Our next bonus episode of the PPC Ponderings Podcast is now live, and the amount of knowledge shared by our guest is remarkable!
Brad Geddes is the co-founder of Adalysis, has written one of the most trusted PPC books in existence, and is well-known for his insightful takes on PPC (since, ya know, he's been doing it longer than the rest of us!).
Come listen to Brad and Kirk chat about his take on the keyword’s past, present, and future. Brad goes into so much more detail than we had time to include on our original episode, so make sure you get the rest of his hot takes here.
Brad was one of our guests on the first Ponderings episode, and we’re excited to share our full conversation here!
If you haven't already, make sure to catch the first full length episode here: Episode 1 - The PPC Keyword, Past, Present, & Future
In this podcast, we are getting away from an interview approach, and come from more of an investigative journalism format. Enjoy!
Brad Geddes has been involved in PPC since 1998. He is a co-founder of AdAlysis, a PPC recommendation engine. Brad is the author of Advanced Google AdWords, the most advanced book ever written about Google's advertising program.
Brad has worked with many of the world's leading companies in managing and perfecting their PPC management and workflows.
One of his trademarks has been demystifying the complicated aspects of SEM. Not one to hold secrets, Brad prefers to educate marketers on the various aspects of crafting successful marketing campaigns to ensure the success for all parties involved.
Brad Geddes (00:03):
I mean, the first ad was, gosh, in what? Egypt three, almost 4,000 years ago was when the first ad was created. When you look at the history of stuff, nothing touches a keyword. The thing that we're going to see marketers do more and more is become a bit more sophisticated in semantic analysis of search terms.
Chris Reeves (00:36):
Welcome to the ZATOWorks PPC Ponderings Podcast, where we discuss the philosophy of PPC and ponder everything related to digital marketing. Today's show is a bonus episode of our full interview with author, speaker, and co-founder of Adalysis, Brad Geddes. Brad was a huge part of our first episode on the keyword. If you haven't heard that episode yet, go give it a listen. Otherwise, enjoy our behind the scenes conversation with Brad.
Kirk Williams (01:06):
Okay. So one of the things I really wanted from you specifically, being the godfather of PPC, as I like to say, the godfather ofPPC, but not in the bad mafia way, so can you talk through... Let's go back in history, can you kind of talk through the early days of paid search, and specifically I'm curious to know, I did a little bit of research on the keyword, where it came from, that sort of thin, but I'm just curious to knowfrom your take, especially kind of being there through all this, what do you remember about the early days of the keyword, if you remember how it got started, just kind of all of that stuff? Maybe just share your thoughts and memory.
Brad Geddes (01:48):
Sure. Gosh, so 1997, Bill Gross out of Idea labs founded the idea of paid search. Essentially, he knew people were typing in keywords and he got, I don't know if it was a patent or it was a patent pending or whatnot, on it. And that was initially named goto.com.
Brad Geddes (02:12):
And so, in, gosh, 1998, when I opened my first paid search account, we actually had to call them, so they didn't have online stuff open.And, at that point in time, no one really knew what this was. So you had a lot of brand marketers who used to measure impressions, and so they would buy terrible keywords for the point of, if no one clicked, they would get impressions.
Brad Geddes (02:39):
And so, relevancy, it sometimes was really bad. And you only had, you had a bigger list of keywords, and you could have ads that go to individual keywords. Whatever you bid, it was a simple auction where you paid a penny more than the person below you, all bids being exposed.
Brad Geddes (02:57):
So one day, I'm working with a plastic surgeon. And you would often watch how stupid people would bid. And so, his competitor came in, and his average CPC is about $15, which even back then was a pretty pricey CPC.And one of his competitors bid 100 bucks because it was a highest you could go then. And so, I'm like, "Well, you're just being annoying," so I bid 99.99. The next bid being about 15 something dollars.
Brad Geddes (03:24):
And the doctor called me one day and said, "Brad, I got the weirdest call. I got a call from my competitor who asked us to lower our bids," because people would call, and then, he's like, "What's going on?" I'm like, "Here's the deal. He's paying $100 a click.You're paying about 15. I could change this, or I could let this ride."And he laughed and hung up the phone. And so you had things bid jamming and gap surfing were bid techniques that have just been lost over days.
Brad Geddes (03:56):
About, oh, gosh, maybe a year-and-a-half later, Goto.com launched, the first bid system. And at first it was 12 bid changes a day. And so, you would sit there and watch when people changed their bids and make time and make timestamps, saying, "Okay, here's when people change their bids. We're going to bid after this by five minutes." And you could literally graph out when people changed bits.
Brad Geddes (04:27):
And so, this then went on for a while, a little more advanced, but not a whole lot. It turned into Overture, which then turned intoYahoo Panama for your old school people who remember what Yahoo Panama was before they dropped it, and it became just Yahoo paid search.
Brad Geddes (04:49):
Now then, in 2002, Google initially launched paid search, but it was a CPM-based system. And so, it was a system really, because everyone was still CPM world, you'd pick a keyword, it was at least $30,000 per quarter, it was on a contract, minimum. You'd pick your keywords, and you'd just show a banner at the top.
Brad Geddes (05:13):
And for some words, this was incredibly cheap. So suddenly, you're buying CPM, and maybe 10,000 a month sounds a lot, but not when you could literally get 100,00 clicks for this because Google didn't know any better. And so, then later in the year, they launched their self-served paid search program, and they did two things which really sort of changed the world of paid search.
Brad Geddes (05:41):
One, they had a better concept of ad groups, which we still think of it's the ad, what's in the ad group describes the ad. The ad is what really matters in that. And the auction, initially it was click-through rate times cost-per-click. So click-through rate being, hey, this is how relevant your ad is. If no one clicks on it, you pay more. We're not going to show you.
Brad Geddes (06:05):
And you had to even maintain minimum CTRs, which meant you could have keywords that were too broad, couldn't get your CTRs, they'd be disabled. That whole auction morphed into Quality Score, which of course, we all know what Quality Score is today. And they brought the landing page in and just had various iterations over the years.
Brad Geddes (06:24):
But that was what Google brought to the table more than anything else was this concept of the ad groups and the CTR or Quality Score. And then,from there, we just layer on larger scale automation et cetera.
Kirk Williams (06:38):
What is your opinion, especially, I guess, early days, soGoogle's claim right from the beginning was Quality Score was to help the small business, the small advertiser who couldn't just kind of dump money... That seems at face value to make sense. What's your opinion on Quality Score, and their originally bringing it in? Do you think it's a good thing? Do you think it has negatives to it?
Brad Geddes (07:03):
I think it's a great thing. I may not always agree with how it's calculated. Let's just put that aside for a second, all right?
Kirk Williams (07:09):
Brad Geddes (07:09):
But overall, the idea, we would have people who could spend $1,000 a month who would beat people spending $1,000,000 a month because they were more relevant. They were more useful. And I think it was a big picture idea from Google. Google is good at thinking 10 and 15 years in advance, which is tough in a digital world.
Brad Geddes (07:31):
And so, the idea of to switch... For a user to not become interested in something, you see bad search results, it's a URL two seconds later. You're on a different site.
Brad Geddes (07:47):
I remember I had done some consulting for Google in 2004.And the first question that Sheryl Sandberg, who was still running Google back then, asked me was, "Since an advertiser only has to hit a pause button to stop advertising, and a user just has to change the URL, and we can't serve ads, how do we maintain relevance?" And they were really focused on, on this.
Brad Geddes (08:09):
And so, I think Quality Score really comes into, if we don't have that good user experience, we'll lose people. I mean, Google still today will not show ads on results that no one wants to click on them. And, yeah, we could argue they show more ads than they had before. But in reality, they've always had this sort of end user in mind. And so, you may not always agree with the methods, but the ideas, I think, behind it are really good fromGoogle.
Kirk Williams (08:37):
Yeah. Do you see Quality Score disappearing in the future as things go more automated? I mean, it'd be interesting to see what happens with that as things become less about specific keywords selected, less about even specific bids. But if Google has their way, things are going more of a we tell them, "Here's our budget. Here's some level of goal that we're trying to hit. Here are general information about our business. Now, go find us, people." Is Quality Score still needed in that auction environment?
Brad Geddes (09:13):
From a landing page stand point, definitely, like are people actually staying on the website or not?
Kirk Williams (09:17):
Brad Geddes (09:18):
One. Two, I think we're further away from that than a lot of people believe because Google is terrible at looking at a website and knowing all the stuff you want to advertise for. Or they don't get things like, when I lived in Chicago, everyone who's a roofer in the wintertime is actually a basement specialist in the wintertime because you don't work on Chicago roofs in winter. But that's not true in Miami.
Brad Geddes (09:50):
And so, this is where Google doesn't understand about businesses. They still need to be told what's going on. Otherwise, you're still going to do the work. You'll just do it on the website as opposed to in your ad campaign. If Google's going to say, "We're just going to pull your headlines off of your pages to make ads," well, okay, we can still control it. Let's go really focused on what those headlines on the website are.
Brad Geddes (10:13):
So Google's not, I mean, they don't make up new words.They don't make up ads. You still give it the assets or it borrows the assets from your site. So, in reality, they're creating nothing from that standpoint, from a content standpoint.
Kirk Williams (10:33):
Yeah, which is definitely what RSAs are, right? I mean, it's just us giving them content, and them then utilizing the data that they get off of those to kind of show which content goes. Okay, interesting. So skipping ahead a little bit, kind of thinking the future of keywords a little bit, what do you see as the future of keywords? What's going to happen?
Brad Geddes (11:00):
You know what? I mean, there's sort of two ways to think of this. One, you still give keywords to Google because Google just doesn't know everything about your business. I mean, that's just one. Two, Google's like, "Nah, we'll take care of it," but of course, they're going to pull it from your website. So you focus more on going back to meta keyword optimization, which hasn't been done in, I don't know, 15 years.
Brad Geddes (11:26):
And so, in many ways, because we know how poor Google can make keyword relationships, they're going to want some insight from people. And part of it is brand protection, part of it is nuanced business methods. I mean, if you talk to people in business-to-business, they're not the same accounting keywords as in business-to-consumer. They're completely different words even though they're both accountants.
Brad Geddes (11:54):
And so, it's those subtleties that I think some human input is still needed. I mean, we've had TV buying for 70 years, and yet you're still looking at what's the demographics of the channel? What's the demographics of the actual content I'm being put on? What's the time delay watch versus the real time watching? And so, you still have metrics for buying TV, and that's got years of optimization behind it.
Brad Geddes (12:22):
I think we'll see a lot of that. I mean, maybe eventually, yeah, if you say 30 years, who knows? By then, we may be answering to machines. But until then, I think we've got some human input still left to give.
Kirk Williams (12:34):
Yeah, I would agree. I think there certainly seems to be a balance. At some point, you have this idea that Google is running all of these things in an auction on the backend that we don't have and can't have access to. There are millions of data points, as they'll say.
Kirk Williams (12:51):
And I think there's a lot there. I think there's a lot of truth to, yeah, if we can nail down our goal inputs like conversions, things like that, that we're saying, the more we can communicate to them, "This is what we want to achieve," the more they're going to be able to assist with that. But I almost see them right now pushing a little bit too aggressively to remove what you just noted, which is still the helpful, important user input.
Kirk Williams (13:20):
Because even if their machines get keywords and close variants matched correctly 80% of the time, they just do, let's just say, a bang up job. They're just killing it 80% of the time. What about the 20% of businesses who are just getting slaughtered with those keywords that are not matching up well through all those things that you had noted? And you have those variables in that that I think are just still really important, which is where humans come in.
Brad Geddes (13:51):
Yeah. And I think we can even go beyond that. So let's say you're a national company, and you want to focus on California for a while. You want to shift focus there. Google doesn't know this. Or if you're a drug company or you run TV ads and you just made up a new slogan, you want that as a keyword because it's on the TV now. And so, how does Google find that when you're first doing it?
Brad Geddes (14:15):
So I think there's a step in the middle here, which is where we can feed better data into Google. So we have, I mean, a lot of clients, who they know better than Google about their target demographic and the geos and the zip codes and the time of day, and their real time bidding is way better than Google for them only. But Google won't take that data.
Brad Geddes (14:42):
And so, I think there's a middle step of us closing a data loop at times of saying, "Hey, we're doing predictive bidding for some people." So if you're going to do predictive conversion rate bid-based systems and not just react to things, we can technically bid better than Google right now. But if Google could take that, they're better at machine learning than we are by far, but they won't take the inputs. So I think we have another step still, which is better data integration across systems.
Kirk Williams (15:16):
Yeah, that makes sense. Okay, so thinking about the keywords, so we're going to talk a little bit about close variants on the episode, man, maybe just give us a little bit... It can be a short or long history of like close variants, what those are, but then also what are your thoughts on them? What are your opinions? Is there some benefit to them? Are there negatives? Hit us with what you got about close variants.
Brad Geddes (15:44):
Sure. So I mean initially when exact match launched, and we'll stick to Google now and not make this multiple platforms, right?
Kirk Williams (15:50):
Brad Geddes (15:50):
But when Google first launched exact match, it was those exact characters, so you needed every misspelling, every singular, every plural, and every variation to show up. And, of course, that's a huge amount of work. And we used to have to do keyword research for misspellings. And so, then Google launched, I do not remember the year, probably '08-ish or so, could be a little earlier, close variants, which were essentially, let us take care of plurals and misspellings and nothing else.
Brad Geddes (16:24):
And so, for some people, this was a nightmare in itself just because an S versus not an S on the end of the word sometimes makes a massive difference in what the data looks like. It was crazy.
Kirk Williams (16:35):
I remember you specifically sharing certain client stories of ways, I don't remember what they were, a law firm or whatever it would be, where you'd be pulling data, especially when some of those CPCs were very, very high saying, "Look, the difference between a plural and a singular is massive here for this specific client," yeah?
Brad Geddes (16:57):
Oh, yeah. Yeah, and it is all over the place. And if you needed to, you could manage negatives, but it was good for 95% of accounts, so,I mean, it was not a bad change. It just had nightmare implications at times.And we went on, we lived in that fun world for quite a while, until only three-ish years ago, probably 2018, 2019, somewhere in there, that Google decided, hey, let's match you to words with the same meaning.
Brad Geddes (17:25):
Now, this is when we started seeing the real problems with variants, and variants also were, we can reverse the word order if the meaning doesn't change. And that's when you get into contractor license or licensed contractor, deck paint or paint deck that Google missed a lot of them, and they still do today, where its same intent is very interpretive.
Brad Geddes (17:51):
And they did that for exact match first. They did it later for phrase match. And then they, the second change only a year-and-a-half, two years ago was complete word substitutions. This word means the same thing, so let's not worry about what the word is itself.
Brad Geddes (18:12):
Now, in some cases, NYC, New York City, I could see where that does have some good things. Part of the problem is when they just make some bad semantic analysis. I mean, they were deals, sales, packages mean the same thing in their mind. And so someone who sells tires rims for cars would put packages to make it easier for consumers, but Google assumes that's a sale. But it might not be a sale. And that's where things got really, really ugly, and that has not changed.
Brad Geddes (18:47):
And then, in the last six months or so, Google has expanded what is the same meaning significantly. So if I say I have a keyword home insurance quote exact match, to Google, that is how much should I pay for home insurance is the same exact keyword, except someone saying quote has a high intent to convert, someone who's just saying what's the average price for something has no conversion rate intent.
Brad Geddes (19:23):
And Google's not using machine learning to make this better. I mean, we have a specific keyword which spends close to a million dollars a month, so it's a lot of money on the keyword. The variants of who, what, why, when, how, spend 300,000 a month for this one keyword. Their CPAs are about 400% higher than if they're not included. Google's never figured it out.
Brad Geddes (19:51):
Now, they have the data too, which tells us they're not trying to fix it. They want those impressions. And so, a lot of times what we're looking at is not just where Google gets it wrong, but [inaudible00:20:04] they have so much data, they're not trying to make it right? And a lot of the variant stuff, they're not trying to make it right. They're letting it run. And so, we'll see how that goes.
Brad Geddes (20:15):
But when we say mobile conversion rates have gotten worse, it's not for the same words. It is for all your people asking questions on their phones triggering ads who have zero intent to actually click on an ad and convert. They are asking a general question, and that's where we see a lot of mobile conversion rates plummeting in the past six months.
Kirk Williams (20:39):
So in those examples, so you're not seeing evidence, at least at this time, that they are applying any machine learning at all, in those examples, like that specific keyword you shared?
Brad Geddes (20:51):
If anything, they're applying machine learning to attempt to get more impressions, not necessarily always make the impressions better.And, by the way, the example I'm giving is a target CPA account. So they have a target CPA set. Google technically is optimizing the bid in CPAs, but because they're moved to a portfolio system not a keyword based system for CPA bidding, they don't care if a bunch of queries are worse.
Kirk Williams (21:18):
I mean, that is an interesting example because we're told, and I think there is a lot of truth to that... We've changed in our agency specifically to rely more on automation than we had in the past, smart bidding, things like that. We've seen evidence of it working better.
Kirk Williams (21:38):
And yet, those are instances where at the very leastGoogle is in no way helping their cause as they are trying to attract advertisers to use more machine learning. They're trying to spread the gospel of machine learning. Those are great examples. And a lot of us will see examples like that in our own accounts.
Kirk Williams (22:00):
And it's not just simply because PPC advertisers have their heads in the sand, they don't want to change, things like that. We can often see examples like that where we'd say, "Okay, maybe the argument would be, well, right, with machine learning you have to give it enough time to learn. At some point a million dollars a month on a keyword, months go by, specific target CPA goals," I mean, it doesn't take a rocket scientist nor should it take an advanced machine learning model to figure out that. So certainly, there's concern there. That is interesting.
Brad Geddes (22:33):
Yeah. And so, this is where, these days, we know we have to use machine learning. We know we have to use some of Google's tools. I mean, we get this. So I kind of think about it almost like a self-driving car. So you hop in the car, you say, "I want to go here, but I don't want to take these roads. I want you to use these types of signals," and you're essentially giving the guardrails to where the car can go. You're letting it steer itself.
Brad Geddes (23:00):
And that's what negative keywords and negative audiences and positive modifiers are is making sure you're still on the right road going the right way. You're putting the guardrails and that overall GPS map in place, but letting it drive by itself within your parameters. And if you're not setting those parameters, you're going to end up in the wrong city. It's just going to happen. So that's where PPC marketers are. It's really, you're setting the bumpers of don't go outside of these lines.
Kirk Williams (23:33):
So what would you say if Google said, "Okay, yep, we totally agree. And therefore, in 2023, we're going to utilize just your website as the guardrails? That's where you put the information. We'll take what you have on there, and we'll build audience list based upon who we see you're targeting, blah, blah, blah. Insert credit card, TROAS or TCPA goal, somewhat of a general daily budget, although, trust us and go at it"? What would you say if that's what they said, "Hey, your website can be our guardrails"?
Brad Geddes (24:13):
Then I would want to be able to feed it data that it can't read on the website. So let's say you've got PDF downloads behind pay walls. SoI don't want the user to find it, but I want Google to read it for keywords.Therefore, I still need a way to feed it data that's not publicly available.
Brad Geddes (24:30):
So for a content site, sure. E-commerce site that puts everything out front, sure. I can see that being doable. Business-to-business websites with long sales cycles is if we've got a three year sales cycle for a product, is Google really going to use three years of data? Probably not.
Brad Geddes (24:49):
Therefore, we need to give it incremental data. We need to tell it how much it's worth. And then we have to feed it stuff that we don't want publicly available, or maybe we want it available behind like registration walls or something. And that's where the downside of some of that comes to is not everything's open.
Brad Geddes (25:07):
Now, secondly, what about Google competitors? Are they going to let Google do this? So if I am Microsoft, I'm Adobe, I'm any mapping application, I'm whatever, and right now, I don't want to feed all my conversion data back into Google because I don't trust the fact that they won't actually go read my own data to compete against me. Therefore, do we not get to advertise anymore on Google? Do we have some controls? What does that look like for the big corporations?
Brad Geddes (25:45):
That's something else that I think sometimes we sort of miss is once upon a time. And this just goes back to how long I've been in this and how the world's changed, our Yellow Page, I had one company bought byYellow Page company. And our website in region did more queries than GoogleMaps. We blew away Google maps in region business queries. It wasn't even close.
Brad Geddes (26:08):
Right now, Yellow Pages, for reasons, they don't have that volume anymore, deservedly so. But you think we are going to give Google all that data? No way. We were the big dog in that region, so we're not going to hand it to the little dog. Now, granted, they ate us, but that was deservedly so. But there's a lot of other things that are big implications.
Kirk Williams (26:30):
Yeah, yeah. Yeah, very, very interesting. At the very least, for sure, again to your points of data inputs of some sort, at the very least, exclusionary inputs would be really essential.
Brad Geddes (26:44):
Kirk Williams (26:44):
These are not the terms. This is not the sort of audience we want. And that's something that I have found difficult always to summarize or to put into words. But I think it's really important that engineers can sometimes miss... I find when you talk to engineers, they're brilliant, but they very, very often see things in such black and white. And I think they start often with this assumption of, well, I can somehow write a code that will eventually see every single aspect of that and then make an informed decision.
Kirk Williams (27:25):
And I just see the world and business, people as being more complex than that, at least for a while, to the point of we're actually really, really far from true AI and that sort of thing. And so, that's where it's not just a matter of looking at specific keywords, and do we want to include or exclude these?
Kirk Williams (27:51):
But, like you said, there's so many other aspects of business that are made as you're talking to someone in the boardroom or what have you that then works into my decision to go into an account and then select those keywords to target or not target.
Kirk Williams (28:07):
And I'm not even fully sure how you would feed that into a system, which is why just handing someone some money and saying, "Here's a basic, like a website or a basic level," just couldn't begin to include the complexity of true business models and relationships when they're done well.
Kirk Williams (28:27):
Okay. Well, let's see. Let's look at just the keyword.What do you like about the keyword itself? Why is it a great targeting method?And then, maybe does it have any limitations?
Brad Geddes (28:41):
You know what? So the keyword, to me, is the absolute best targeting method of any advertising in history. It is someone who's sitting down saying, "I want to know about this right now." I mean, they're telling you what they want, and then you could read data signals about who they are, when they want it. I mean, there is no targeting that comes close to this.
Brad Geddes (29:09):
So the whole when people are saying, "Oh, we'll just do audiences only." No way. An audience says, yeah, they may care about this, but they didn't say, "I want this now." So, I mean, the first ad was, gosh, in what? Egypt three, almost 4,000 years ago was when the first ad was created. And so, when you look at the history of stuff, nothing touches a keyword, but it has huge limitations.
Brad Geddes (29:36):
You have to know what you want, and be able to type it in. For brand new products, there's no search for them. No one knows you exist, and that's where it does not create huge demand. Now, someone may see your site and start searching brand over time. We've seen brand growth through search, but its biggest limitation is you can't influence someone that they might want to search for this later. That's TV, radio, video ads, whatnot. You can't introduce a brand new product if there's no market share for that item yet. And that's where it is a very difficult thing to work with.
Brad Geddes (30:17):
So when you have something that is not new, when you have something that is being supported by other ad channels to bring awareness to it, the keyword is, I think, the best ad method or targeting method in history.
Kirk Williams (30:33):
Totally agree. Love it. So yeah, the idea of demand creation, demand generation, as opposed to, whatever, demand capture, however someone wants to say it, can you utilize PPC to generate demand or no?
Brad Geddes (30:52):
Well, okay, so let's stick with search because obviously paid search-
Kirk Williams (30:55):
Brad Geddes (30:57):
... [inaudible 00:30:57], totally possible, right?
Kirk Williams (30:58):
Brad Geddes (31:00):
In some instances, it can be done. I remember we were working with one of the early vampire power companies, which all that really means, and everyone's power strip has this now, you can make an absolute power turnoff. Your TV's not draining power when you're not there. And the problem was you could advertise for powers strips and this and that, but no one really knew, no one searched for power strips. That's such a boring thing. They just went to Best Buy at that point in time and bought power strips.
Brad Geddes (31:33):
So we started doing things based upon saving electric bills, lowering my electric bill, and a lot of stuff in that area, and they sold a lot that way. And so, yes, you can, if there's another way people are searching that might introduce something there.
Brad Geddes (31:53):
But think when phones first came out, no one knew to search for a cell phone. How do you advertise a cell phone 25 years ago? No one even knows it's a category. When a new TV type, plasma or LED is introduced, yeah, you can advertise on big screen TV, but they're a very competitive market with the ROAS right now. There's not a lot of sort of empty queries that aren't at their almost max capacity. So it's difficult at times to jump into a market unless you can burn some money and advertise brand new products.
Brad Geddes (32:30):
So it can be done in some cases. I would never make the statement it could be done for everybody. I think there's a lot of examples it can't be done.
Kirk Williams (32:39):
And some of what we find is, not that it's completely impossible to do, as you had pointed out. Sometimes it's just, yeah, thinking of creative ways that people are interested in a problem, and then presenting yours as a solution, right?
Brad Geddes (32:52):
Kirk Williams (32:52):
The problem more and more in 2021 paid search is, I think, cost. Oftentimes, the CPCs that you're going to be hitting to demand gen with paid searches, that's where I would say, "Okay, take that and go use... go target some audiences on Facebook or what have you."
Brad Geddes (33:12):
If someone makes a brand new email system or an alternate to email system, the new, we've got Slack stuff. We have email. So I'm going to make something brand new that's even better. Well, email's so expensive, andCRM stuff is so expensive, and Slack stuff is so expensive, there's no way to break in the market on search. I mean, it's possible, but you're going to lose alot of money doing it.
Kirk Williams (33:39):
Yep. Where do you see privacy, if at all, impacting the keyword? Okay, well, maybe there's two questions in there because it's already reportedly impacting it, right? Google has claimed that in terms of the search terms report and pulling back and then giving us some back, but most of them are no click. So the privacy and the keyword have already started to mess with one another. Where do you see this going at this point with that?
Brad Geddes (34:06):
I mean, so the privacy and the keyword, it's a tough one because not everything is always classified into queries that shouldn't be tracked. Medical, legal being probably two of the top ones that get iffy how you're tracking getting into them.
Brad Geddes (34:25):
It's not the keyword itself that causes the problem. It's the audience and other fingerprinting data that often go with it, and that's where the issues start to play, or the user's journey over time back to these specifics sites, and you can see the whole thing. That's where more the issue comes into play.
Brad Geddes (34:46):
So if, let's just mad world, if Google said, "Hey, you can pick your exact keywords. We won't expand them. You can only advertise on this. If you do, you can put a conversion pixel on your site to see conversions, but we're not going to give you full data because we want to protect user privacy. So you can't time stamp it. You can't do that, but you can have any word you want to"? Would people take them up on it?
Brad Geddes (35:16):
I mean, it's a good question because, yeah, if you get full control, great, that's what marketers want. But if you remove some of our middle data, do we suddenly say, no, give me less data, but give more, and I know I'm getting less data because of privacy, therefore, you expand my keywords for me. I don't want the misspellings, so I'll take everything you want to give me. Or, hey, we'll make it perfect for you, but you don't get this in the middle. But you'll get conversions and sales at the end. Which world do you live in?
Brad Geddes (35:50):
And that's sort of like almost a trade off at times. And so, Google's decided, let's expand this for you. Let's give you as much data as we can and respect privacy in the middle, knowing markers would be upset about some of our targeting methods.
Brad Geddes (36:08):
And it's a tough trade off, and I don't think there's clear cut easy answers to a lot of these things. Consumers say they don't like ads except they do if they're relevant and useful, and search is among the most relevant, useful of all the ad formats you have, right?
Kirk Williams (36:29):
I've noticed a very interesting and not a desired trend lately.And maybe that's just the world I hang out, especially with like D2C operators, direct-to-consumer for those listening. So it tends to be more e-comm, whereFacebook especially has been such a core part of the paid media and a lot of these D2C operators, they've really grown a bit because of Facebook over the last few years. It's kind of been this little trend of people talking more and more about ads and paid media.
Kirk Williams (37:00):
And I've noticed, oh, hey, I think they're actually referring to social. They're referring to Facebook as kind of like paid media.And in that sense, the privacy conversation, I'm hopeful that those who have to begin regulating all that, they're going to also start to separate those because there is a significant difference between paid search and paid social.
Kirk Williams (37:24):
And again, that goes back, to me, what the core of paid search is, someone typing in a query that they literally do want an answer for.And a great advertiser is going to provide them that answer, and that itself is where Google and the advertiser and the user can all be kind of perfectly aligned there.
Brad Geddes (37:42):
Kirk Williams (37:43):
And, whereas Facebook, it's kind of those social things, and in some ways where a lot of people have gotten frustrated with, well, I hate ads. Well, if you're talking about you're trying to figure out whatGrandma Jones is doing and you keep seeing these annoying ads for these companies that you may have never heard of or guru courses of someone trying to sell you with their Lamborghini behind them, right? Yeah, I agree. Those are annoying too. Let's-
Brad Geddes (38:08):
Facebook's got great audiences. And so, when it comes to privacy, that's the real question is audiences. But when we say search, I mean,Amazon's a search engine. YouTube's a search engine. So when you say advertising on Amazon, you could do display in Amazon just like Google, or you could say pure search, and that's the same relevancy. So there is good search well beyond Google, where, again, Amazon, that keyword's really important.
Kirk Williams (38:33):
I like it. Do you have any other final thoughts on the keyword or on anything we've talked about? Otherwise, I think I'm very happy with what we got.
Brad Geddes (38:43):
Yeah. I mean, the thing that we're going to see marketers do more and more is become a bit more sophisticated in semantic analysis of search terms. So we've seen this, things with like say n-grams. More people are understanding the reason to look at n-gram based data for either, A, new products or, B, negative keywords.
Brad Geddes (39:10):
I can tell you many stories about slight misspelling differences or changes to singular plurals of how that affected users. And the problem is most marketers have not done a full semantic analysis to understand, oh, you know what? If we're targeting doctors, and someone misspelled a weird word medical condition, they're probably not a doctor. Therefore, we really don't want to focus on any misspellings.
Brad Geddes (39:37):
And it's those small little things which sound good when you hear them that are not being done quite as much. So I think we're going to see more marketers embrace semantic analysis of their terms to pull more insights to then better inform Google of what to do and target.
Brad Geddes (39:59):
It's one thing for me to say, "Hey, we do n-grams and we see these trends Google's not doing," but we're one voice among three million advertisers or whatever. So it's pretty drowned out. But if 100,000were doing it, and saying, "Hey, here's the n-gram trends we're seeing, why aren't you doing on n-grams, that makes a big difference."
Brad Geddes (40:20):
And so, a lot of changes Google makes are marketer driven.And so, the more sophisticated marketers get in understanding, not just the keyword, but the search term and the insights of how slight user differences in words makes a large difference to their end audience and their conversions, that's going to inform machine learning and push Google to use better semantics and just be better at semantics. I mean, that's one of their weak spots, I believe. I really believe their weak spot is semantic analysis.
Chris Reeves (40:54):
This has been a bonus episode of the PPC PonderingsPodcast. Keep checking back for more interviews and our next full episode. If you like what you hear, please consider sharing this with your network, leaving us a review on Apple Podcasts. Until next time, may the auctions be ever in your favor.