Google Ads Value-Based Bidding: How It Works, Best Practices & Common Pitfalls

Episode Description

Stop using #TargetCPA bidding.

Instead, move to #ValueBasedBidding to maximize your profitability.

To help you learn more about Value-Based Bidding, how it works & more, we brought in experts from Google — Alex and Rebecca.

In this episode of PPC Town Hall, you will learn:

• Why you should move from Target CPA to Value-Based Bidding
• What is Value-Based Bidding and how it works
• Best practices and common pitfalls while using Value-Based Bidding

Episode Takeaways

Why You Should Move from Target CPA to Value-Based Bidding

• Google’s Stance: Google suggested that using a Target CPA may not be as effective as previously thought, prompting a shift towards more value-driven strategies.
• Effectiveness: The discussion highlighted that while Target CPA has been a go-to for automated bidding, it might not align well with actual business goals which can be more effectively met by value-based bidding strategies like Target ROAS or maximizing conversion value.

What is Value-Based Bidding and How It Works

• Definition and Application: Value-Based Bidding focuses on maximizing the actual value derived from conversions rather than just increasing the number of conversions. This approach aligns bidding strategies more closely with business outcomes.
• Google’s Explanation: The Googlers explained that Value-Based Bidding allows for differentiation among leads or sales based on their actual value to the business, which can lead to more efficient use of advertising budgets and better alignment with business goals.

Best Practices and Common Pitfalls While Using Value-Based Bidding

• Best Practices:
  • Focus on conversion value and align bidding strategies with business objectives.
  • Ensure sufficient volume of conversions to validate the effectiveness of the bidding strategy.
  • Consider conversion delays and learning periods to accurately assess strategy performance.
• Common Pitfalls:
  • Including the learning period and conversion delay in performance evaluation can skew results.
  • Making significant changes during the learning period of the bidding strategy can disrupt its effectiveness.
  • Overvaluing certain conversions without clear incremental value can lead to inefficiencies in bidding.

Episode Transcript

Frederick Vallaeys: Hello and welcome to another episode of PPC Town Hall. My name is Fred Vallaeys. I’m your host today. I’m also the co-founder and CEO at Optmyzr. So today we wanted to discuss something that’s been a little bit of a buzz in the PPC industry. Google’s been doing a couple of webinars to tell us what to expect in 2022.

And on one of those webinars, some statements were made that talked about TCPA maybe not being as cool or as useful. As a lot of us have thought, right? A lot of us these days in PPC. We think about smart bidding. We think about automation. And when we think about automated bidding for many of us, that really means switching over to a maximized conversion strategy, sometimes with a TCPA, a target CPA goal.

Now, Google came out and basically said that using a target CPA May not be as effective as many of us thought it was and actually wants us to maybe shift away from using target cpa so when we hear that that’s a little bit shocking, of course because you know We thought this was the state of the art So let’s go a little bit deeper on that topic today and let’s find out what google really meant What google believes we should do?

And who better to help us really? dig into this topic than Google itself. So today we’re very fortunate to have a couple of Googlers on PPC Town Hall. We’ll be having a discussion with them and they’ll explain what they meant when they said TCPA may not be as effective as we thought it was. So let’s get rolling with another episode of PPC Town Hall.

All right. It says right there, target CPAs. That’s quite the statement. Right? So I, Hey, welcome to the Googlers on the call today. Alex Ioch and Rebecca Chu. Thanks for joining us. Absolutely.

Rebecca Chiu: for having us here.

Frederick Vallaeys: Yeah, and Alex, you’re returning. You’ve been with us before on the PPC Town Hall, so thanks for agreeing to come back.

And we had some questions that were submitted by some Optmyzr users, so you’re going to help us answer those today. But Alex, remind everyone who hasn’t met you before what you do at Google.

Alex Ioch: Absolutely. It’s a pleasure to be here. Thank you for having me back. I’m Alex Soich, original product lead automation in America, so specialized specifically on any smart bidding variations and helping our top agencies and customers succeed.

So pleasure to be here. Great.

Frederick Vallaeys: And Rebecca, you and I have been working together for a couple of months now, but I think this is the first time that you appear on the PPC town hall. So welcome to the first timer and tell us a little bit, well, I know what you do, but you never know what else you do.

Rebecca Chiu: Yeah, absolutely.

So I’m a platform partner lead for Google and I’ve been working with Fred’s pretty extensively in terms of products solutions between Google and Optmyzr’s. So really excited to be here to talk about value based bidding and what the next frontier is.

Frederick Vallaeys: Great. And the listeners always love to know where people are coming in from. So Rebecca, where are you located today?

Rebecca Chiu: i’m located in Redwood city in Georgia office and in Woodside, California, living wise,

Frederick Vallaeys: right? So, right here on the peninsula where I am too. So I’m calling it from Los Altos. Alex. What about you?

Alex Ioch: Same here same office right at the city.

Frederick Vallaeys: All right Well, all three of us are hoping for a little bit of rain at some point this year I don’t think we saw any rain in january february.

I get to travel to SMX Paris soon and i’m gonna see my family . They requested I bring them some sunshine, and I think I’m asking for some rain in exchange for that. All right. So, well, let’s jump into it, right? So you know, the big question here, TCPA is dead. Like, what did that mean?

Alex Ioch: So to clarify, really not that though, right?

Frederick Vallaeys: Okay, that’s just me hyping it, right? Okay, but so yeah, if people aren’t supposed to use TC. Just

Alex Ioch: a tiny bit. So it’s as effective as it used to be. So by what we mean by actually moving to ValidBiz being strategist, is it going to be a little bit more? Effective since it’s going to be aligning with your business goals.

So I think that’s going to be topic of our conversation today is looking at more of like your business objectives and how value is being strategist, namely target return on ad spend and maximize current value, how they are better in terms of actually hitting your, Goals from the business side. So that’s why we’re encouraging.

Not, not that saying the TCPA is that is more encouraging people to transition over from target speed to target revenue and expense strategies. Right.

Frederick Vallaeys: And so TCPA is a strategy, maximize conversions. We’re still going to have access to that. But I guess the message is. It may not be as effective as you, as it could be for you.

So there are other options. So let’s illustrate that maybe with an example. We got this slide here and it’s a, it’s a Google slide. I’ve actually modified it a little bit. But Alex, do you want to walk over? Actually, Rebecca, why don’t you walk us through this one?

Rebecca Chiu: Yeah, absolutely. So this just shows not every single lead and customers is valued exactly in the same way, right?

So you may have a lead that’s value at a hundred and someone who comes to you for, for estimates and took some kind of conversion action, maybe value at 300. And the customer who actually transacted with you is a 500. So. In terms, their bidding in terms of value is so different. So you may want to set the way to, to bid for these customers very differently to, or, but these, these leads.

So one could be, you know, set the 5 versus, versus 15 that really aligns with your business goals because not every single lead and customers are valued exactly in the same

Frederick Vallaeys: way. And I guess the crux of the problem is that. Okay, so you’re a lead gen business and your focus is heavily on target cost per acquisition getting conversions And so what you’re communicating to google what you’re tracking as a quote unquote conversion is that first stage to lead?

But the point is that different leads behave in different ways and some of them Actually, you know, if you sell roofing, right, some people will ask for the free estimate, and then some portion of those people who got the estimate will also become a customer. And it’s sort of those things that happen after the lead stage.

So, so in classic target CPA bidding, all three of these leads that end up doing different things are valued the same because you don’t look beyond that initial contact. And so that’s why you’re saying if you do look beyond that, now you can actually bid differently for them and bid more for the high quality lead as opposed to the, you know, the person who’s just kicking the tires a bit.

Rebecca Chiu: Exactly.

Frederick Vallaeys: All right. Let me turn off my sounds here. People keep interrupting. All right. So with that, how do people sort of shift into target rows? And by the way, have you guys seen any results? Like what happens? For the typical advertiser who switches from TCPA to target ROAS.

Alex Ioch: Yeah, I think we actually have a slide for this one.

So thank you for alluding to this one. So for in implementing them, you’re like from switching from conversion values being strategy, so your targets be a maximized conversions, right? And by virtue of prioritizing different leads or different stages of your funnel differently, we see average uplift of 14 percent of conversion value.

We have the same investment level, right? So you already had a pretty tremendous, a good outplay from coming from manual bidding to smart ring strategies, target speed. In addition to that, coming from target speed to target ROAS and maximize curve value for that matter. You can see extra 14 percent on average, of course uplifting conversion value.

So pretty good results. And that’s how it makes it effective by actually either prioritizing how much each specific person in auction for retail businesses, both to you or actually different stages of the funnel prioritizing leads, maybe over calls, depending on the, on your strategy and actually telling the Google, what’s more important to you from conversion value perspective.

Frederick Vallaeys: Right. And one thing that’s fascinating to me in this is that TCPA has been so strongly aligned with legion advertisers and TROAS has been so strongly aligned with e commerce. Right. But I guess that’s the wrong way to think about it. Right. So Rebecca, you were alluding to this, how it’s really about getting as close to your business outcome as possible.

Rebecca Chiu: Yeah, yeah, absolutely. So in terms of lead gen, we see different people using different kind of methodology in terms of value of their leads. So you could go from a CPA to potentially cost of sales, ROAS, and then integrating your margin data to have profit and then ultimately LTV. So. So aligning as closely to your business outcome, it is best.

A lot of times, you know, given shopping campaigns, especially smart shopping campaigns, you have that value and transaction in a sales number already in terms of optimizing for target Ross. However, we, we see in terms of shopping campaign e commerce customers. A lot of times they don’t use target ROAS for search campaigns either.

So we highly encourage people to actually look across all those different products in terms of Google and just using target ROAS or max conversion value. And on top of that, in terms of lead gen, there’s a proprietary, you know, CRM solutions, off the shelf CRM solutions. other ways in terms of bringing some of that data into the funnel to look at exactly what the value of the leads are.

So we, we definitely see a higher adoption rate from e commerce customers, but we see opportunity across the board with different kinds of marketing objectives.

Alex Ioch: Yeah, absolutely. And then for Legion specifically, right. It’s not that it happened only in a one touch that you have multiple touches all the way down to close deal, or actually if you track lifetime value of the customers, you can pass that information back as Rebecca mentioned.

Even, even as simply as voting for the spreadsheets, right? So you don’t have to have any sophisticated CRM system to do that. And by virtue of passing that information back and you know like actually Matching against the data, what happened within the front end, you’re going to be able to allow smart being strategies to target rise or maximize current value to go after more valuable to you.

customers. So importing that data will allow you to actually tell the system what’s more important. And next time the similar person with similar attributes happened to be in an auction, then SmartMix Trans will know and predict a lifetime value like or your conversion value. the thing, you know, auction itself, person by person basis, right?

Not only conversion rate as they do with target speed, but also conversion value prediction is going to happen. And by predicting that you’re going to be, have a higher chances of being more visible for the people who are more likely to convert for the bottom of the funnel. So that’s a goal, right? Tell the system where people are more valuable to you.

And as the similar people come in auction based on the data you provided, then it will be higher on them. Make sure you win over competition. Right. And then you know, assuming they like you and we repeat the process, you get more and more customers.

Frederick Vallaeys: Exactly. And I started to put that into a different perspective, maybe, you know, say you’re a bank and you sell mortgages, right?

And you get all these leads for mortgages and you you go from manual bidding, CPC bidding, like we saw, which is the simplest, and you switch over to target cost per acquisition. Great. All of a sudden, the automation gets you more leads, more. People filling out the form at a lower cost per acquisition.

And you’re happy until the next month you look at your mortgages and you see that actually you sold fewer mortgages. Why is that, right? It’s the machine is getting you more leads. Why aren’t they closing? Well, because you haven’t communicated. What you actually want. Like he, I mean, listen, I run Optmyzr.

I love leads for my business, but ultimately what I care about is people who sign up for the software and that’s very different from the lead. Like the lead is one stage to get there. The free trials one stage to get there, but that’s not the end game. Right. And so with the bank, what that bank should be doing is communicating.

When somebody actually is qualified for a mortgage and when somebody actually closes for the mortgage and can communicate that back to Google. And now the automated bidding systems from Google can actually go and find the type of traffic that matters to you. And so the beauty is you don’t even have to worry about like, what’s the actual CPC bid that should be set in that case, like that’s what the automation does.

But you just need to tell the automation what you really care about, because if you don’t tell it, it’s just going to get you the wrong thing. So I hopefully everyone listening sort of like gets this and is Excited about doing this. I mean, it makes sense makes a lot of sense to me. So how do we go to get there?

Right? Like, what are some of the first steps in terms of getting started? So Alex, I think we have a framework for that. So let’s that slide here,

Alex Ioch: let’s take a look at more visually. But the first step is really to ensure you actually passing down information back to Google, right? Sharing the better data.

And again, don’t have to be super sophisticated, but so those are some of the options you have here listed under share better data. So, you know, Google’s site tag. So Google g tag from Google ads account natively tracks really better. If you have an option from the many kind of conversion tracking, privacy is super important for Google.

So make sure you don’t lose any conversions because all the. In a disabled cookies or anything. So enhanced conversion will help you to add on top of that. If you have an ability, especially for lead gen customers, like offline conversion imports, right? The your what happens offline after a year of the people submitted their leads, how do you track in your serum system or in the back end passing that information back?

It’s really important. So some of those kind of highlights for sharing the better data. Yeah. And the second stage will be really once you do have it ensure that you assign the correct values. What is expected value from each either stage of the funnel for lead gen or from different customers for the retail cut, right?

So I, so that’s so that the system knows. How you can actually prioritize within different stages for different, different interactions is exactly once you have that information, you share the data once you value them differently to prioritization perspective, right? Then you can actually set the bidding strategies to optimize what you’re looking for.

To either return a net spend or actually. Just maximizing in that conversion value.

Frederick Vallaeys: Hey, and maybe we jumped the gun here a little bit. Right. So these are all great things to do. But like really first step, or one of the foundational steps here is switching from maximize conversions to maximize conversion value with a TROAS.

Or if you’re doing this for the first time, choosing that bit strategy. But we did have an audience question, and people, and I’ve spoken to a lot of people who have this question too, but If you don’t have a lot of conversion data, yeah, what do you start off with before you go into maximize conversions with a tROADS?

Alex Ioch: Yeah, it’s a great question. Most of the strategies actually don’t have a number of requirements for number conversions. And the reason is behind that is we we know the intent behind the pool as the people are actually searching for products or such as yours, right? So, We know the intent on the back end side.

It’s just a matter of how they look and how they work and perceive your brand specifically that we need to learn. So target span, maximize conversions and maximize current value. So the max current value doesn’t have any required number of conversions. You can jumpstarted from the zero for target, return, and ad spend.

There’s only requirements is still there. If you’re coming transitioning from. Manual bidding to target return on that spend goal, then you have to have 15 conversions on the campaign level that is being worked on to elevate on account level right now. So it’s in the works so it’s good news there.

And then we have 00, the disability beta you can enroll into, so you don’t even have to do that.

Frederick Vallaeys: What’s the sorry, say that beta again. What was that?

Alex Ioch: So is zero conversion eligibility beta for target and an S band. If you work in some of your Google representatives, ask for that one has been pretty popular.

But again, it’s only if you’re coming from manual to target directly. If you’re coming from X conversions or kind of like targets be a two target for us. Because because we already learned information about user intent via those strategies, then they’re like passing that information basically to target rest.

So you don’t have to have any conversions coming from target speed to target trust specifically. So there’s a zero number of conversions requirements there too. So pre most of the cases you know, straightforward implementation. So you don’t have to wait. If you want to jumpstart. With say if you just a cold start campaign, if you want to jumpstart with maybe target speed and then transition to target for us, you can always do that.

Or you can go with max conversion value at the beginning, if you like. Yeah, so maximize conversion value without tROAS or maximize conversions. Now some people I talk to, they just want to go. Enhance TPC or maximize clicks. I know that’s not the recommendation, but like, let’s specifically call that out.

Frederick Vallaeys: Is that just not on the slide or is that actually something you don’t recommend?

Alex Ioch: Yeah. So the issue is that you, you always can try that then the issue specifically, I think more of if enhanced, enhanced cost per acquisition is technically manual bidding strategy, right? So you’re optimizing only 50 percent of your traffic and then still respect and all bid adjustments.

So you’re not necessarily learning all the intent behind the users. So it’s not recommended because it’s going to be still learning much harder learning period transition from CPC to say target for us once you Have those data enrolled. So it’s always easier if you already let the system know and learn is user intent Say if targets be a or max conversions or maximum value for that matter so that information will be shared with other beating strategists, right?

So but if you want to jumpstart with X clicks, you can always do that. And again, but there will be, it’s also not a smart being strategy is not the optimizing for any conversion data. So that’s why any automated being strategies and auction levels which we call smart beings, like there’s targets.

X conversions conversion value, those are going to be probably better option because you’re going to be optimizing for actual conversion information.

Rebecca Chiu: And I just want to add that Google has incredible intent data, right? So we know how much how, how valuable each lead is. So if you’re passing information, we see the most innovative marketers actually using max conversion value and target ROAS.

You could go back five years to use maximize clicks, but it’s not going to be aligned with your business goals that you want to ultimately get to. So we definitely

Alex Ioch: traffic for sure. No, no worries. I don’t have max clicks, but these are going to be more valuable traffic for you. That’s a question. You should ask yourself.

Frederick Vallaeys: Yeah. And I mean, what I always like to say is how do you maximize clicks with a limited budget? Get the cheapest clicks. That’s how you get the most clicks. And why are clicks cheap? Well, you’re not the only one in the auction, right? So somebody’s probably figured out that click is not worth a lot because it’s lower quality.

That’s why nobody’s competing or fewer people are competing. That’s what makes it cheap. So sure, you can get lots of clicks, but again, the story here is not, we don’t care about clicks. We don’t care about leads. We don’t care about sales. We care about profit. We care about new customers, right? So and then that shifts into the whole conversation and let’s go there next again About communicating your value back to google.

So I worked at google right and one of the 10 principles of innovation at google I think it was number five back in the day was share all information that had nothing to do with smart bidding or had nothing to do with running ads campaigns. But what it was about was as a company, Google is a big company.

Lots of people work there. The best way to get all of these really smart people to work towards the same goal and move in the same direction without having to micromanage everything was to say, well, we got to share all information. If everyone knows what our goals are and everyone knows the information that we’re using.

To try to get to those goals and these smart people will individually maybe take different paths, but eventually they’ll come to the same place. And it’s the same with machine learning, right? The machine learning is a really smart capability, but if you don’t feed it the right information, then how good of a job can it do?

So, so that whole principle of sharing information applies just as much to the machines as it does to the humans. And so we have a slide here about different ways to share your data. So so let’s maybe take a look at this one.

Alex Ioch: Yeah, so it’s really more of a kind of higher level, like different ways you can approach it, right?

If you have a platform, you can always use the Google marketing platform. You can use it for API integration, but more so the more common buckets is really Online conversion tracking. So using the GTAC or Google ads tag, what was I tag as also different names with the same thing for a Google tech manager important.

So tracking that information, or if you don’t think, or like in addition to that, or instead of how, how are you like, or to organize it, You can also import your offline conversions, right? So what happens after the initial first stage of submitting the lead online? So submitting that information there or getting the purchases, maybe people are adjusting their values for retail too.

So that’s one of the other use cases. Exactly.

Frederick Vallaeys: That’s the funny thing, right? I mean, we call it offline conversion tracking, but it could actually be things that happen online. So someone going to your website and filing a return, exactly. That’s an online behavior. But that doesn’t automatically get captured in the conversion data unless you actually push it back.

So it’s called offline conversion tracking or value adjust. But that’s what that’s about.

Alex Ioch: Exactly. So having this information is the first stage, right? So once we pass that information back, then the second is ensuring at the time of import, or even within the interface of Google ads, you assign different values to those different stages, right?

So that you can inform the smart being strategies. What to go for and what’s more important to you.

Frederick Vallaeys: Cool. And then there’s a funny little thing. So I think we’re all familiar with the conversion actions page where there was this checkbox that would say, include this as one of your quote unquote conversions, which would mean manipulating bidding.

Google recently changed it. I don’t think they really talked about it. So we have a slide from it here. Exactly. So in your box now, there’s like a primary and a secondary conversion action.

Alex Ioch: Yes, it’s I thank you for relating to this one. It’s definitely been a change from a UI perspective, but it’s more like, again, just Putting in the different buckets, right?

So the team has tried to make it a little bit more organized. So the way it’s structured now that you have a goals and under each goal, so goal could be like a purchase or goal could be elites and say, you have a leads as a goal, overall category, then you can have different conversion actions underneath each.

And those conversion actions under each goal. Could be either primary or secondary. So the primary one will, if you enable that option, that will mean that not only you’re going to track it as a conversion actions, like including conversions, right? But also inform smart being strategy to use that as a, in the bidding optimization.

So that’s exactly. They called it out here as an interface as well. So just if you don’t want to optimize for that conversion actions, just mark it as a secondary if you’re including, depending if you include an account level or using everything that more like a select optimization and campaign level as well.

Frederick Vallaeys: Yeah. And then Rebecca, there’s different types of conversion. Conversion values.

Rebecca Chiu: Yeah, absolutely. So there’s if you’re just using, you know, target CPA, for instance, there’s, there’s conversion. Information that you could put put in was in this, but then then stack static value. If you use conversion value rules for for instance, you could assign average value for people who took specific actions, so it could be across.

You know hundreds or thousands of that conversion actions of this type of customer. This type of leads dynamic values is where you actually pull in information for each customer early each lead. So this is more a lot more sophisticated in terms of. Figure out pulling in final sales value, if you’re an e commerce e commerce advertiser, for instance and then the advanced dynamic values, a lot of times bring in the lifetime value.

So if you have more sophisticated. model in terms of calculating what the predicted lifetime value when the person actually comes into you pass that information back to, to Google and our machine learning algorithm is going to get smarter based on the value that you actually pass back to us.

Alex Ioch: Yeah.

So the, just to think about it for a second, right, you can import your profit into the Google and actually be to profit. So you can tell, like machine learning to go and learn more and more profitable clients to you if you pass that information back.

Frederick Vallaeys: And so in that most sophisticated bucket, why not just wait for a year and then as these really good things happen, communicate your conversion back to Google then?

Rebecca Chiu: Yeah, I think for us, machine learning is, it’s, you know, you want the freshest data possible, right? So the quicker you actually pass it back to, to Google, the better it will be. Some of the offline conversion, for instance, you could only put any information within 90 days after. The click has been executed.

So, so a lot of times you should, you should try to pull in data as, as recent as possible. So our machine learning algorithm will,

Alex Ioch: will help you to optimize. And one more thing from a, you know, actually from our PMs, right. Who work on this fund. So they really stressing out the consistency of passing information, especially if line imports for the matter of that any smartening strategy is actually going to predict your conversion lag.

Right. So if you have this recurring consistent import, so ideally daily, that the system will know, okay, on average, this type of interaction, say, cells qualified lead. It’s taking this number of days actually to convert. So next time that, you know, that similar person comes into the auction, okay, pretty predict not only conversion value, conversion rate, but also like conversion lag.

So that will factor that in because, okay, we expect that maybe it’s between like next two weeks that lead will actually fruit itself into something a little bit more bottom of the funnel. So it will know that. So that’s why it’s consistent. The imports are important. But if you have a broken one, it’s kind of, you know, messes up a little bit machine learning because you, you have like imported one in one day, another one in a week.

So it’s going to be affecting your conversion life learning.

Frederick Vallaeys: Right. So the more frequently you feed the machine fresh data about what’s actually happening, that, that’s great. And so in a way, I guess it’s better to have almost like. fresher data, but maybe not the exact right answer. Like, don’t wait for the exact right answer, but give, like, directionality to the machine because that can also help it inform which direction it’s going to go in.

And I think that’s kind of a big problem that we often see is advertisers get stuck on being at the perfection. Actually, that was another principle. Google innovation was Do, don’t let Perfect. Get in the way. Get

Alex Ioch: get it done then. Perfect. I guess. Something like that. Yeah,

Frederick Vallaeys: exactly. Something like

Alex Ioch: that.

Right?

Rebecca Chiu: Don’t get don’t use Perfect. Get in the way of good or something like that. . Yeah,

Frederick Vallaeys: exactly. That’s it.

Rebecca Chiu: Move fast. Been

Frederick Vallaeys: since I worked at Google, so sorry. I don’t have a like top of mind anymore. But exactly that’s the same thing here again. It’s you don’t have to have a perfect answer. Like, think about directionality.

Like, was that lead just a little bit better than another one? And if it was just a little bit better, well, then tell the machine, because that means the machine cannot prioritize that one a little bit more over the other one.

Alex Ioch: Yeah. And then get, don’t get stuck to an actual, like new degree devalues. Right?

So if you have, say, calls, if you work for 10 and lead, 50 to you really changing to 51. That leads, I’m not sure how much of the dam is going to actually impact to you. Right. So don’t get hang up on like really needy, greedy details of like conversion values, as long as the magnitude of difference between them are important.

It’s more important to you, right? 1 versus five or 10 versus 50 will be the same thing for machine learning because I know which one to prioritize. So it’s in the day, still like ones and zeros on the backend. So I’m optimizing for that higher conversion value. If that auction will actually entail that and having the people like that.

Rebecca Chiu: And you can always update the information as it goes along too, right? So for instance, I work with a advertiser that has a subscription business. So the refunds tends to come in within the first five days and you could completely You know, pass that information back and adjust that value as you, as you get better and better data.

Frederick Vallaeys: Yeah. As more and more stuff gets returned, which hopefully isn’t the case. I hope not. Goodbye,

Rebecca Chiu: Crowdy. Did

Frederick Vallaeys: you return a lot of stuff? You can hear Crowdy Carson. That was actually one of the problems this year. And so let’s talk about this. But like, how do you correct these values? Right? Because this year, Black Friday, so many people were worried about not getting the stuff that you wanted.

So I bought more gifts and even like last week, I went back to Costco and I returned some of the gifts I ended up not needing, like two months after Christmas happened. Three months after I bought it. Right. So how do you, and these are some of the systems that you see here on the screen for restating values.

And let’s talk about data exclusions for a second. Yeah. So let’s test it out a little bit more give a little context. So primarily use case and Google really careful, always the communicating what to use and not to abuse it for. So to primarily is, and you should be using if if something is broken from your conversion tracking perspective.

Alex Ioch: So if you’re tag malfunction or you double count and, or didn’t count the conversions on that particular timeframe, you can actually apply data for that period of time, right? So that will eliminate basically that period from machine learning. So basically skip that incorrect data you’ve been passing or not passing for that matter to Google ads.

So I think that’s a primary use case. Other reason, and this is actually data exclusions is, I know we have a question probably from the audience is or might be curious of what happens actually from the seasonal adjustment period of time. Right? So data exclusions are automatically bacon into the seasonal adjustments.

If the I know the Black Friday intent user is kind of unusual, usually comparing to like the rest of the year. So if you do apply your seasonal adjustments, you think that You’re within that period of time, your conversion rate is going to spike double. And you would like to be more competitive during that period of time through seasonal adjustment tool.

So that will also have a big data exclusion on the back end. So it doesn’t impact the future learnings. And then once that period of time is over, then the machine learning goes back to normal essentially. Yeah, so that’s good advice on when to use seasonality adjustments and when to use data exclusions.

Frederick Vallaeys: And then the The other types of things you see here, we we did a pretty good, I’ll say great blog post on the Optmyzr blog. So let me put that in a banner here real quick. So you got the link to that. Just go to our blog and we have a really good description of the three different ways of adjusting values, whether it be offline conversion or value rules or value adjustment. All right. So. Next the next step then is we were communicating all these values. So now we’re actually going and turning on these campaigns. We did have some questions from the audience about campaign structures and specifically one of the questions was around, should we enable a DSA campaign at the same time?

And where does that fit in? So that was a question from from Chris.

Alex Ioch: Yeah, it’s a great question. So if, when you think about your structure of your campaign, you’re encouraging to think about as, as of your business, right? What are the buckets, either products or different intent, even if you’re a lead gen, and how much those type of customers or interactions are actually worth to you.

So approaching that from an actual how much is valuable to you and how much return on ad spend you’re looking for on the backend, right? So tracking that, how much you’re getting that. So structuring your campaigns based on your goals. So similar maybe themes of keywords or similar products by Target or an ad spend you’re looking for.

So it’ll be the structure For your account and in terms of the dsa we do see different preferences overall in the market some people prefer to have a little bit more control for the budgets and return on ad spend goals for that particular campaign. So they break it up and it’s in its own standalone campaign.

If there’s no strong preference, then you can create ad groups with the dynamic search ads within existing ones. So it’s, it’s all up to your preference and how much you think you need more control for that extra layer of kind of catch all buckets. So you need to remember the DSA. Is a secondary priority if comparing to a regular search, right?

So what you’re not covering with your budgets, with your keyword structure, or with your match types that will capture the DSA. So your search terms you cannot possibly think of, like I think 15 percent of the queries every day are net new, right? So you cannot Oh my

Frederick Vallaeys: god, did you just say that again?

Okay. No, no, you’re right. But like, this is a, it’s probably like a 20 year old stat from Google now. And for some reason it never changes.

Alex Ioch: Definitely not 20, but yes. This is

Frederick Vallaeys: a drinking game.

Alex Ioch: It’s still there, still official. So still use it, but at the bottom line is, you know You cannot anticipate all the different search terms in real time.

So that’s why the DSA will capture what you are not covering with your match types. Yes,

Frederick Vallaeys: exactly. No, I mean, that’s a good point, right? Like there’s a reason that some of the old structures like an alpha beta methodology still work. That’s why broad match has so long been really useful. But it is something that you have to manage, right?

And so. Now in the day of automation, it’s nice because you don’t have to manage the bit of every single variation that comes out of a broad match. The system kind of handles that for you. And that’s also a little bit the point that Google then often ends up making is like, you got to put these automations together.

If you just use them independently of one another, then, you know, it may not lead to the expected results, right? Broad match is fantastic. But if you’re bidding, 50 for every prod match variation. Like that’s not going to end up well. Right. But if you combine that with smart bidding and then you even combine it with RSAs, and we’ve done quite a few presentations on responsive search ads.

Probably let the machine put the right ad for the right user at the right price for the right query. It’s kind of how that all comes together.

Alex Ioch: One more point to remember is actually that the machine learning learns on this search term level, right? So it’s on that SQU level. For, for your shopping ads and the search from level on the, on the search site.

So you can have different keyword match times. If in your ad group, it’s perfectly normal. Now, even like have a separate DSA one, it will learn which are the more important search terms and also shaded data on the back end, those thousand signals who are tied to that search term and actually learn from this one.

So not don’t overthink your structure as much as focus more on the. combining, giving as much data as possible. So more data and machine learning will actually learn who are the more important customers for you on the back end.

Frederick Vallaeys: Right. I’m glad you brought up that point because that’s another common misunderstanding.

So you positioned it as like, it learns at the query level. The other place where people often confuse things is they think, Oh, I have to have one campaign, but as much data as possible, because that’s how my conversion data. It impacts that campaign when in fact, it’s the conversion action. That’s the thing together with the query at one level and different things at different levels, but it’s really the conversion action.

That’s the fundamental piece of machine learning. And so if you have five campaigns and they all have the same conversion action feeding into it, yeah, they all learn from each other. And then the other, other thing, and Google has a great slide. We don’t have it today. Ready to show, but basically even if you have manual campaigns, manually managed campaigns, and they have the same conversion action collecting data, those conversions will influence your smart bidding campaigns, right?

So there’s these interactions that people don’t understand because it all goes back to the conversion action.

Alex Ioch: Yeah, that’s why they come kind of the important conversion values is so important, right? So tracking that information consistently over time will import like influence what target you’re in that urge on that spend or maximum value will do.

So that’s why sharing that information is important.

Frederick Vallaeys: We had another question from Chris. So let’s quickly go into this one But would you use different bit for portfolios for different campaign groups? And I guess this goes a little bit into the whole structure thing. We’re discussing right now

Alex Ioch: Yeah.

So again, if you do, if there’s a reason you’re branching out campaign in a different I guess the different bucket, so having different campaigns, maybe per geographic location or different budget allocation. You can have a portfolio strategies for subsegment of campaign, A couple, I guess, best practices to keep in mind while you’re doing that.

I guess the goal is to ensure those campaigns you’re bucketing together and their bid portfolio, they have same target, right? So that’s the one goal. Either you want to just maximize. Converting value of those or like target your own net spend. So if you have the same target or net spend goal for those campaigns, that’s a good strategy to use it.

But you want to watch out for inefficiencies potentially what might happen on the campaign level from the budget perspective. So if you’re bucketing around the one beats portfolio, then you better ensure those campaigns are not going to be capped by budget, right? So if you’re trying to stay hit 500 euros on five campaigns.

But some campaigns become budget constrained. It will be harder for overall portfolio strategy to effectively hit that on all consistent over time. So either use a shared budgets. It’s going to be our recommendation if you have on those campaigns or make sure they are just uncapped so they’re not going to be kept on an individual day basis.

It’s not that, you know, we want to get more budget, of course we all do, but it’s more of a consistently making sure the strategy is able to find those queries who fit your return on that spend goal during the day. And if it doesn’t it becomes budget constrained. It will become a little bit more conservative.

It will try to not going to overly aggressive auctions during the day to ensure you hit that goal consistently, right? So it can become a little bit more slower to hit your goal. So that’s why if you’re bucketing campaigns, so number one, make sure they have the same goal, right? Number two, they are not budget constrained or they have a shared budgets.

Frederick Vallaeys: Hey, Rebecca, I don’t know if you want to take this question, but Brett was asking, Brett’s one of our frequent watchers. So you and I have talked about this concern too, but so as we start putting more conversions, conversion actions basically into the system, like is there a risk that people start inflating the numbers?

Rebecca Chiu: Yeah, I think a lot of times people don’t understand that the conversion alley rules is measuring the incrementality on top of what you, what Google already observed. So for instance, if Google already observes that your, the sales transactions are worth X for these types of customers and your LTV, let’s say is proportional to the sales transaction, then You shouldn’t be using conversion value.

It just conversion value rules on top of that. However, if the lifetime value is twice as much as what school is observing right now in terms of your sales transaction, then you should say, Hey, two X of these types of customers from this audience from this geo from these kind of devices. So so there’s definitely things that you need to read into the fine lines when you’re actually using conversion value rules.

And a lot of times also, you know, margin data, for instance, Google does not have information on your margin data. So if you want to put that information in and a lot of information could be index as well. If you want to keep that information proprietary. So we know relatively these type of transaction has a higher margin of, let’s say, 70 percent versus these values of 30%.

So we’re actually optimizing to your true value business, business value at the end of the day, rather than just sales value.

Alex Ioch: Yeah, absolutely. Like don’t overthink conversion value rules is that it’s only for cases, but you’re not communicating that information already to Google ads. And most of the cases, if you do a full time conversion import, you probably already have it, or even dynamically.

Tracking your transactions of most of the e com already do you’re already communicating that so it’s only for edge cases where you think Like maybe one location is more important to you from like lifetime value and you don’t have a way to import that lifetime value for example back or profit margin as Rebecca mentioned.

Other thing is that if you want to implement conversion value rules, so you’re actually going to be able to segment it and it’s called value rule adjustment segmentation where you’re going to be able to see original value rule applied, original value rule not applied, and adjusted values. So You’re actually going to be able to break it down to see how much in each bucket falls you know, for the thing that conversion value was adjustment.

Frederick Vallaeys: Yeah, and the reason this is such an important question for a lot of people watching today is that you’re still generating reports. Your boss is asking you, hey, how much value was driven the moment that you start. Kind of like value rule, adjusting it. These values are no longer what you intended it to be, right?

They’re helping the machine move in the right direction, but it’s not what your boss wants to see in the reports. So like Alex is saying, there are actually multiple columns now in Google ads. And so if you pull the right column data, you can still report what your boss wants to see in the reporting engine.

But you can have the machine learning system use the adjusted values for value rules. Optmyzr has done a lot of work in this arena. So if you want to set better value rules, if you want to have these reports be clean, just reach out to our support team and we can help you. All right, so we don’t have that much time left.

But let’s talk about some pitfalls and best practices, which I guess are somewhat similar to one another. Right. So, but we have a good slide of best practice and a couple of key things to keep in mind. So let’s take a look at that here. Rebecca, you want to start us off on this one?

Rebecca Chiu: Yeah, absolutely.

So the common pitfalls in VBB, there’s a lot of different key metrics that you could focus on, but in terms of using value based bidding, so you should focus on. Ultimately, your your max conversion value, your conversion value or target Ross in terms of testing for this. So, so really keeping that to change.

Having your goal in terms of right metric is very important. The volume of conversions is important to give the test enough volume conversions to say definitively the test, you know, the experimental group is better than the control group, given your old strategy and your new strategy. Having that we I think we are we’re recommending 50 conversions to basically make that call.

A lot of times also learning period conversion delays are the periods that you should exclude when you’re evaluating performances. So it takes Google about a week in terms of figuring out the algorithm to to optimize for your business strategy. excluding the performance during that learning period will help you to figure out exactly what the performance differences between control and experimental group.

And then on the conversion delays as well, it takes time for the conversion to come in, right? So sometimes even for legion, potentially could take two to three weeks. So excluding that period at the end, so you actually allow time for those, those conversion to come in. To really evaluate the performance values is really important as well.

Alex Ioch: Yeah, I mean, sorry, but those are expectations, right? So it’s not like every single campaign is going to have seven day of learning period or two weeks of conversion lag. So one thing you can actually check and then I wish they had a better way of intuitively knowing where it is. But once you’re at least tragic type.

So we’re in the name of your actually bit strategy, where it says maximize conversion value with TRS. So that’s actually the blue hyperlink we can click and open up bit strategy report. So, perfect. And it will look something like this, where you can actually see the blue, Line there. It will tell you how long your conversion lag is for that particular campaign, because it depends on your product based on your search terms is going to be different lag until people actually convert.

And then Google, your conversion is recorded of day of interaction and respect on your, of course, attribution models. And if you have like more spinning wheels, there are gray bars is going to be your learning period. Right? So take a look at that. Be sure for super, super helpful. It gives you a ton of information.

What happens and then, then you’re going to be able to adjust your window of analysis to to see, are you actually hitting the target and are easier conversion value going up?

Frederick Vallaeys: Yeah. So, I mean, a lot of the same things we have to think about now, time lag volume of data before we make decisions, like, Machine learning and automated bidding doesn’t magically solve these things.

It’s job is to set the bids. It doesn’t mean like it can do, it can get great results in five seconds as opposed to five weeks, right? So keep those things in mind. It also helps by the way to start with the TROS that Google recommends, which is close to what you have been performing at. Again, there’s like pie in the sky thinking like, Hey, I would love to have a 10, 000 return on ad spend.

Sure, who wouldn’t, right? But let’s be reasonable. Let’s start with something that’s a little bit closer to where we came from. And then let’s optimize it from there.

Alex Ioch: Yeah, and then the reason why is because you want to start with the same competition level. Your TRAS is actually the level you can use to be more or less competitive, right?

And you want to start with the same level you used to be before. Of course you can adjust it, but it’s going to be either more or less aggressive. And I would definitely encourage you to use recommended level if you’ve been tracking your conversion values properly, because recommended by the settings level will also incorporate your conversion conversion lag, right?

So we’ll factor in there’s more conversions going to be in the future. So sometimes it’s 30 days, but a little bit different depending on your conversion lag. So definitely take into consideration recommended in the settings.

Frederick Vallaeys: Cool. All right. So these are some best practices. I guess those lead straight into the pitfalls. So should we talk about pitfalls for a second? Rebecca?

Rebecca Chiu: Sure. Yeah, sounds great. You have this slide on that one or the first one is a lot of people don’t include. Actually, a lot of people include the learning period conversion delay period.

in evaluating the performance data. That’s probably one of the biggest things. And then don’t give enough time to reach a conclusive result. So you should definitely look at your statistical significance level when you’re running these to make sure that it’s getting to the to the level that you want when you’re making a call, whether in a new biz strategy actually working for you or not.

And then on top of that, A lot of times people make significant changes to their, their, their campaigns. And sometimes they, if you have absolutely have to make these changes. Then make it to the control group as well as experimental group. The, but the best practices is actually to minimize as much change as possible during that time to give the algorithm a time to learn and to fully test the strategies.

The third one is what we talked about, Fred, in terms of conversion value rules, people think, Oh, well, I should be value this particular customer lease from this region much more than this. But a lot of times smart bidding is actually observing that already. Through its algorithm. So those, if it’s observable, then do not use value conversion algorithms to do it, but communicate incremental information and things are not observable from a Google perspective.

Frederick Vallaeys: Cool. Well, there you have it. TCPA. It’s great. T Row has way better value-based bidding. Way better now. You know why? Google is saying that. We at Optmyzr believe it too. So we’re building a lot of capabilities to help you with that. For those capabilities, you can reach out to the support team. You can read our blog.

We did a recent post that you can see right there. But yeah, I mean so regardless of what type of advertiser you are, what you sell, what kind of leads you do, how big you are, value based bidding certainly is the smarter way to optimize your account. So Now, you know, give it a shot Google. Thank you so much for explaining this to us and sharing a lot of great content.

Really appreciate your help on this webinar and producing the blog post. So that’s it for this episode of PPC town hall. We’ll be back in a few weeks. If you like these videos subscribe, thumbs up these things. We’ll keep doing more of them. We’re also going to be making this available as a podcast.

So keep coming back. Thanks for watching and we’ll see you next time.

Alex Ioch: Thank you, Tim. It’s a pleasure.