, but it does not have to be exactly the same. As a broad match, it has to be in the same category.
Google has this thing that it calls the exact match keyword, so the query has to be exactly the same as the keyword. Even if your keyword is exactly the same as the query, normally that would have been called an exact match keyword. But now, Google says, “We can also show an exact match keyword when the words in the query are different but they express the same intent.”
The example Google’s giving is lawn mowing service would be the same as lawn cutting service because mowing and cutting are synonyms, so they express the same intent from the searcher. That would still be an exact match keyword. That’s why you have all these different keyword match types that let the advertiser know how closely related the search from the user has to be to the keyword that the advertiser provided.
This is a good place to interject that something the advertiser probably already realizes but needs to always keep in mind is that Google’s incentives and their incentives are not aligned. Google’s mission is to maximize the profit they get from you, the advertiser. In the case of this feature of synonyms like lawn mowing service, lawn cutting service being considered equivalent in terms of exact match, it seems like it’s a good thing for you but it’s really a good thing for Google because then they increase the competition and the search volume, and it’s a way to further monetize you. If you’re reliant on Google’s tools, their reporting, and so forth, you’re probably going to miss out on a lot of opportunities and lose a lot of cost-cutting measures, I would imagine, right?
Yeah. I’ll put my AdWords evangelist hat on here. I did drink the Kool-Aid a long time ago and it’s still in my system. What you’re saying is right, so you’re going to see more competition, but ultimately, Google is doing this not entirely just for their revenue, they’re also doing it because they realized that a lot of advertisers just are at a loss of expressing what they sell and how that connects to what users are searching for.The only choices you have as an advertiser is to either spend a lot of time manually checking that things are going the way they should or automate on top of the Google system. Click To Tweet
Google has this stat, it’s really old but actually recently, we looked at it and it says 15% of queries that happened on Google are unique in something like the last 30 days. There’s literally 15% of the searches that you and I could probably not even think of how to formulate it because people are so creative in that. They’re basically saying, “Listen, this 15 % of unique queries, they’re still useful, they still might match to some advertisers, but it’s just really hard to have all these keywords. Why not give you that potential for 15% additional clicks, additional conversions?”
Now, all of this, as you said, it’s a bad thing if you’re doing manual bidding because now you have to monitor your search terms, you have to break them out into actual keywords, then you have to set separate bids, and sometimes you don’t have a lot of data. It’s a lot of work, it’s very difficult to do it correctly, but nowadays, Google has Smart Bidding. They have a lot more automated bidding capabilities that you can turn on and in conjunction with these broad keywords and close variant keywords, they actually do a pretty good job because they might say, “Listen, this thing is related but it may not have the best conversion rate.” They just still show your ad but they put you at a much lower bid. If you get that conversion, you’ll still get somewhere close to your cost-per-acquisition or return on ad spend target.
Yeah, because I’m thinking there are occasions where, let’s say, one cutting service doesn’t attract this high quality of a prospect as lawn mowing service.
Yeah, absolutely. There are different words and they might speak to different customer demographics, they might speak to how people search in different parts of the country. If it’s someone looking for a lawn-cutting service in San Francisco, they probably are willing to pay more than if it was in Cleveland. Those types of things we would all have to take manual control over and that’s what makes it complicated. With the Smart Bidding system, Google presumably handles that and they should still get you to your cost-per-acquisition.
The one concept that I’ve been talking about more recently is what I call automation layering. There are some automations that Google makes you use. You have no choice but to have close variant keywords. You cannot opt-out of these. The only choice you have as an advertiser is to either spend a lot of time manually checking that things are going the way they should or you can have your own automation built on top of Google system that says, “We’re automatically going to check. We’re going to monitor these close variants,” because Google does put them in the reports and they label them as close variants. Then you can say, “If we see a close variant that’s spending a ton of money compared to its originating keyword and it’s not driving conversions, then automatically make it a negative.”
You could write a script for that. In fact, I’ve put a script, you can find it on Search Engine Land, and you can put it on your show notes. There’s a script that does that for you where you can use more of a semantic approach, where you calculate the number of character differences between the query and the keyword. If that exceeds the number of characters, then you can say, “Okay, maybe it’s too far away from the original intent, so let me scale that back and make it a negative as well.”
All right. That’s a great segue to finishing up defining some of these terms, tell us about negative keywords briefly. Also, we need to do a couple more. Ad extensions is one term we haven’t defined for our listener yet, too.
A negative keyword is actually a great control because you can specify what words cannot be part of the query. If they are in the query, then your ad will not be shown. That’s the purpose of a negative keyword. Again, with these close variant, exact match keywords, and maybe your keyword is lawn cutting service but Google decides to show you for lawn mowing service, if you think that’s a horrible thing to do, then you just put negative mowing as a keyword. Now that’s the negative keyword mowing. If that shows up in the query, your ad is not going to be shown. It gives you some of the control back.
Of course, the problem is you have to spend time to figure that out to put in that negative keyword. Account management does get a bit more involved. That’s why I’m such a fan of figuring out a way to automate that component of it.
That’s a big opportunity if somebody’s not using negative keywords. What’s the maximum you can have in Google Ads? I think it’s 10,000?
Yeah, in a campaign?
It varies. At the account level, you can have a million keywords but that’s a soft limit that they can extend for you and then like you said, it breaks down to the campaign at that group level as well.
If you were to automate, as you were saying, the creation of a huge list of negative keywords, that could save you a lot of money. I like to use this example because they were a client of mine for a number of years—from the SEO side, though, just to clarify. If you’re Zappos and you’re buying the keyword “shoes” on broad match, then that would get things like brake shoes, snowshoes, and horseshoes, all sorts of shoes that you really are not interested in, free shoes, whatever. You would negative out all of those words like horse, brake, snow, and free, but if you want to have an automated tool generate a huge list for you so you don’t have to do all that heavy lifting yourself, all that thinking of every single negative keyword, that would be pretty awesome. Does your toolset do that?
Exactly. We would figure out negative, like the root word “free.” We would notice that, “Hey, that appears in quite a few queries.” You’ve spent, say, $100 across all these queries and it might be mixed. It might be 10 cents here, a couple of dollars there, but in aggregate, you spent $100. You got no conversions from it and we might notice, “Hey, in your account, you typically have a $50 cost-per-acquisition. You’ve exceeded that number by quite a bit. This keyword always tends to appear in those queries so let’s make that a negative.” We do offer that intelligence.
Then it’s also a one-click optimization. Not only do we tell you that it’s an issue, but we automatically then with one-click, put it into your account and make sure it doesn’t happen again. My thinking is you should probably have an account or campaign level negative lists with things like free images, the things that you can pretty easily think of as solid negatives. Where it gets more complicated now is that example, the lawn cutting versus lawn mowing. Maybe the numbers need to tell us if this is a good match or not because semantically, to me, it didn’t make sense but you might actually see very different numbers.
One crazy example I remember for my Google days is we had this tool that would show us the average system-wide CTR for keywords and there was a massive difference between the word plumber and plumbers, the plural and the singular. You think about that plumbers probably means someone’s looking to hire a plumber whereas if somebody’s typing in plumber back in the day, at least, it seemed to indicate it was more of maybe looking up at one specific plumber they wanted to get more information on. It tended to perform less well for others, but really minimal changes between keywords can have massive impacts on the performance.
In the Google Ad keyword tool, you get these keywords getting grouped together. It was very irritating because then you wouldn’t know which close variant was the one that actually had all the volume and then also later down the line, which one would generate the revenue for the advertiser. I would use the Google Ads Keyword Tool all the time—Keyword Planner. It originally was called the Keyword Tool and then it’s called the Keyword Planner. Eventually, I just stopped using that as a primary keyword tool because it had so many flaws. Now I opt for using Moz’s Keyword Explorer and a bunch of other tools.
That all makes sense. I was involved at some point with the Keyword Tool. It was called the estimator at that point. The engineers at Google really wanted to do these exciting projects, so they wanted to predict the future volume and future cost of certain keywords. That was then people like you would say, “Hey, the numbers are wrong. You predicted X but I saw Y. The tool doesn’t work. I’m not going to use it anymore.”
Advertisers really just wanted historical data like, “Tell us how many times has it been searched and I’ll make my own judgment on what that means for the future and the applicability to my own campaign.” That was sometimes the disconnect between Google engineers and customers that sometimes we just need something simple and they want to do something really advanced.
That’s also where Optmyzr can then step in. If we listen to advertisers and if enough of our customer base say, “Hey, this is how we look for negative keywords or positive keywords, we’ll just put together a tool around that and make your life easier.”
We go one, maybe two more terms to define for our listeners. The ad extensions one, let’s go to that one.
The ads on Google have evolved quite a bit. In the beginning, they used to be simply three lines of text, a headline and two lines of description, and then the visible URL, of course, as well. Over time, more stuff got added to that and it was basically on the premise that ads are information, too. By the way, Google’s mission to organize the world’s information and make it universally accessible and useful, I don’t think it says anything about revenue in there. Though I thought that was a good point you made before. To the point of all the world’s information, they just said, “Hey, ads are information too,” and how do we put more information in these to make them more useful to our users? Oftentimes, that means putting in a couple of additional links, putting in some attributes of your business, what kind of payment do you accept, what brands do you carry, what services do you offer.
Google said, “You can still have your lines of ad text. In this case, now it’s up to three headlines and a description line which is much longer.” Then you have ad extensions. You can put in additional links or price points, a location, a phone number. Those things are done through what Google calls Ad Extensions. What’s nice about this is that Google will variably include these as needed. They will figure out, “Okay, does this seem to be a local search that somebody’s doing?” If that’s the case, then they might show the address next to your ad. Whereas if you have a high ranked ad at the top of the page, but it’s for a generic keyword like travel and you happen to be priceline.com, then they might say, “Here’s the ad for Priceline, but then we’re also going to put in four site links to the different portions of that site like book a cruise, book a flight, book a hotel, book something else.” That’s what those do.
You could add site links underneath your regular Google ad and that’s an ad extension option. Another one is your phone number. You can actually have the phone number be a trackable phone number so you can associate the keywords that you bid on with the phone calls that were coming in.
Pretty cool. Now, did we define dayparting for our listeners? That’s where this all started.
I don’t think we did. Dayparting or ad scheduling is simply the notion that you can tell Google certain times a day to turn your ads on and off. If you are call-center based and you don’t want to show your ad because your call center closes at [9:00] PM, you can automatically turn your ads off at that time of the day and then turn them on automatically at, say, six o’clock the next day. That’s the simple scenario.
The more advanced scenario is that you understand there’s a difference in value from clicks at certain times of the day. You actually put what they call bid adjustments. You combine a bid adjustment with a daypart, so you might do something between [6:00] PM and [7:00] PM, I want to increase my bid 50% because we know that historically, people that have clicked our ad at that time tend to convert at a much higher rate. That’s another way you can use ad scheduling.A negative keyword is a great control because you can specify what words cannot be part of the query. Click To Tweet
That tees up another point that I should probably make, but you were talking about trackable phone numbers and knowing when your conversion rate is better at certain dayparts. The one thing that’s really important to remember is that Google has different attribution models for conversion tracking.
Historically, people have used last-click attribution. What that means is that you look at the last interaction that someone had before converting, whether that’s making a phone call, buying something, or filling out a form on your website, but they only value the last interaction that happens to the last keyword that was searched.
Now, that’s often tricky. Historically, it’s been okay. Say that you’re selling Adidas sneakers. You look at a report and it says, “Your keyword ‘sneakers’ has had zero conversions and $1000 in cost.” You, probably, realize sneakers is actually relevant to Adidas sneakers so I probably shouldn’t turn off this keyword even though it says no conversions. Why are you making that decision about? Because you realize people at the earlier stages of research might not know they were looking for Adidas for a specific model of shoes so they search something generic.
Nowadays, when you have much more automation, you have automated bidding for example or you have a script that you installed that says “Turn keywords off. Make negative keywords if it’s not converting,” it could easily pick this up. Purely on the numbers, sneakers is going to look like a bad keyword because you gave it no value.
Nowadays, there are many more sophisticated ways to track attribution modeling. You can do a time decay, you can do data-driven attribution. They’re all different, I don’t go into too much detail here, but the basic premise is that these different models can associate at least some value to each interaction that happened not just the last one.
Now, if you put in the right attribution model, that keyword ‘sneakers’ that never leads to a direct conversion but does take people down the path to finding the type of sneaker that they want and that you sell, that now has value and it will no longer just be bit down or be made into a negative keyword by some automated system which would have been very detrimental to your business.
Right. If the listener is still using a very basic attribution model like the last click, then they’re probably making some mistakes in what they’re doing based on the data, right?
Let’s talk about ad testing a little bit. When you want to improve the performance of your ads, you’re doing AB tests and so forth, you can test the landing page, you can test the ad copy or the ad creative as they call it, you can test a lot of different things, and you probably want to get more sophisticated than just what the Google Ads interface supports, right?
Right. Thanks for teeing it up because it’s actually one of the tools we have in Optmyzr. We have an AB testing tool for ads. It does something very simple. It says, “Let’s take each ad group and different ad variations that you have in it. Each of these is its own horse race and we want to find a winning horse for each of these ad groups.” That’s what it does. It uses a statistical significance calculation to say, “This looks like the ad with the best performance for this group of keywords within this ad group.” So, then we automatically let you remove the assets that are not working well. We have an audit tool that will tell you later on, “Hey, it looks like you’ve got 15 ad groups that don’t have at least 3 ad text variations,” or however many you want to check for, and then you know, “Okay, I need to go and build new experiments for those ad groups.”
We even have a tool that says, “Okay, now you want to build a new experiment, here’s a couple of suggestions based on things that have worked well in different parts of your accounts. You can just copy and paste those and then slightly tweak them.”
Ad testing is very important. It is evolving because of responsive search ads. Google is no longer asking the advertiser to put together an ad in its entirety, but it’s saying, “Give us two components. Give us 15 variations for the headline, 4 variations for the description,” and then they on-the-fly will put it together in the way they think is best for advertisers. But even that is optimizable because Google will tell you, “Okay, out of the 15 headlines, these 2 we don’t think are working so well,” so you can go ahead and pause those and put in new variations and keep evolving that way.
Can you give an example of a client or a client’s client, if it’s an agency as the client and then their client found some insight or capability or found some additional money for the client by creating some experiment? What would be an example, a case study, a short one?
We have a couple on the website, but one that comes to mind that we recently did was specifically with these responsive search ads. It was a time-saving thing. The agency knew they should be doing this for their client that was in the dental service organization but it was just going to be so time-consuming. Then they found Optmyzr’s tool that could do this in a matter of seconds and they ran it across the whole account. Then they quickly saw, I believe, it was a pretty good increase in conversions and a decline in cost-per-acquisition. I don’t have the numbers off the top of my head but it’s in the case study, that’s on our website.
Okay. I’ll include that in the show notes as well, the link to that. Can we go back to some of the expert things that you had mentioned like that GRIP capability? Can we learn a bit more about some of the ninja things that very few tools out there or even PPC folks know about?
Yeah. The GRIP structure stands for GRoup of Individual Products. That’s something that we do for shopping campaigns. The more granular that we can get, we’re putting a single product in a product group, it just gives us more control over what negatives to put in, what bids to run. It gives you a lot more visibility.
The real ninja stuff does rely a little bit on the Optmyzr here, but after you’ve built this out, which you can do very quickly and easily through our tool. We can also do an interesting analysis. Say that you have 100,000 products that you sell. You’ve maybe grouped them by brand and first-level product category. Now, we can do an analysis that says, “Besides these two things that are part of the structure for which you can easily see the data by going to the reports, tell me something about the colors of the products or the combination of the brand plus the color. Do we see different conversion rate patterns between these?”Minimal changes between keywords can have massive impacts on the performance. Click To Tweet
Because we have the merchant feed which has all the attribute data from the products, we can generate that report in the back-end. It’s quite a bit of data crunching that we need to do, but we automate it. Then we’ll tell you, “Hey, listen. It looks like orange T-shirts from Adidas are much better than black T-shirts from Nike.” But now because you have the GRIP structure, you can actually take action on this. We can go and find your 100 product groups that happen to have orange T-shirts from Adidas that we can go and increase the bids for those.
It gives you an entirely new level of control that would have not existed in the past. Of course, to clarify that example maybe a little bit, a lot of people, if they have a structure that doesn’t go down to the individual product level, they may have all of the Adidas T-shirts in a single product group. These are yellow T-shirts, orange T-shirts, or blue T-shirts.
Now, if we have a finding that says, “Blue sells better than yellow,” you can’t do anything with that because your blue and yellow T-shirts live in the same place and that same place, by nature, has to have the same bid from Google. It’s a great insight but you can’t act on it. That’s why we like having the GRIP structure. That’s one ninja thing.
Another ninja thing that we can do is Google does a lot of stuff around target CPA and Target ROAS bidding. We believe in profitability. Most listeners would agree, it’s not about ROAS, it’s about profits.
ROAS is return on ad spend for those listeners who are like, “What’s ROAS?”
Exactly. You got ROAS, but ROAS is an intermediate step to profitability. To derive profitability, you have to know what is the margin of your products. A high margin product, you can spend more money on advertising and still break-even on it. What our tools can do is we can bring in your margin data and combine that together with ROAS automated bidding and group things in the right way so that every product, as diverse as it may be, that has the same margin, we can group it together. Then we can set the same ROAS target and now Google can go automate bidding and make sure that you actually hit profits as opposed to just some magical number.
By the way, a great story I picked up from a Googler who was working with a lot of agencies but they had taken over account management, the client was like, “Hey, we need to hit a $4.4 return on ad spend. They were like, “Okay, sure, we’ll go and do that.” They work on the account for a year and they’re struggling. “Why is this a $4.4 why does it have to be a $4.4? We could make so much more profit and drive so much more revenue if they allowed us to go a little bit lower on return on ad spend.”
After a year, they asked a question and they’re like, “Why is it $4.4?” The client says, “That’s what it was with the last agency that managed us so I just figured if it became worse, it’d be egg on my face and I would look bad.” A really stupid reason that had nothing to do with the actual business, but I think that’s actually pretty common. People grasp onto these numbers and then that becomes their truth. But at the end of the day, it’s like what does your business try to achieve and how do you translate that into some Google numbers?
What they did for this client now, they went to maybe a $3.5 ROAS which is a lower return on ad spend, so fewer dollars being generated per advertising dollar but they were actually able to drive a lot more conversions, a lot more revenue, and I believe they even became more profitable.
Oh, very cool. Any other ninja techniques that you want to share other than those two?
Oh, my God, there’s probably so many. In the Rule Engine, we can do things like relative expensiveness factors. Oftentimes, if you ask someone how to manage bids, they can’t really even tell you the cost-per-acquisition or CPA target that something should have because justifiably so, they say, “My CPA for a brand campaign and a non-brand campaign should be different.” That makes sense. Your brand clicks are going to be cheaper than your non-branded clicks.
What our tools can do is we can do a relative analysis. We can say, “Look at every keyword inside this campaign and compare it to the overall performance of that campaign.” This is under the presumption that similar keywords tend to live in similar campaigns, brand keywords tend to live in a brand campaign. Now we can say, “Even though this is a really cheap keyword because it’s a brand keyword, it’s actually significantly more expensive than all of your other brand keywords.”
In the grand scheme of things, that piece of data may have been hiding because averages tend to lie, they obscure the underlying data, but now this brings it out and says, “Yeah, still cheap keyword but relatively expensive compared to other brand keywords so go and take a look at it and try to fix this.” This is another cool capability that we can do.
Nice. We got only a few more minutes left. I wanted to get a couple of points in before we close out the episode. What about Amazon? To completely switch gears, you guys are working on some pretty cool stuff with Amazon as I understand.
Yeah. I’m glad you asked about that. Have we got so much going on, I sometimes forget to cool stuff we’re doing. Amazon Ads, obviously one of the fastest-growing ad platforms in the world right now, especially in the US. Shopping ads are going crazy. This is just in line with that. We’re now supporting, managing Amazon Ads, so finding keywords, negative keywords, setting better bids, we can do all of that using the Rule Engine, so the if-this-then-that type of methodology for Amazon ads as well.
Optmyzr today works with Google ads, Bing ads, Facebook ads, recently also with Amazon ads. We have direct integrations with Google Analytics. We have Data Studio integration so you can pull our data into your Data Studio reports. We do have an Alexa integration as well. You can ask your Echo device from Amazon how your ads are doing in she’ll usually answer.
Lightning round here real quick before we close out the episode. Digital Marketing in an AI World, that book is a lot about future-proofing your PPC agency or just your PPC department if you’re part of a retail or something. What would be something that we haven’t talked about already that will help the listener future-proof their business even if they don’t use the Optmyzr platform?
The biggest thing here is to acknowledge that it’s not brand. AI and automation will disrupt your business but brand right now, just being ready for that, and just understanding your role. The book really breaks it down quite nicely into three human roles that we’re typically used to. The PPC expert as a doctor, as a pilot, and as a teacher. Each of these three distinct roles has specific capabilities where you can add a lot of value to your boss, to your organization, to your clients. There’s a lot of details about these examples in the book.
Then the other portion of it is just explaining how this technology works and really understanding that these are very specific solutions. You have bid automation and keyword automation. To this day, you have very few automations that span across everything. You, as a human, to have oversight and put the right pieces of the puzzle together, that’s becoming much more important than doing the actual work. It’s more of a strategic role than a tactical role.
Right. In fact, that dovetails really well with something I discussed in another episode, I forget which one, but I was talking about how I had been at an Abundance 360 conference which is Peter Diamandis from the XPRIZE, he runs that event. I was a part of Singularity University. One way to ensure you don’t get disrupted—or not ensure it—but to delay the inevitable as the AIs become so much smarter than us is to find things that you could creatively put together that are just not obvious.
For example, if you’re an expert in, let’s say, quantum mechanics and you’re also really good at downhill skiing, is there a creative way that you can bring those two things together, those two disciplines? That is where the AI is going to fail and you’re going to be able to still have a job.
One last quick question, if you could share some nugget around retargeting for Google Ads like retargeting for search or for display, what would be a nugget that our listeners could take away and implement next week in terms of remarketing?
I think at startup having some audience lists, build those out and at least put them on observation for your campaigns so that you start getting the data. Here’s the funny thing about Google when it comes to geographic and device-type bid adjustments, even time of day and the dayparting we talked about, you can turn that on today and you can see historical data from the very first day you started running ads on Google.
With audiences, on the other hand, you have to tell Google what the audiences are and only then will it start to collect data around that. Even if you’re not ready to really make it part of your strategy, just having them built out so the data starts to collect is going to put you in a much better position down the road when you actually do have time to think about it and figure out how to leverage these audiences.
Yeah, because that data is an asset if your whole point of having a business supposedly is to build up the asset value and then have an exit. If you are neglecting certain assets within the business such as your email list, your retargeting/remarketing list, or other aspects, that is missing out on some of the asset value of the overall business.
Awesome. Thank you so much, Fred. This was fabulous and really fascinating for me. I’m sure very educational, too, for our listeners to figure out that there’s a lot of money to be left on the table if they don’t delve deeper into this area of Google ads and Amazon ads for that matter. Thank you so much.
Thanks for having me.
Go get back to me with those additional things that I can add in the show notes. Again, listeners, that will be at marketingspeak.com/206. Thank you so much, Fred.
Thanks, Stephan. It was great.
All right. Listeners, we’ll catch you on the next episode of Marketing Speak. This is your host, Stephan Spencer signing off.
- Frederick Vallaeys
- Optmyzr – Youtube
- Digital Marketing in an AI World
- A script to create a Google Ads Experiments dashboard
- Responsive search ads for dental services organization – Case Study
- Kaspar Szymanski – previous episode
- dot-com bubble bursting
- Google Analytics
- Sheryl Sandberg
- Search Engine Land
- Jeff Allen
- Moz’s Keyword Explorer
- Amazon Ads
- Google ads
- Bing ads
- Facebook ads
- Data Studio
- Abundance 360
- Peter Diamandis
- Singularity University
Your Checklist of Actions to Take
Make an effort to learn how to gain insights from my data analytics. Through training, I can become better at decision making after looking at the metrics presented to me.
Create cause charts, flow charts, quality scores, etc. and ensure that I present your data in a way that my team can easily understand. When something is represented in a simple visual format, the ultimate goal becomes clearer to everyone.
Utilize tools that will help me achieve my PPC goals. Grab the complete recommended list by Frederick Vallaeys on his episode’s show notes.
Constantly test ads so that I can cancel out what doesn’t work and build on what works.
Get familiar with the Google Ads terminology. Explore its components and learn how they can help you create better ad campaigns.
Automate my campaigns and processes so that I don’t have to deal with constant, tedious work to monitor the performance of my ads.
Do keyword research first before I run a campaign. This will eliminate several trial and error strategies that most likely lead to nowhere because of a lack of research.
Understand bidding and other factors that affect the cost of my ads so that I can weigh the performance of my keywords better. A cheap keyword may turn out to be expensive in the end because of a low click-through rate
Determine my product margins before I start running my ads. An advanced insight on my profitability will give me a clearer projection on what is yet to come.
Check out Optmyzr to learn more about the tools that can help you run more efficient Google Ads campaigns.
About Frederick Vallaeys
Frederick was one of the first 500 employees at Google where he helped build AdWords. Today, he is a cofounder at Optmyzr where he creates tools for PPC experts.