
Episode Description
No one has a crystal ball that tells the future, but we’ve got the next best thing: geniuses.
Hear what some of the brightest minds in PPC have to say about what PPC will look like in 2021, including where Google will go and how quickly they’ll move.
This panel covers:
- What will Google automate next
- How our experts use automation and what they still do manually
- Launches from 2020 that will have the biggest impact on how we do PPC in 2021
- Expanded Text Ads vs. Responsive Search Ads
Episode Takeaways
What Will Google Automate Next
- The panel anticipates further developments in Google’s automation, particularly around keywords and match types, predicting a gradual shift from advertiser-defined keywords to Google’s AI determining optimal keywords based on performance data.
How Our Experts Use Automation and What They Still Do Manually
- Experts utilize automation for bidding and ad testing but emphasize the importance of human oversight in strategy and data interpretation. They advocate using automation to handle repetitive tasks while allocating manual efforts to strategic decision-making and complex problem-solving.
Launches from 2020 That Will Have the Biggest Impact on How We Do PPC in 2021
- The introduction of Responsive Search Ads (RSAs) and changes in Google’s approach to match types are highlighted as significant. These changes require advertisers to adapt by relinquishing some control over keywords, thus relying more on Google’s AI to manage ad relevancy and targeting.
Expanded Text Ads vs. Responsive Search Ads
- A debate persists on the efficacy of ETAs vs. RSAs. While ETAs allow for more precise control over ad content, RSAs offer potential for higher performance by enabling Google’s AI to optimize ad combinations based on real-time data. The consensus suggests a mixed strategy, utilizing both to maximize coverage and effectiveness.
Episode Transcript
Frederick Vallaeys: All right, we’ll see you for the next episode, thank you
Brad Geddes: to kirk. I see right. This is what everyone should be reading these days Pbc years united. Yeah
Matt Van Wagner: Let’s let’s let’s look forward to the new thing. Let’s go. Let’s hail to the chief, right? And we have to go
So tell your friends to go watch it We’ll be back in two weeks with the next ppc town hall same date same time And we’re going to talk about smart shopping campaign shopping campaigns e commerce have some great panelists on that Matt, can I ask you a final favor? You want to play us out? But what do you want to hear?
Frederick Vallaeys: Yeah. Strange new world from one strange year to another strange year. So on that note, thank you so much for being on thank you everyone for watching and the amazing engagement so this video is going to live on youtube.
But yeah, it’s the, the world of real data in GA is sort of, it’s, we’re now fuzzy analytics, right?
Matt Van Wagner: Yeah. And you’re counting people, the people that love to, you know, the people that chase like your bank account when it’s one penny off, you know, even though it’s a million dollar account, you’re counting people are going to go Lulu with with Google analytics and conversion modeling and and but it’s, it’s you know I think if they’re modeling reflects your reality or their modeling reflects your modeling, then, you know, you’ll, you’ll develop some confidence over time.
I mean, they, they, it’s in their best interest to overestimate. Yeah. How we start going back again, right? Like we used to and say, here is what they say. Here’s what we say. So every time we look at numbers, we need to divide it by, you know, 0. 6 to get the real information. I mean, I think that basic, it’s not perfect math, that fuzziest math is going to be important for a lot of companies to sort of get back into that mindset with more model data than actual data.
We assume some other platforms will start modeling conversions. Now, when Google was the first to do this back in GA1, right? If they sampled data model conversions and they always severely overestimated. So what we’re looking at is, Hey, how do we actually do our own modeling again of conversion data to compare against them to get some sort of here’s how much they’re, cause we assume they’re going to overestimate.
Brad Geddes: One of the things we’re kind of looking at is, okay, so we know the cookie loss, right? Facebook’s going to a seven day attribution and they’re going to stop modeling conversions.
Frederick Vallaeys: even know, right? What’s going to happen.
Look, if anyone comes and says, I know everything that’s happening in this space run because no one doesn’t
Ginny Marvin: Well, and there’s not a, I mean, a lot of this is still so unknown that It’s okay.
Frederick Vallaeys: Yeah. I mean, honestly, this is a complicated space and I think a lot of people, myself included, don’t understand a hundred percent.
You know, I think that’s kind of another sort of back to basics analysis to start looking at as well.
Because I, you know, I, I think this has been sort of the. You know, the grab bag of data. And so really think about data that you have, the data that you use and what you are not using and can stop collecting. And, and, and be able to communicate that to stakeholders that We’re losing this data, but it’s not going to have an impact or we’re losing this data and it is going to have a X, Y, Z impact.
Ginny Marvin: Yeah. But I also think that this is an opportunity to look at the data that you are collecting and what you’re actually using.
You know, walled content and the subscription stuff, you know, get people that, you know, log in and have a reason to log in so that at least you have some, some more you know, recent data to work with, even if it’s not, you know, from Google or Facebook, just use your own data.
Matt Van Wagner: Well, yeah, yeah. I mean, it is. It we went a little bit too far field on how we did deploy data we had on people but it is what it is so as far as I think one of the other things you can do to sort of mitigate, you know, cookie losses, look at ways to create.
Ginny Marvin: That’s because we abused the system. It’s our own fault.
It’s like on this circle. And I’m sure there’s old stuff about measuring engagement and user engagement. People went, that’s not good enough. What are you doing that for? That is suddenly relevant again.
We had keywords, we didn’t have queries and we had hits and engagement metrics. We didn’t have, you know, phone call tracking. And then we get all this data. And now it’s going back to engagement metrics from which we used in 1999. And it’s going back to lack of queries, back to keywords. We had a 98 in paid search, right?
Brad Geddes: Yep. I hear this stuff and I almost feel like I need to go back and read like search engine strategies from 2000, right?
Frederick Vallaeys: Yeah. Like they say data is the new oil, right? So you can only get that data the moment you set it up. So do that. Right. Even if you’re not going to use it for another quarter, you’re not going to use it. Just get the property set up.
We’re just we’re you know, we’ve got all of our all of our clients set up with it only one or two that were actually sort of actively using that That data with the rest is we’re just kind of getting our game on. And I think getting your eyes used to the differences is kind of important to do. So if you haven’t already gone to GA4, definitely recommend after this call, just go get yourself a property set up and, and start it now.
This is going to become part of your conversation over the next. You know a year or two but I would I would say fred that that would be a good topic to put on a town Hall, I don’t know if you’ve got already got your schedule set But yeah, I think I think and there are a couple of people i’d recommend to you that would do a good job at that we’re we’re just in it.
And things that happen as opposed to discrete tracking that the cookie loss prevents. So it’s meant to sort of set up a post, you know loss of cookies data environment. So the interface will look a little bit different. You’re going to start to be talking with your clients about engagement metrics, as opposed to hard conversion things.
If you want to a pretty good primer on it take a look at Ken Williams site. Kenwilliams. com. But he got, he got a lot of early looks and he’s got some, some really good tips on how, how to transition across you know, things you need to know about things are going to look different, but the, the, the, the main things that are going to be looking different are this, this fuzzier concept of events.
Matt Van Wagner: Yeah, you create a new profile, right? But the path is pretty good.
Frederick Vallaeys: well set up, right? It’s basically you have an existing profile and you you transition that one.
It’s not going to bring historical data in. It’s sort of new, must be new data formats, new data structures that they need to sort of have you And it’s very
Matt Van Wagner: Sure. Well, the first thing about GA4 is if you have not already created properties for all of your, your various clients and accounts go in there and create one, you know, just because the reason is that the data that GA4 will be using for AI and everything else going forward starts when you create that property.
We’re going to talk about that for a whole hour in two weeks So join us for that one. But yeah, tracking changes, loss of data, GA4 I mean, Matt, is that like a topic that you want to kind of give us the high level on?
Frederick Vallaeys: Yeah, we do. So today is the session where we’re kind of like lamenting Google ’s decisions. In two weeks we have PPC Town Hall. About e commerce and shopping ads with Google . So so Google is going to be in there But we brought in speakers who are going to tell you what Google thinks but we’re also going to get practical like and we do Have outsiders, from practitioners who are going to talk about their experience with smart shopping versus regular shopping campaigns So I see some of those topics coming today.
Ginny Marvin: we can talk about cookies. The only, I mean, I, I think watching commerce this year is going to be. fascinating. And but we don’t need to go down a commerce rabbit hole. I just think there’s so much happening in that realm. That I, I just think it’s it’s gonna be really interesting to watch.
So is there anything that the three of you think we should talk about before we sign up? Well, let’s talk about cookies then. I mean, if nobody’s like, yeah,
Frederick Vallaeys: all right, well, we’re down to 10 minutes. Maybe we do want to talk about the cookie stuff here, but. This might be a good time to start wrapping up.
Matt Van Wagner: No, I, I, I just, I had too many other you know things to do, but I, I support the effort and it’s a good crew and I really like what’s going on. So,
Frederick Vallaeys: They dunno what they’re missing.
No, no. I, no, Fred. Oh, Fred. Yeah. No, they would, they’re smart. They wouldn’t have me
I mean, that’s a, I think that’s, that’s nice. Cause it’s a pure. PPC geeking out, you know, a lot, a lot of folks you’ve gotten to know and love over the years. David Sattel, I think, and Chuni, you’re involved in that, I think, a little bit, right? Right. Yeah. Yep. I think all of us are on the board, not me. You’re on the board aren’t, aren’t you?
Matt Van Wagner: February. Yep. And that’s, that’s a nice homegrown.
Ginny Marvin: Yeah. The only thing I would add just also coming up is the paid search associations conference that will be mid February,
Frederick Vallaeys: Nice. And so, so speaking of data, let’s transition maybe into the final topic, but then like all of these privacy influenced cookie follow up with changes.
Matt Van Wagner: I think so. Yeah.
Ginny Marvin: Six of those right this year.
That’s I think it’s sort of up there. The speakers are awesome. That’s at the end of February and there’ll be a, you know, 45 minute sessions and then there’ll be an ask the experts thing at the end. But that, that should be, that should be pretty good.
If you’re working sort of tactically, you know, on the board, you know, you’re making. Quick edits. You know what sort of reporting is inside there. So what we’re going to be doing for SMX conferences for 2021 is to is to focus in on one day sort of events that drill deeply into a particular topic. So take a look at the agenda.
The we, we’ve got a, a pretty good, pretty good conference coming up at the end of February, and it’s a one day. Look at reporting both from SEO. There’ll be an SEO track. There’ll be a paid search track, and we’re gonna be looking at GA four in some depth. We’re gonna be looking at reporting things that are inside the tool.
Matt Van Wagner: Matt? Yeah, I’ll be glad to jump in real quick.
Frederick Vallaeys: And how do you educate? Let, let’s talk about that. So shifting topics for a bit, but obviously read SA search engine lanes to educate yourself. Do you guys wanna talk for a minute about what’s happening events wise for SMX in 2021? How people can get educated?
If you look into a campaign has been set on broad match and you were reaching attorneys in Palm Beach and you sell you’re a hair salon, you know that that data is useless. So I, I think all of this goes to, you know, educating your self and being a great marketer. It’s, it’s really like still so much about, this is fundamental.
Are they going to get it right all the time? No. And that’s where we come in. And again, that’s where Your own data is going to be helpful. If you come in with a campaign that has been a disaster all that data is not useful. You know, you can’t, you can’t use that data for anything. So it’s, it’s think about it the same, same way.
Ginny Marvin: Yeah, I guess my in terms of you know, the fact that we still have some agency in all of this I think is a really good point. And I, I think that coupled with use the machines and the ways that they were, are, are aimed at helping us rather than fighting them.
Frederick Vallaeys: Exactly so now if you give up the self driving car, it would drive you off the cliff in new york
Tell me to swim
Brad Geddes: I have a screenshot of one where they swim like 5 410 miles, right?
Frederick Vallaeys: Do you remember when Google you would ask for driving directions from new york to london and it would actually tell you to swim Yeah, exactly.
Yeah. Maybe it can drive itself, but you still need to make sure, right. It’s going in the right direction.
Overall, like, yes, I mean, if you’re not driving the car, the car may be able to drive itself these days, right? But you still have to make sure it’s not, it hasn’t misread something and driving over a cliff. You know, that’s, that’s your job, right? To tell it where it’s going. Is there a better route you want to take?
Brad Geddes: I think it’s a very good point.
Right. Hey, we had great, amazing conversion rates for gym equipment. Well, how many barbells can one person buy? Right. Like it’s not because there was a great conversion rate in 2020 that all that we’re going to continue that. Right. So I think that’s where we need to think about ourselves as a, as where we fit into the industry when automation comes into play,
That’s not where the machine is good at making decisions. That’s where we fit in, right? And so kind of transitioning towards that place where we stay on top of the strategy and then we figure out what are the inputs we can give to the machine that informs it that maybe the strategy has shifted the whole point about like, Hey, we don’t care about the data from 2020 because we don’t think it’s reflective of what’s going to happen next year.
And so we have to change attacks because of that. We change business operations. Right. And so I know we’re talking to PPC experts here, but there are insights that we can bring back to the teams that we work with at the corporations who we work with. But, but, but just keep that in mind. Like Brad was sort of saying, like the limitation of the machine is that it’s very good at the math stuff, but when it comes to semantics, when it comes You know, what does this mean for a business strategy?
They can change the bids to make sure you’re still hitting your ROAS. But the place that we as the humans play is the strategy, right? So Bopiz, buy online, pick up in store. Was huge in 2020. That is not a decision that the machines make. That’s the decision that we make because we see things changing.
So this is amazing and great conversation. Now in the debate of humans versus machines, there’s, there’s one point that I do want to make. So I think 2020 taught us the lesson that the machine is really good at responding. to what’s happening in the world and so when conversion rates for certain industries started to change, the machine is very on top of that.
Frederick Vallaeys: and thank you everyone for watching. I mean, I was telling the panelists how many people usually watch live, but we’re like at three times that number.
But yeah, no, they’ll, they’ll continue to be very viable. I mean, I think, you know, I think. Yeah, that’s what I think. But again, some more, more comments there. Just a really, really great. Yeah, I didn’t think this would be a complete rant, but,
So I, I think, you know, the pace that they innovate is, is pretty good. They, they do have to spend a fair amount of their time and energy sort of matching or trying to mark Google changes because they, you know, it’s a market reality that people spend more on Google and oftentimes will begin and on Google and port stuff over.
Matt Van Wagner: Well, Microsoft is continues to add tools and they’re adding you know, more. machine learning things and they’re adding more recommendations and whatever.
Ginny Marvin: no, it’s, it’s a, it’s a compliment as how I would think about that. Yeah.
Brad Geddes: But you won’t spend a huge amount. You, it’s not an alternative to your Google spend. You just can’t. No,
Ginny Marvin: There’s an indication of what’s going to happen. Somebody was asking about being Microsoft being a viable alternative. I, I, I. Would say yes. And if, you know, it just is a little bit of extra resources that goes into managing that you’re going to reach some other different people and you’re going to get some other data.
I don’t think we’re gonna see good results from them. And that’s the part that’s sort of scary, right? Is they try to automate keywords. We’ve proved they’ve proven they can’t do this over and over again, no matter which way you want to walk down semantics. Well, and you can, you can kind of see this just enter your landing page in the keyword tool, right?
I mean, I didn’t think their search engines worse than it was five years ago, right? So, like, I, I, so the more they try to do semantic stuff until they rejigger their system, they get some new people in who understand semantics better than the current ones, and they rely not purely on data from that standpoint, right?
Brad Geddes: Yeah. And the problem is Google is great at learning when it’s pure math. When it comes to semantics, I feel Google ’s incredibly weak compared to some other platforms when it comes to any type of semantic learning.
So I, I don’t think that the intense signal is going away. I think the control is eroding.
On top of it. But and not to say that no broad match works, but or no Brad much doesn’t work. Anyway, some broad match can work. But I mean, I think that’s the thing is the losing control over what keywords we want to show for is as what we’ve seen the matching how it’s been working. Is, you know, concerning in a lot of ways.
What I see is Google not getting rid of that as the intent signal, you’re getting rid of the advertisers being able to say, these are the, you know, Keyword signals that I want to be targeting and said, it’s, they’re saying we know, we know what keywords these companies should be targeting. And what the scary thing for any of us that have actually seen what happens when you just turn on some broad match is it doesn’t work so great in a lot of, unless you’re really, really.
Ginny Marvin: Well, yeah, I mean, I, I guess to Brad’s point about the keywords being just the, you know, pinnacle of intent signal.
But what we’re saying is, is the match type going to go away? Because there is a push from Google . Towards putting people back on broad match, which is essentially what every keyword type is at this point, just grades.
Frederick Vallaeys: I agree with that. I mean, they’re not going to get rid of keywords because it’s a target targeting mechanism. That’s huge, right?
Keywords they can be funny about and show more ads, get more revenue. Every negative decreases potential auction pressure. So I don’t think Google ’s ever going to make fun of negative.
Brad Geddes: I don’t think that’ll happen. Cause that reduces Google ’s revenues, right?
What I’m really looking at, but I think that’s the signal we’re looking for. So if you, if anyone sees it, I’ll be the first one to report, get the scoop, you know, like negatives have gone fuzzy that maybe that’s the, that’s the tag we use negatives gone funny, fuzzy.
So I’m thinking of creating a couple of exact match ad groups with 15, 000 negatives. Just every negative so that there’s nothing like the of a, you know what is, you know, and, and just to see whether or not I get, you know, a hundred percent of, of what I’m looking at there, you know, do I get that? Do I get it?
But once you start to see Negative keywords become fuzzy. Then you know we’ve lost complete control. And so I’m actually, I was just thinking about this the other day. I’m actually going to do an experiment on exact match because there are some ad groups. I only want if and only if this character’s in this row, in this order and these two words.
That’s why we have negatives. And so I think the signal here’s the signal if you see this then essentially keywords have become completely devalued. If you see degradation we’re not gonna see negative keywords go away. But right now, negative keywords are still very strict in the way that they they negate or they you know, they prevent a match from happening.
Matt Van Wagner: don’t want you to see we don’t want you to see The decision we’re making because we may have made these decisions four years ago That that’s not a good keyword.
Frederick Vallaeys: Yeah, we
The loss of cookie data and Google ’s inability to feed its own AI stuff I think what they’re doing is they are learning on all of our dimes And they’re saying we don’t care what you think, you know human we are going to let our machines run wild and we’ll decide what’s working or not
And this is, this is a little bit rant and maybe it’s conspiracy theory too. But I think if you look at this problem in a certain way we really, they may never take away keywords, but they may stop paying attention to what we’re actually telling them. Okay. And and, and, and I think if you look at the loss of, of, of search query data and you look at.
Matt Van Wagner: Well, you know, here’s what I think.
It’s better than any programmatic, any audience, any, anything else. So as soon as Google stopped saying, I can show an ad on, I want X, they’re going to see a massive decline in revenues. People go, well, then I might as well just do programmatic. If I can’t, you see a keyword stuff.
Brad Geddes: Now, you know what? I will argue that as soon as Google removes keywords, their revenues decline significantly. There is, of every advertising method out there, the intent of a search of something I want this is the strongest signal in anything.
Frederick Vallaeys: Yeah, let’s make some predictions here for 2021. So when are we going to lose keywords? Is 2021 the year?
It’s, it’s kind of funny.
So now I get some stats for it. And here’s all the common misspelling. So now I get some stats for it since you can’t get it and it’s not accurate. Because we know this is the keywords impression, not the query impression. So it’s still not accurate, but it’s better, right? So we seem to go the opposite direction as Google does these things.
Brad Geddes: Yeah. Some of they’re using a negatives, but some are just saying, here’s everything that Used to be a search term.
Frederick Vallaeys: with negatives along the way.
So we, some people have just over gone insane on here’s every exact match of variation and misspelling and whatnot, and my entire account. Now I can start to figure something out. It’s supportive
Brad Geddes: This is where we found a lot of people are using more keywords than they have in the past because they want to give Google Every possible variation to get some data. And it’s crazy. It’s like, Google ’s like, Hey, you don’t need to give us all this stuff, but if you don’t give it all to us, then you can’t get anything.
So then what do we do? Shut up human.
We’re just trying to find a bunch of negative keywords, right? And why are we looking for negative keywords? Well, we’re looking for them because you don’t make exact match exact anymore, right? Now it’s close variance. So. You’ve kind of forced us to be more aggressive with negatives, but you don’t have to give us the data to make smart decisions.
Influence, influence change or not, but we get it right. I mean, privacy is important and you have to take away search terms data for that reason. But what is it that you can give us an exchange? That’s maybe obfuscated in a way that those privacy concerns are not there anymore, but give us bigrams, engrams, trigrams, like tell us the word strings that commonly appear in these unique queries that Are sensitive and let us make decisions based on that because ultimately that’s what we’re trying to do.
So we’ve lost what, 70 percent of data, but there’s still 30 percent of data we can use in the same way as we always have to make some smart decisions. I think what we actually as an industry or as practitioners need to do is go to Google and say, listen, we get it on the privacy angle. And that’s regardless of whether this was a privacy, see.
Frederick Vallaeys: So there’s all sorts of data loss and it goes from what the engine is limiting to the fact that cookie policies and privacy policies are just making it harder to connect the dots across systems. And I think that, so the keyword loss or the search terms loss. We’ve all talked about it quite a bit, right?
Do you think that’s a good? Topic to, to transition to, or let’s talk about that. I thought you wanted to talk about keywords, but we’re going to cookie about. Okay, well, I think afterwards, I don’t know either way. I mean, you have some opinions on this, too. So, yeah, I mean, so I think the keyword last thing that kind of speaks to, I mean, maybe what Brad was talking a bit about.
Matt Van Wagner: Well there have been a couple of questions that came you know, the question about keywords you know, what, what, what, what’s the status of keywords? Is interesting. Without sort of going and you know, getting whiplash by jumping off this whole data. And tracking thing. I think maybe we can get to that, but I think maybe we should spend a minute or two to think about some of the automation and some of the cookie loss and maybe talk about how that changes our ability to to measure and and how are we sort of reacting to that?
Frederick Vallaeys: time. Hey, so man, which questions and what conversation that are you watching that we should talk about?
Go. There’s some really good back and forth there. If you guys are paying attention Fred, some of these folks need to be speaking on a town hall to these are good questions and good feedback. Let’s bring them in next
Matt Van Wagner: by the way, I think, I think we should just shut up and let the conversation that’s in the comments.
So we’re going to see an increased CPC because businesses come online, but increased opportunity, right? As long as you don’t do a year over year comparison. And it’s weird to say more expensive CPCs, but better opportunity than normal. In the same sentence, right? Normally, that’s not something you can say
Brad Geddes: Yeah, I agree. I was doing some modeling earlier this week with someone like, all right, we’re going to have a better than usual opportunity this year in, in with Facebook’s tracking changes, the JPM and stuff for are spending a three to 5 percent decline in revenues to do the tracking changes.
But I think we, we, it’s, it’s getting the measurement stuff right is, is important, but then also making sure we don’t make the wrong decisions based on year over year sort of mentality. Let’s temper that a little bit.
We’re actually referring back to 2019 data because 2020 was kind of foobar. So the, I think you have to be careful, especially this year, measuring your year over year metrics beginning March 15th. I think the data after that becomes abnormal. And I think we, we may be abnormal through mid year of this year, so it has value.
Matt Van Wagner: Yeah, we’re actually starting to model based on, we’re, we’re, we’re for seasonality and for sort of predictive, you know, coming into market, coming into season type of stuff.
Ginny Marvin: That’s a good point. The year over year is a really good point and it’s going to be year over year and the next year will be year over year. I mean, we’ve got a couple of years of this. That’s unknown.
Brad Geddes: right.
Matt Van Wagner: That’s
So I’d say even from a. Not a predictive modeling, right, but a basic standpoint of modeling. You need to get it right because the privacy changes are going to mess up the data inside some of your platforms that just to do some basic analysis, you actually need to have it yourself now and can’t rely on the platforms to give it to you because they’re not going to have all the data they had previously.
Brad Geddes: I would argue this is actually more important than in the past. Because with all the privacy things happening, you doing your own data modeling, your own attribution modeling is going to be essential, right? With iOS 14 doing a Facebook seven day cookie and then Facebook doing modeled conversions, not actual conversions, your year over year data inside of Facebook is done, but you can still do attribution modeling in your analytics if you tag it and collect it correctly.
So just forget about perfect and do the best you can.
Ginny Marvin: no perfect. There is no perfect. So if that’s your bar, that’s gonna be a lot of sad nights for you.
And then I can get more sophisticated and I, I can layer another signal on top of it. So like you said, I mean start with the basics and then don’t stress out over like we looking for perfect signal. And there is
The person who didn’t look at that pricing page. And that’s pretty simple. I mean, that could be me saying I’m going to value it 20 percent higher. And where did the 20 percent come from? Well, I completely made it up, right. But it’s better than not doing anything. And now I can start seeing over time, like, is that actually moving things in the right direction?
So if someone’s looking at the pricing page, I think that’s probably a better customer Like i’m not quantifying how much better that customer is I’m just saying like that should be a small boost In the conversion value that I should pass back to Google and the only reason i’m passing back higher conversion value is because I want to tell Google ’s machines that This made me a little bit happier than the other thing.
And now, in many cases, your business is too small, right? So even if you run the predictive analytics or you run the regression model or correlation like you just, you can’t quite figure it out. Right. But what if you simplify the question? What if you just say you know, we don’t need to get everything exactly perfect, but like my example from Optmyzr , right?
How is this relevant to what we’re all talking about? Well, the relevance is that everybody fundamentally understands that there’s probably something going on That’s a better may impact about your business. And so everyone’s like, Yeah, I want to install that script. I want to do it. But then they get stuck because they can’t figure out how does the better impact my business.
Frederick Vallaeys: and then okay so like in for a long time we’ve been talking about the bidding by vendor script okay.
Small like start incrementally. What’s your capacity or should you think about outsourcing and hiring a analytics consultant or somebody who can just focus on getting your tracking set up Like just do that
Ginny Marvin: yeah, so I, that’s a really great point. And I think, you know, like some of these things are out of reach and some of them sound really basic to a lot of the people on this call, but I would, I would say start really early.
Who’s watching. Like we all get it. You’re preaching to the choir. We’re preaching to the choir, but like, how do you get that customer to give you that clean conversion data? Because some of them still can’t set up conversion tracking properly. I mean, I,
Frederick Vallaeys: I love everything you said, but like, that’s such a high bar for some advertisers. And I think everyone’s sitting on this call, right?
I’d start with that. And like fundamentals of data cleanliness, understanding what signals are actually leading indicators for your business, I think, is kind of like Q1 of 2021. Just focus on the data fundamentals.
Going to happen. And you know, that’s how audience targeting is getting delivered. That’s how ads are getting served. All of this is getting based on predictive signals. So you know, I, I think in many ways, 2021, if you don’t have a, if you don’t have your analytics, squared away and are starting to test GA4 and getting those profiles tested.
It is not clean. It’s and so I think a lot of this is if you’re going to talk about. Predictive analytics and predictive marketing, which you know, I wrote a piece and did a talk at SMX last summer about being in the predictive marketing era that everything is now being built on historical data, being able to predict what behaviors are occurring.
But I do, I do think, I mean, I, two things that have come up, I think one, we’re talking about data and I think. More and more, the data that you have is going to have to be good data. And, and it sounds so basic, but I can’t tell you the accounts and analytics I’ve been into that it is, it is not good data.
Ginny Marvin: more in there.
Frederick Vallaeys: and find
Ginny Marvin: and then you can build audiences off of those
But at least you can recognize that that was a more qualified lead in some way. And we need to be feeding that back. And we don’t need to wait until the very end. And you can do conversion updates every couple of hours if you want. Well,
And so now if you recognize that someone who’s in the free trial, who’s got a conversion in Google , if you can say, well, that person came back to the pricing page three times, we should score that higher. We should tell Google that’s the perfect candidate. Like they’re actually interested. Now, whether we close them or not at the end, like that’s still going to be the job of the sales team and the job of how good is the software.
And if that person who’s in the free trial has come and looked at our pricing page a couple of times, they’re probably more likely to be serious about buying the software, right? As opposed to someone who’s never looked at the pricing page. And so what are these signals that you can pick up and that you as a human can have a bit of an instinct about, well, what is it that Probably moves the needle on my business.
Okay, so that’s a conversion for us. But what we really want is not a free trial It’s the person signing up and becoming a monthly subscriber now We tend to wait between that the two weeks of them doing the trial. We don’t really look at anything But what we should be doing And what would help Google is if we got smarter, we said, well, let’s look at Google Analytics.
Frederick Vallaeys: And so the way we’re thinking about it is, you know, grab the GCL ID when that Conversion or whatever the event that you’re tracking as a conversion is and don’t wait. So say that you’re a lead gen, right? Don’t wait until the final Conversion has happened until you close the deal with that customer But start looking at the signals along the way as you can identify them in something like Google analytics to say So let me put it into perspective of The software vendor at bread is i’m sure you see the same thing, but we do a free trial, right?
The best we can do, I think, is to, to make our site and our. And our analytics sort of work the best with the tools that are being developed by Google or, you know, other so I agree with that, Matt. I mean, so looking at Google Analytics, that’s sort of the next step of informing the machine of what’s happening, I think, is really important.
Represent good proxies to the next sort of actions and give those signals a little bit more attention feed them and develop audiences out of them and try to understand those a little bit. So we’re for most of our clients, we don’t have the ability to, to take and build AI. I mean, we’re not going to do that.
And no matter what you try to do to try to connect the data with phone tracking, everything else, it’s still, you still have this problem of, of connecting, you know, that, that last, that last bit. So our, our, our goal really this year is to take a look at how we can understand what behaviors on the site.
a full online you know sales chain, you know so many, so many companies in lead gen and everything else, lead scoring and, and orders that, that happened by people calling on the phone and everything else. You don’t get that last, you know, mile to, to be able to connect all the data together.
You know, scroll depth and time on page and to do that multimodally. So as opposed to just taking a mean, an arithmetic mean, how do we take and separate these data to be able to create audiences that, you know you know, how, how do we build up signal within the site that we’ve got and, and, and triggering events that we can feed back to our bidding algorithms that especially for folks that don’t have.
And it’s actually going back to really seminal work in the whole web, UX and web design area. If you look at Google analytics for, and it’s moving away from discrete. Real things to events and connections to events that represents sort of proxies scroll time time on page. What I’d love to start to see is, can we do some things like, can we get combinatorial data that would allow us to say combined?
Matt Van Wagner: all good points. Yeah. And to you know, a couple of folks have mentioned some things Brett, but off ski and Dave Wilson’s sort of I think we’re making a similar point, which is all right, instead of fighting let’s focus on giving them the right goals. So one of our focuses and Dave says, Dave Wilson says to feed them, and I think that’s I think that’s where we’re going to be focusing a lot of our time.
So
One, businesses don’t always want to share their inputs. With Google , right? So when you run up against that how do we anonymize that those inputs? And, you know, I think I think there’s certainly conversations around that. But there’s the sensitivities to, you know, what happens. Yeah. What data businesses are going to want to share with Google to make those, those inputs actually work within those algorithms.
I. In house and You know, that may be getting easier for sure, but it’s, there’s, you’re still limited to the number of businesses that that’s going to be accessible to at this point. And so do you do like, is the, is the real voice here of advertisers to keep pushing for give us more inputs. And again, I think without what we’re, there’s two issues there.
I mean, that is not, it’s easier said than done. It’s kind of it’s, you know, it’s easy to say, these are our top signals. Yeah. But that takes real analysis and investment and in internal or agency, whatever it is. But there that takes a lot of digging and and then be able to present that in a way that is can actually be used, whether it’s creating your own A.
And, and, you know, Because of Google signals out to Brad and Fred, your points about having the, our own business signals mixed into those, that data I think is really critical. And I, and I agree, I think that is where Where the real leverage can come, and particularly the competitive leverage over, you know, your competitive set, those people that are and businesses are able to one identify their own business signals that are going to make an impact.
Ginny Marvin: Yeah, I mean, I think you’re right. I mean, I, I, Matt’s the stop fighting the machines mentalities. That’s more where, like, I really feel like that is the way that we’re going to have to evolve. And I’ve kind of been thinking this way over the last probably two years of just, this is what we’re, this is where we’re going to have to go.
And I think that’s sort of the question that we have to start asking more.
No offense to people in New York. Right? Actually, we have Wildfire, so maybe you’re not using it in California quite as much. But you know that that user is going to be buying those accessories because they’re going to be camping more often, right? And so it’s, how do we communicate that to Google at the periphery of Google Ads so that inside the circle that they control, they can actually do their magic?
And, and, and so if we know that a conversion from California just means more to us than a conversion from New York, even though maybe they bought the exact same camping tent, right? But maybe we know that someone in California is going to use that camping tent more times during the year because the climate is better.
Frederick Vallaeys: I my point is that kind of like in the circle that we’ve been controlling I think we have to move to the periphery of that circle And now we got to start thinking more about how do we tell Google about what a conversion really was, right?
Brad Geddes: but so
Frederick Vallaeys: Yeah,
Brad Geddes: doing? They’re using predictive analytics, right? Just to be clear, right? It’s predicting analytics in house bidding not historical
And Google has these ways to predict what’s happening What are they
Frederick Vallaeys: So I think sort of what Matt was also alluding to was that, like, we have to stop fighting the machine. And basically, when I think about AdWords and being like the circle, we used to manage all the little components within the circle. So we used to manage targeting and bidding and the messaging. And everything was done, like, Kind of chasing history like historically we knew that okay these things happen and that converted well and this didn’t so we’d set these Bit modifiers or adjust things but but basically we’re always chasing history, right?
Easy is a tough word, but more viable when you find it hard, he can help you. No, I’m not a programmer, right? Is this something that you got to be careful of? Because I know people are starting to catch up with what Google ’s been doing, right? But I think, and maybe the question was directed at Matt, but let me jump in, right?
And and it’s better than Google because they know what data makes the biggest changes. Google doesn’t. So until Google . Let’s you say, Hey, on the machine learning levers for bidding, I want to push a versus B versus C more. Should we really be saying, let’s turn it all over to the Google automation when we’re seeing because of machine learning so easy to do in house now?
But, but just to give scope here, that have a couple good developers in house. Who have started running machine learning or predictive analytics over their own data, and they find they get better results using all the bid modifiers, which is still automated bidding, right? They’re just automating to themselves.
Brad Geddes: So, Matt, can I make a quick, a counter question, right? So, we’re seeing accounts that may be spending as little as 50, 000 to 100, 000 a month, which I know is a fairly healthy budget still, right?
So it’s, I, I think you’re right. I think you have to look more holistically at. As opposed to just doing this or that, look at it across the entire account.
They’re all, it’s all, especially if you see, by the way, we’re going to talk about Google Analytics 4. If you see, What’s happening with, with Google analytics Google has already planned for the fact that we have very few real signals to deal with cookie loss and everything else. So modeling and estimation and, and history, history in, in particular keyword sets is, is what they own and we don’t.
Ads are sort of that’s sort of a whole island and that, that seems to be settling out. But I do think that it, it gets to a whole larger area of, of how we are managing the entire, our entire accounts. It’s not just RSAs or ETAs or disk automation bidding, or this type of way to develop audiences.
But when you start to pull in bidding and we start to pull in, you know automation to particular events, if you’re an all online player, you’re not, it, it all does get sort of. Really difficult to understand what the impact of one versus the other is. So I, I think your point, Fred, is that this is the year where I think we have to agree we, we, we think that we have to stop fighting the automation and really look to where the automation is going to be in six months and start to redesign our campaigns, the way we do our keyword match keyword sets.
I mean exact matches even. eroded. I don’t know if you noticed this recently, but you know, we’ve, we’ve designed campaigns over the last number of years that are set to exploit and optimize this one thing, you know, get impressions and, and, and you know, keep, keep a traffic into a particular data set.
And so but I do, I do think the point about automation and the fact that you’ve got, you know, automation in audiences, you’ve got automation and bidding, you’ve got automation and in your ads you’ve got really automation happening in your keyword match types, because for the most part, keyword match types are almost completely gone.
Matt Van Wagner: It’s multivariate type of testing. And, and so, you know, one of the complaints, and I think Brad, this has been one of your sort of pings for a long time with with automated testing. You know optimization, you know, to click through rates, whatever that oftentimes you see that decisions are made being made too quickly on winners and losers.
And that’s like giving up huge control. So Matt, what do you think about that? Well, I think I think that’s a really interesting point. And I think Melissa Mackey looks like she’s at she mentioned a comment that in a lot of cases, you just don’t really have enough data for RSAs to really even get their wheels spinning.
So you get frustrated that the result isn’t there. But if you put automated bidding on top of that, Google also knows what you actually care about. So they set the right bid and that’s where it gets tricky for us, I think. And I’d love to hear your thoughts, right? But there’s this whole angle that for automation to work at its best, you almost have to use every form of automation.
But part of it also then depends on you have to be using broad match or something Not exact match and it probably is beneficial as well to use some level of automated bidding Because Google may show these ads that now qualify for new auctions because you’ve allowed them to get onto these new onto these new auctions through broad match keywords But now you might be bidding wrong, right?
Frederick Vallaeys: It was back in March of 2020, I think. And I’ve had several scripts to that whole point, right? So there’s this whole incrementality argument that Google tends to make. So you’re like, you’re saying it just, it enables your ads to qualify for more auctions because they’re able to put together a variation that has the high enough quality score and ad rank so that you can get onto these new keywords or searches, right?
Lifts and total conversions, right? That’s something we, you know, again, if they were new impressions and it lifted conversions, that would be one thing. I haven’t seen it. Yeah, I’m sure there’s exceptions, right? But, you know, overall, so I love the point that you’re bringing up. And we had this script on search engine lands.
And so when you’ve got a 95 percent impression share and you suddenly start showing for new stuff, It may not be things you want, so primarily they’re looking at a conversion per impression type ish, you know, metric right to sort of marry the two, but usually impressions has not been one and we haven’t seen.
Brad Geddes: accounts have already encompassed the queries they would show for. So they’re not seeing new queries or if they are their queries, they don’t want anyway.
Otherwise, I can kind of think about that. Inventory differently. Or no, is it really just like, we’re just going to look at the bottom line and forget about those extra larger
I’m wondering, particularly for Brad, if, if advertisers are also looking at, because one of the things about RSAs is you’re supposed to be opening up yourself for eligibility to more impression inventory. And so is that a factor where people are saying, you know, Or looking at it from a I’m getting more exposure and I’m not going to be thinking about it from a conversion rate standpoint that those are kind of like incremental impressions that I wouldn’t have gotten.
Ginny Marvin: My, my question on it, it comes back to, so what are, what are results? And I know that’s a, we know results are, but you know, if we’re really strictly looking at conversion rate, cost for conversion, then, you know, I can see where ETAs are often going to win or, you know, ROAS, whatever it is, we’re gonna, that’s going to be the winner in many cases.
So I think you know, there’s certainly benefits to RSS and we have some metrics we can show pop up in a moment that they do.
Frederick Vallaeys: I think what I sort of mean is that in RSAs There’s really not enough data in many cases to help you take back the control to get to the results that you want.
A lot of them They don’t care right if you’re a lead gen company You may not you may spend 10 million a month, but in the end you care about the results, right? and so
Brad Geddes: It’s not just about the control. It’s about results, right?
Frederick Vallaeys: Which in a way is not surprising, I think, right? So if you spend a lot of money, you tend to be more of a control freak and actually like do PPC for a living. As opposed to something else you’ve just tacked on. So you probably want that control back
They’ve shrunk their usage and so we again I’ve done a little bit of segmentation. Since we did smx on spend and account size versus declining rsa usage And it’s definitely the smaller accounts have declined their usage much less than the larger ones. So our debt is a little opposite of yours in ways, but we have definitely seen those who spend, you know, half a million to a million plus have decreased usage more than those who spend 10, 20, 000 a month.
Brad Geddes: I’m wondering if you’ve ever done the data of saying, all right, a people who tried them. This is the, this is how many they got to, right? They had X number of ad groups percentage wise with them. And then here, where it is now, is it a shrinking number? Because while Our data shows that more people have have tried them and stopped than yours does We also see is a lot of people who still have them.
So
So, RSAs existed in the accounts. And 75 percent of those who used it kept, kept using it. Right. So, they enabled ads, they kept them live, they didn’t turn them off. Only 25% Have basically never tried them and then a very tiny fraction of all the accounts had tried rss But stopped using them. And so to us that was an indication that automation in the case of text ads Does seem to work people seem to like it at some level Brad, what are your thoughts on that?
It’s just because there was an ad format that was just text as well This one is just text that you define, but it was just newer it had more you know more space basically So that’s called ETA Right. So, but we looked at accounts that had tried RSAs and we saw that 75 percent of all the accounts in Optmyzr had at one point tried RSA.
But I think at the end of the day, most of the time we tend to agree on topics, but let’s share what we found. Right? So the first thing that we looked at from Optmyzr s perspective was how many people have tried. Responsive search ads and obviously the responsive search ads. That’s what I would what I would put in the category of automated advertising Right you as the human you still put in the ad components the headlines the description lines But you let the machine figure out how to put it together in unique ways And on the other side, there’s the eta the expanded text ad for people newer to ppc Why do they call it ETA?
So I figured where we start is RSAs and ETAs. So. Ginny, Brad and myself, we did a session on that at SMX towards the end of the year. So and the whole thing was positioned as a debate, right? So Brad and I were supposed to basically disagree on everything because that’s what makes things entertaining.
Frederick Vallaeys: Yeah. Yeah. Lots of people calling in life today. So this is great. So folks, if you have any questions or any directions you want to take this topic. So obviously we’re talking about automation. Automation is fairly broad as a topic goes, and I think we’re going to talk a bit about ETAs versus RSAs, maybe to get things started here but yeah, let us know where you want to take that conversation, right?
Ginny Marvin: Oh, I love seeing where everyone’s coming from.
Frederick Vallaeys: and hence, that’s why Matt respectfully requests that nobody follow him.
But for the most part we’re working people, we’re helping people with their with their paid search stuff. So,
Matt Van Wagner: Well, we, we we’re an agency. And so we help mismanage paid search for a number of clients. The you know, we, we, we, we have a number of We, we service large, small, we do, we do SEO for people that know less about SEO than we do.
Frederick Vallaeys: Yeah. And so Matt, besides SMX and being a full time comedian and musician what else do you do?
I think, you know, are some of those things. So I’m glad to be here today.
And, and I’m, I’m just really glad you, you, you, Fred, you mentioned that. You know, the three of us have a common thread where, you know, we, we did a lot of the programming for the SMX paid, you know, paid panels over the last few years. And so it was always, it was really always a lot of fun because Brad and Ginny and myself would always be, you know, trying to think, you know, what, what are, what’s on people’s minds, what are the biggest problems and, and We’re talking about today.
Matt Van Wagner: Yeah, yeah. So, Seven Days. I think you’re gonna hear both of those. I think Trump has picked that. That’s the song that Trump has picked when he leaves the White House. I mean, you know, nah, nah, nah, nah. And chief Agency is a new one. At any rate it, it, it, I think we’re all looking forward to a new period of time, 2021 automation, all sorts of good stuff.
Frederick Vallaeys: Very nice.
Right?
Which one do you think that might be? Right? And of course, of course, on the same day, on the same day, we’re gonna honor, and we’ll hear this one played a lot of times.
Matt Van Wagner: hang on, hang on. Let’s see. Let’s see if you can recognize which is which.
Frederick Vallaeys: Well, well,
I think actually it’s two events. One is we’ve got a president leaving and we’ve got a new president coming. I, I, maybe we have two different pieces of music. What do you think? Yeah, let’s do it. I mean, all right.
Matt Van Wagner: which I think there might be.
Frederick Vallaeys: And by the way, everyone listening in, we love it when you put in the comments where you’re calling us from and say hello and then Trying to get everyone primed up to ask questions. So yeah, seven days at Countdown. Hey, I know you’re a musical guy Do you think it might be a? Some music that accompanies that sort of an event,
So in seven days, it’ll be a wonderful day here. Seven days, we’re counting down every hour.
Brad Geddes: I’m outside of D. C.
Frederick Vallaeys: So where are you calling us from today, Brad?
Matt Van Wagner: Yep.
Brad Geddes: Yep.
Frederick Vallaeys: Well, SMX panel about ETAs and RSAs that I figured you know, we got to bring you onto one of these, right?
Brad Geddes: Thanks for having me.
Only reason that I hadn’t had the guts to invite him before. But Brad Geddes, welcome to the show.
All three of you are deeply involved in SMX and organizing those events. So, so thank you for putting on some great educational sessions over the past couple of years, and continuing those, of course, in in the virtual world. So our next speaker that I want to introduce Man, everyone knows him PPC, but yet he’s never been on PPC town hall.
So many of you have seen her lead panels at the SMX conferences, the virtual ones, the live ones. Ginny was also the editor in chief of SMX. So if you’ve ever read that blog, and I’m sure you have, then her name should be very familiar to you. And by the way, that’s actually kind of the common thread between the speakers today.
Frederick Vallaeys: Yeah. And so Ginny, obviously formerly of SMX and third door media.
Ginny Marvin: It’s going great. It’s going great. I’m taking a little bit of a break and doing some work here and there. And we’ll see where things go.
All right. And here are my panelists. So welcome to the show, Ginny Marvin, Brad Geddes, and Matt Van Wagner. How are y’all doing? Ginny, how’s it going?
So today’s PPC town hall, we’re going to talk about automation and what the experts are automating and what they’re still doing manually. We’re going to talk a little bit about brain safety with all the things happening in the world again this week and in 2021. So with that, let’s get rolling with PPC Town Hall, the first one of 2021.
And so we’re really happy to bring it back in 2021 Because we found that PPC professionals just don’t have places to go and meet in person anymore, yet there’s still so much that we can learn from each other. So for the first episode of 2021, I decided we should bring on some some people we’ve never had on the show before, but who you certainly will recognize and also bring back some of our favorite speakers and some people who really We’re deeply involved in the industry and are going to be able to lead conversations with us and have some really interesting insights, not only from the things that they do, but also from the people they’ve talked to in the many interactions that they still been having in this past year.
Frederick Vallaeys: Hello and welcome to the first episode of PPC Town Hall for 2021. So my name is Fred Vallaeys. I am the co founder of Optmyzr , and I’m going to be your host today for Town Hall. So thanks for those of you who are joining us, especially the ones joining us live to ask questions. So we started Town Hall in 2020, about in March, and we ended up doing more than 20 episodes of it.