Use Cases
    Capabilities
    Roles

New Strategies for PPC Bid Management

Mar 24, 2021

Watch or Listen on:

Episode Description

Succeeding in an increasingly automated world is no cakewalk. To get full value from your budget, your bid strategies should be aligned with campaign and business goals. On the eve of PPC Town Hall’s first birthday, we spoke to Google and other industry experts to get their insights on nailing bids.

This panel covers:

  • How to optimize automated bidding
  • How to keep doing bid adjustments with Smart Bidding
  • How to feed your business data to Google for better bids
  • What’s happening with the retirement of tCPA and tROAS
  • How attribution model selection impacts automated bidding

Episode Takeaways

How to Optimize Automated Bidding:

  • Emphasize the importance of feeding accurate and comprehensive data into the automated bidding system to enable it to perform optimally.
  • Highlight the significance of understanding business objectives and aligning them with bid strategies to maximize results.
  • Discuss the need for continuous monitoring and adjustment based on performance analytics to ensure the automated system remains effective.

How to Keep Doing Bid Adjustments with Smart Bidding:

  • Explore the balance between leveraging smart bidding capabilities and manual interventions to fine-tune campaigns.
  • Suggest practical approaches for integrating human insights with automated bidding to address unique market conditions or business goals.
  • Stress the importance of staying engaged with campaign management, despite the automation, to capture nuanced opportunities.

How to Feed Your Business Data to Google for Better Bids:

  • Explain how detailed and accurate business data, such as conversion values and customer behavior insights, can enhance the performance of Google’s smart bidding.
  • Discuss methods to effectively communicate this data to Google, such as through offline conversion tracking and enhanced conversion setups.
  • Encourage the use of segmented data to refine bidding strategies across different customer profiles or product categories.

What’s Happening with the Retirement of tCPA and tROAS:

  • Clarify the implications of these bidding strategies being phased out and what alternatives Google offers.
  • Provide guidance on transitioning to new bidding strategies that might replace tCPA and tROAS, ensuring minimal disruption to ongoing campaigns.
  • Discuss the potential impacts on budget allocation and overall campaign efficiency.

How Attribution Model Selection Impacts Automated Bidding:

  • Delve into the various types of attribution models available and their influence on automated bidding effectiveness.
  • Recommend specific attribution models based on campaign goals and the typical customer journey observed in different industries.
  • Highlight the consequences of choosing an inappropriate attribution model, such as misallocated budgets or misleading ROI calculations.

Episode Transcript

Frederick Vallaeys: Hello and welcome to another episode of PPC Town Hall. My name is Fred Vallaeys. I’m your host. I’m also the co founder of Optmyzr . We’re almost coming up on the one year anniversary here of PPC Town Hall. And one thing that we constantly have to do is figure out what’s the next topic that we should talk about to keep it engaging, to keep it lively, and to keep it interesting.

We’ve definitely heard your feedback over the past year that you want to talk more about the tactical stuff and less about the touchy feely and you know, the big thoughts you want to come away from these sessions with something you can actually do. So for that reason, we are going to go back to a favorite topic, which is bid management.

I know we’ve talked about bid management a lot, but every time we talk about it, we seem to come up with some great questions, some great new strategies. So that’s what we’re going to do today as well. We’re going to cover that topic again. Now, one thing I should remind people of is that our panelists are going to be expressing some opinions and views and those views and opinions are their own and don’t necessarily represent the views and opinions of the companies that they represent.

So with that said, hopefully all the panelists will be able to be lively and give us some great insights. And we’re going to talk today about bid management. PPC Town Hall.

All right, bit management and a great set of panelists, a few people that you probably recognize from before. So we’ve got Peter and Andrew coming back, joining us again. And we’ve got first time PPC Townhaller, Laurel Galloway. So let’s say hello to the newbie first. Laurel, welcome. Thanks for joining us.

Where are you calling us from today?

Laurel Galloway: Thank you. I’m calling from Columbia, South Carolina.

Frederick Vallaeys: Nice and so one thing we ask Our panel or our viewers to do is use the comment section to say hello and tell us where you’re calling in from today And that just to get you warped up so that you see that you can use those comments on the youtube page to also Ask us questions.

We’re alive. We’re you know, some people did experience a couple minutes delay So if we don’t get your question right at the start or when you ask it, that’s probably because that delay But do ask us questions. Laurel, what company do you work for? You

Laurel Galloway: I work for closed loop agency based out of California, and we have a lot of remote employees.

I’m lucky enough to be here as one of our east coasters.

Frederick Vallaeys: Nice. And what role do you play at closed loop?

Laurel Galloway: I am a senior vice president of digital advertising, kind of leading up the client services division.

Frederick Vallaeys: Okay, so I bet your clients ask you once in a while about bid management.

Laurel Galloway: It does come up. Yes.

Frederick Vallaeys: Not surprising.

All right. Andrew, it looks like you’re still in Southern California slash Northern Mexico.

Andrew Lolk: Yeah, I think for this round, we’ll, we’ll we’ll say Mexico. I’m I’m located in Mexico at the moment. It’s we almost have the same weather as Southern California, but a bit more tacos, bit more tequila and mezcal.

So enjoying it.

Frederick Vallaeys: Sounds like a good life. And it’s so good. I hear that the views that you represent actually are the views of your company. So what’s that company?

Andrew Lolk: So I found it I just thought that was funny because like, again, it’s savvy. It’s, it’s, it’s me and 10 others. So so if you hear me say something then it’s definitely also the opinion of,

Frederick Vallaeys: of savvy.

All right. Well, I know you’re very opinionated, so looking forward to to hearing what you have to say today. And then Peter from Google, thanks for joining us again and giving us a little bit of the behind the curtains look and understand how smart bidding really works where are you located today, Peter?

Peter Oliveira: Yeah, Fred, thanks for having me again. I’m in Raleigh, North Carolina again.

Frederick Vallaeys: Nice. Anyone else calling in from North Carolina? Let’s see it in the comments if you are. So good. Now for today’s town hall, we’re going to switch up the flow a little bit. Typically we do a bunch of talking heads. Very light on the presentations, but for this session, I actually was asked by a couple of people to present the slides that I had for the paid search association virtual conference about a month ago.

So, I did a session there on bid management. You can go to the PSA dot org paid search dot org and you can get the videos from that live stream. So if you want to see that presentation raw, it’s there for you to pick up. But what I thought we do today is I’ll go to the presentation, few slides at the time, and then we’ll just pause and I’ll actually have.

Experts tell me, am I right? Am I wrong? What did I miss? What else should we have added? So so that’s going to be the format today. So let me show you the slides We’ll keep the panelists right there on the side so we can talk about it. But yeah, let’s talk about bid management Okay. So the first section that I had covered in the presentation was who is smart bidding for when does it work the best?

And thanks to our friends at google and some internal research we did We pulled some stats across the Optmyzr user base to see who was getting the best improvements in conversion rates between smart bidding and manual bidding. And so I wanted to highlight a couple of things here. So first of all, if you look at the relatively high spending accounts, Right.

And high spending is relative, right? We’re not talking necessarily about accounts, just spending over a million, but kind of like in the broader spectrum, spending over 15, 000 a month on Google ads. So we see that there’s actually really good improvements in smart bidding bank conversion rate performance there.

I think that’s probably because smart bidding. Bigger accounts have more data, so smart bidding is able to do a better job of boosting those results. On the other side of the spectrum, you have the small accounts. And I think Andrew and Laurel as agencies probably have some opinion on this, but it tends to be that if you have a smaller account, less spend whether you’re working with an agency or working in house.

Less spend usually means you devote less time to fixing these accounts. And so that probably speaks to the fact that mediocre management is not going to do as good of a job as automated management from smart bidding. And then the one that’s maybe the most difficult to explain is what happens in the middle of this section here.

So with the accounts that are spending. At mid ranges there. It seems like management manual management is pretty good. And smart bidding is not necessarily boosting those results all that much. So I want to pause there and see what thoughts or panelists have. And of course, audience, if you have any questions or thoughts on this, put them in the comments as well.

I’d love to see them.

Andrew Lolk: Yeah. So you, you mentioned the thing about the smaller accounts and I remember having, I am. Debate with the guy who came up with Google at what’s express, which was like the, kind of like the first really big automation that came out of Google. And I wrote a critical source speak article like back in 2013.

And then I met him when he came to the office. It’s one day he was like, Hey, we have some things to talk about. And it turns out, I think we all agree on that point. Like, like the smaller accounts really benefit a lot from, from, from many of the automation, because any agency, any freelancer, anybody that’s just sitting managed these things, it just costs like percentage wise of your spend, it costs a fortune.

If you’re spending a thousand bucks and you get somebody to manage it for 500 a month, which is not it’s not that easy to find somebody good to manage it for 500 a month. Like that’s 50 percent of your budget. If you just spend those 500 extra on ad spend and then let someone automation do the work, you should get more out of it versus getting an agency or freelancer to do it.

In my opinion.

Laurel Galloway: Yeah, I would add to that. Just as long as you have the minimum number of conversions to make automation work well, I’ve seen the same thing. I’ve seen it work really well on smaller accounts, and there’s always the opportunity to add some additional conversion types in there, like micro conversions to give the system more data.

But in general, you have more data, especially good data is going to impact how well the automation works.

Frederick Vallaeys: And Peter, what do you think about that?

Peter Oliveira: All plus one to what all the panelists have said. I’ve done quite a bit of analysis. I pulled some results for this one here and have done other pieces of analysis across, with some partners across hundreds, sometimes thousands of accounts.

And I typically see some results similar to this in the smaller buckets. Obviously the more data, the better. A smart bidding has more data to optimize towards. We’ll see some stronger results. But it doesn’t mean that if we don’t have a lot of conversion data in there, the accounts smaller that it won’t see good results.

Right. Obviously, we’re using the data that’s in the campaign, but smart bidding also has other sources of data and it has other ways to predict conversion rate that are outside of the outside of your particular performance. Right. So that coupled with the fact that a lot of those smaller accounts do receive less management, we typically do see that increase in results.

And. There’s also just more volatility at, at the lower spend thresholds, right? If you think about it, if an account, if accounts can consistently get six conversions a month instead of four conversions a month, that’s not that much of a jump in volume, but it can be quite the percentage increase, which is somewhat what we see here.

Yeah,

Frederick Vallaeys: that’s a good point. And let’s pop those slides up again. Right? So you’re saying if you get one more conversion, half of a small base of clicks that can be a big percentage increase. Yeah, that makes sense. I know we had a question. So as far as how we look at this. Manual bidding means not Google’s automated bidding.

So it could be manual bidding alongside with some third party tools. You know, could be using Optmyzr s first page Bridger, for example, on top of that. But I think the fundamental point is smart bidding because it has the auction time capabilities Always adds that layer of typical improvement right and I think on the left side of the spectrum here on the bigger spending accounts We’re going to talk a little bit more about how do they achieve these much better results because I think it’s not just smart bidding But it’s smart bidding Plus some of the optimization tactics that we’re going to cover in a minute.

Right. Tends to be that bigger accounts usually have better knowledge of what these tactics would be, and they probably have more of the staff and the tools to be able to do these tactics. And so that’s why they’re able to deploy smart bidding. And I’ll just get the base level improvement of that, but then take it to the next level as well with an optimization.

All right. So thanks for weighing in on that. There’s actually one more question here from Sarah. So doesn’t. Search partners use automated bidding. Even if you are on manual, though, Peter, do you understand that question?

Peter Oliveira: Yes, I do. So just from what I remember for search partners, there’s two options there, there’s smart pricing which is pretty similar to smart bidding in that it’s trying to keep your, your, your CPA consistent across different networks.

So there’s some element to it, even on. Even on traditional even on manual bidding. But I do think there might be some incremental benefits you get with, with fully smart bidding. Not sure if any of the other panelists have more experience.

Frederick Vallaeys: It doesn’t look like yeah,

Andrew Lolk: on the topic. I don’t, I don’t, yeah, I don’t have much, like, I, I don’t, I, so I, I, I casually forget Search partners exist nowadays. I, I was all into it when just started PPC, but nowadays, I, I casually forget that it’s there.

Peter Oliveira: Yeah, I actually I have tested out to research partners quite a bit.

I just haven’t tested it on the manual end. I can talk a bit to the results that I’ve seen for smart bidding. For a few partners that I’ve worked with before they hadn’t seen the strongest results with search partners in the past like a few years ago. But now when we’re coupling that in conjunction with smart bidding, the results that I’ve been seeing in my test have been overwhelmingly positive.

If you think about it, it’s right by using search partner, you’re opening yourself up to more potential conversions. And if smart bidding is working well, when the machine learning is working, well, more potential conversions could also mean more conversions for you to capture, right? So I have seen some pretty solid success on smart bidding with search partners being used in conjunction, but I can’t speak much to the manual bidding experience.

Frederick Vallaeys: All right, and that’s a point where we’ll have a slide on later But basically once you get smart bidding sort of figured out and your goals are set up correctly Then you can really open up the throttle and expand into new networks because it basically figures out pieces that might be relevant Still sets the right bid for them and gets you more conversions at the right price it’s really good to see my my old buddy from google greg powell on the call and he’s clarifying that in fact People are asking about the smart pricing component.

So Peter’s answer there makes a lot of sense. It’s also someone asking if the Optmyzr rules affect the way Google’s machine learning algorithm works. So that’s kind of a complicated question. So we as Optmyzr or as a third party tool, You can’t really influence directly the machine learning model, but what you can do is you can actually tell the machine learning what you’re trying to achieve.

And so that comes into the importance of goals, which is actually the next section I wanted to talk about anyway. So let’s quickly look at this, right? So what’s the importance of goal setting. to get the machine and the smart bidding to do the right thing. So this is pretty fundamental, but you know, I had these slides, but let’s take a look at them anyway.

Right? So the problem is if you focus on having a high margin, then you probably set a high target return on ad spend because it maximizes your profits from a single sale. The problem is that limits your volume. And so you’re getting that little bit of profit, but And at the cost of not having a lot of revenue in your business.

So just setting a high target ROAS is usually wrong on the other end of the spectrum setting a low ROAS target is going to focus on revenue. You’re going to bring in lots of revenue, but eventually you’re going to start losing money on the incremental conversions that you bring in. So another way to look at this is what you’re really trying to optimize for is not target ROAS, right?

Your boss is probably not saying, Hey, Fred, can you make sure we have a 500 percent ROAS for our ad campaign? What the boss is probably saying is give me the most profits or maximize my revenue. Those are business styles, right? So when it comes to maximizing the profit there’s this curve and it goes up in the beginning.

You get more profit, but then it starts to decline and you end up getting less profit. So both at the low target ROAS. And the high target ROAS, you’re not maximizing profit. The problem is how do you find the target ROAS? The thing that you have to tell Google to that the machine learning is going to work with to maximize that profit.

Right. And that’s a bit of an unknown. So this covering that point is difficult. So Let’s pause on that for a second. And I think Peter had some thoughts on that as well.

Peter Oliveira: Yeah, I think it’s really important to look at the impact that this has on profitability and honestly, ultimately the bottom line for the business, right? Even if it comes down to things like customer lifetime value. All right. Some may perceive like an increase in ROAS or a decrease in ROAS, for example, as being something negative.

But in reality, what we need to do is look at how that impacts the profitability. Cause in some cases we might have a less efficient ROAS target, but that will open us up to more sales and those sales will ultimately have a positive impact on our bottom line. Right? So the, the true, like the best ROAS is something that’s really should be an iterative process.

And really we need to match that up with our business data to see what targets and what results actually maximize impact for us.

Laurel Galloway: I can add to that a little bit. We’ve had a, Cool client experience the last six months where we’ve actually divided campaigns based on profitability goals. So we have one group of people that’s new customers and one group of people that’s returning customers and together with our client they were able to determine that the difference in the ROI target is profitable in one week versus profitable in one year.

So if we, they’re totally okay having a lower ROI for new customers. And then they obviously want that higher ROI for the returning customers due to lifetime value and profitability. And it’s, it’s worked out really well for us to segment them into two different groups and bid accordingly.

Frederick Vallaeys: Nice, interesting way to do it. It makes a lot of sense. So, and actually Laurel, somebody was also asking, like, what were you talking about with those micro conversions? So I know we got a slide later, but let’s quickly explain that a bit more.

Laurel Galloway: Right, so tracking more actions on site than the final step.

So lead gen client, you might track certain button clicks on site or a form fill versus only tracking all the way down to the lead or the MQL. If that makes sense. So basically giving the system more triggers and more data to help get information about who these people are and their propensity to convert,

Frederick Vallaeys: And much more to come on that in a few slides.

So stay tuned a couple more questions, but we’re going to table those for now because we have slides coming up that will talk about these things. So let’s go back to the slides here. Let’s talk about the other thing that informs the machine learning models, and that’s really how to pick the right attribution model.

So this is a point we’ve been making for a long time, but do not use last click attribution. Please, please, please. Because last click attribution under values The things that are at the upper funnel. Remember that people generally don’t go and do one search to purchase something. They usually go and research.

So it takes multiple steps to move them down that funnel or down to consumer, the customer journey to get them to purchase. The problem is if you’re using last click attribution, that you’re basically saying the only value that I ever get from doing PPC is on that last search before they buy something.

So that says there’s no value. At the top of that funnel that’s problematic because as a human you will Recognize that some of these searches that happen early on and the keywords that are corresponding to that That those are valuable that those are relevant. So you’re not going to make decisions to turn those keywords off Smart bidding and automated bidding is not that smart if you don’t tell it that these things are driving value for you Then they’re going to give up on them.

They’re going to turn them off and your funnel is going to get narrower and narrower. So a recommendation is to use a time decay model, which is very close to last click attribution. It’s really safe to switch over to that. You can do a position based attribution model. If you want to really prioritize boosting the business and really building your brand.

And then if you have lots of data, we think data driven is the best approach because it is machine learning based and it is kind of the smartest of all the models out there. So any thoughts? On attribution models, Andrew.

Andrew Lolk: So one thing that, so first of all, I think this is the, this is super smart and, and most people should really like change that last click model.

But I think one thing that we’ve seen that once you start hitting a certain volume, then your branded terms your display retargeting, it eats up a lot of conversions, even if you’re using something like position based, it’s still 40 percent of the value of that. One conversion that goes to, for instance, your branded your branded keyword.

So in some cases what we’ve done is we’ve actually separated display out into its separate own separate account, separated YouTube out in its own separate account and done actually done the same with branded keywords, just to try to see how can we get enough volume into those more upper funnel keywords and activities.

And also like the same thing on display, the same thing on everything, because up until I think it was last month, there was nothing that Google could do in terms of measuring conversion journeys across search and display. Okay. So data driven attribution and position based didn’t work across display and search.

So again, splitting out to separate accounts, if you really want to be aggressive with with, with scaling accounts have, we’ve seen that a lot of success.

Frederick Vallaeys: Bingo card for Fred was on mute while he was talking. Okay.

Andrew Lolk: I thought it was me.

Frederick Vallaeys: No, you’re not. So I think we’re touching their own portfolio management and remember that a lot of automated bid management is going to treat it like a portfolio strategy. So it’s going to try to hit the average goal.

And that means some things may be far above the target and other things far below it, but on average they hit your target, right? It’s a kind of a volume maximizing scheme. But, but you’re making a good point. So if you feel that maybe too much of that really cheap traffic is substituting for really poor quality traffic, it does make sense to maintain a split in structure so that you can really optimize each of these buckets towards the best performance I think Laurel, your boss wrote a book average is losing. I love it. I love that title because that’s exactly what it is, right? Portfolio means you get an average performance, but it means that there’s little pockets of underperformance that you might not be aware of. And that happens a bit management too. So I like what Andrew is saying as far as like split up campaigns. Anyone have any thoughts on this question that we’re getting about data driven attribution of the conversions drop below the threshold changes to linear. Why would linear be the default? I

Andrew Lolk: had to pick one. I think that’s the, the, the honest answer like that, like had to pick one and linear doesn’t, it doesn’t favorize a first click, which Google could get into trouble with. It doesn’t favorize last click, even though you could consider last click as being kind of the default, most accounts.

So why doesn’t it? We’re right back to, to last click, but I personally, I just, I think it’s linear is the closest thing you come to data driven linear is you get the same same across the board without the data driven high emphasizing one step or the other, I believe that’s why. If I had to explain it on behalf of

Frederick Vallaeys: Google.

It makes sense. Think like a Googler, Andrew, who knows? Maybe they’ll hire you next. Experience with cross account attribution.

Anyone want to talk about that? So, I mean, you can do the cross account conversion measurement. I don’t know, Dan, if your question is about like, do any of the conversions or the attribution fall off? I guess somebody made the point that yes, it it didn’t reliably work between display and search.

So I mean, I, I guess the only advice for, and then I’m sure you’re already doing this, but don’t rely on one measurement tool, right? Don’t just don’t measure just with one tool because it’ll measure in channel within Google ads. So you want to take the broader view as well. But I think ultimately.

No model is going to be perfect. And it’s just like, which one is close enough to representing kind of the outcomes that you want to drive? There’s another question that’s coming in for Peter here from Vince. Don’t know if you want to take this one.

Peter Oliveira: Yeah, I saw that. I think in general for, for smart bidding to.

To use to be able to optimize for specific conversion values or for conversions, but it should be included in the conversion column. And if it just appears in the old conversion column, it’s probably not taking that into consideration as from my experience so far.

Andrew Lolk: Isn’t that where we just make sure it conversion

Frederick Vallaeys: sets today.

Andrew Lolk: Isn’t that where we, isn’t he just supposed to, to, to select it as count in the conversion volume or column or or otherwise create a conversion set where it gets included.

Frederick Vallaeys: Yeah, that would make sense. Right? So yeah, I mean, Vince, maybe that’s a question that we’ll dig into a little bit deeper. If you’re seeing any issues that we can just email support at Optmyzr .

com and ask about town hall. My team might not be able to answer it, but I’ll get it to me and we’ll help you out. Got a little sidetrack here, but sidetracks are always fun. What’s everyone’s take on Google stepping away from keyword targeting for search? So I’ll argue with that premise.

First of all, like, I don’t think Google is stepping away from keyword targeting. I mean, I think we still have keywords. I think what you’re alluding to is we’re just getting fewer query points. I think we have less control over exact match, truly being exact match without close variants. But I think ultimately like the keyword is still such a, an amazing expression of intent that I don’t see Google doing away with it.

It’s just that our controls around it are shifting. Anyone disagree? Agree.

Andrew Lolk: No, personally, I agree. And I think the latest the latest on privacy changes and third party third party data and all these things that Google came out with a couple of weeks ago, I believe kind of just underlines the heavy reliance on search intent.

Behind everything that has related to google ads. So I think from that perspective Everything from google ads will still be keyword targeted for a long time Even though many of us had kind of predicted going the other route with all the audience Data that’s that’s happening But i’m i’m less bullish on on that moving forward actually than i’ve ever been though

Frederick Vallaeys: Right, and I federated learning of cohorts.

And it’s a good point, right? Because third party cookies are going away. It does make it more difficult to target to an audience. And those audiences are no longer individuals, but they’re becoming groups of people. So as a result, yes, you’re right. It’s the keyword actually takes on a more important role until.

We can figure out how flock really works. But yeah, I mean, this kind of leads nicely again into my next section of slides. So let’s load those up on the screen. So rethinking what we manage in PPC, right? So under the assumption that we might no longer manage keywords, that’s one premise. I don’t necessarily believe in that, but I do believe that historically Google has given us this Very easy system.

You know, you pick a keyword, you set a bid, geo target, language target. That was it. That was a 20 years ago. But then they started adding layers and bid adjustments and all kinds of things that you could do. And so we ended up managing a tremendous amount of detail when really the only reason we were doing that was because we want to get lead sales or signups.

But actually that’s what Google calls conversions. They call them leads or sales. But what I really want as a business is I want not a lead. I want a paying client. I don’t want to sale. I want a profitable sale. I don’t care that much about signups. I care about signups that produce revenue, right? So there’s this whole disconnect.

And I think we’ve. Been so focused on managing the details in the bubble of Google ads that sometimes we forgot about managing the things we actually want to achieve and communicating that to Google. And so in the smart bidding world, things are flipped around, right? Everything inside the Google bubble, it’s handled the keywords, the bids, the bid adjustments, like Google figured it out and they’ll do a good job.

If we can correctly communicate what it is we want them to manage. So when I was at Google, I gave a presentation quite frequently, which was like the 10 rules of innovation. For Google. And one of the principles of innovation for Google was to share all information. With your colleagues. And the premise was simple.

It was it was saying that you have smart colleagues who presumably can make really good decisions. You’ve hired them for their good decision making skills, but if you don’t give those colleagues a complete understanding of the problem you’re trying to solve and all of the data to help them solve that problem, they’re not necessarily going to make the right decisions.

So think of smart bidding like that. Super. Capable colleague that you just hired a human who’s going to manage your bids But if you don’t tell that human what you’re trying to achieve what outcomes you’re trying to achieve They’re not going to do it. Well same with the machine the smart bidding needs good data, right?

So it’s much more today about how do we tell it what we’re trying to achieve so that it can help us with that? I’ll pause on that for a second. I see Peter nodding his head. I think he agrees.

Peter Oliveira: Yeah. I think from my experience, smart bidding does a pretty good job optimizing towards the goals that we set towards it.

Right. But there’s so much more complexity that can be achieved on how we set those goals and what we set for it. To give you an example, like if you look at an advertiser, that’s just optimizing towards. One specific type of conversion and compare that to an advertiser that has a variety of different conversion types and they’re dynamically importing like customer lifetime value to those specific conversions.

That’s a, a massive journey that it takes to get there and also a lot more sophistication that we can get in there, right? So there’s still a lot of levers that we can pull within those conversions, within those conversion values. And by doing so, and by making those reflect our business outcomes as much as possible, we’ll make smart bidding that much more powerful for us.

Frederick Vallaeys: Yeah, and let’s the next question here is like, what are those inputs? So let me actually show a few slides and then I want to hear from Laurel. And Andrew, if they have other thoughts on more data you can put in the system, but so I framed this as there’s three levels of smart bidding sophistication, and really the first level is simply telling Google what a conversion is, right?

A lot of advertisers or your clients, if you’re an agency, they will come to you periodically and be like, Hey, my conversion tracking was broken. Sorry, like figure it out. Like how are you supposed to help them achieve business goals if that’s the case? So in the first case just make sure your conversions are correctly reported now Let me give you an example If you just do target cost per acquisition goals and the only thing you’re reporting is a conversion Then imagine the scenario on the left.

You have the globe and you get conversions from two locations the uk And somewhere on mainland Europe turns out that you’ve reported two conversions. You feed that to the bid management automation system. And the next thing you know, is smart thing here in the middle goes to work and that it’s actually more efficient to buy lots of leads in the lower value market where you get lower value for the leads, but it’s significantly cheaper to get them right.

So, but the system will figure that out for you. That’s the beauty of it. And then the third level is basically lifetime value and look beyond the individual sale, right? So start to make some predictions about lifetime value of the client. And this is a point from the Google executive summit, but as the advertising becomes more automated, it’s really about the KPI that you ask the machine learning algorithm to optimize and the data that you share with it.

That’s going to give you the competitive advantage over others who don’t necessarily do those things. And it, it looks like we are having some technical issues. But hopefully people can still hear us. I think it’s back. Yeah, it seems to be good, which means they missed my point, but you guys go and talk about what I just said.

All right. Levels of sophistication and what additional data, how do you make smart bidding perform better with the data that you might have?

Andrew Lolk: So I, so from, from my point of view, I think it’s like, it’s, it’s the most crucial thing in the world. Like, like this whole thing with, with people and you see it across the spectrum on in house and agencies that have like a very solid opinion on the smart bidding doesn’t work or it always works and all these things.

And it’s just, it comes so much down to what kind of data you’re feeding the machine. And how are you working with it? I can’t, like, I think all of us who’s worked with advertising know for like a time that we’ve, we’ve had our ROAS go down substantially for a couple of weeks and we talk to our boss or we look at it ourselves or we talk to our clients like, Hey, ROAS is down right now.

What do you want to do? And I, and often you’ll see, let’s stick with it for a couple of weeks. Smart bid and. Doesn’t stick with it for a couple of weeks and see if, if ROAS kind of just starts picking up at a later stage. It will start seeing that you have a low ROAS in these couple of days, and it’ll start bidding lower and thereby your exposure might get lower and you start ending into this negative spiral.

So again, like constantly working on what are we seeing in the market? Should we decrease the ROAS for some time? Should we increase ROAS for some time? Should we do it for some campaigns, not other campaigns? Do we, do we send the right data in all these things are just so crucial when you work with smart bidding, it isn’t that thing where you just set it up and then let it run forever and forget that it’s there which you often see from people complaining that it’s not working.

Laura, what do you think?

Laurel Galloway: I agree completely. You need good data and enough data to get started. I think one thing that I’ve seen ad managers kind of struggle with is at what point should we switch from manual to automated bidding? And I have been in conversations with A couple of my Google reps recently about this, you know, what is that minimum threshold to switch over to automation and then how patient should you be?

What should you set your original target to? I, I. Can’t stress enough the importance of letting the learning period complete before you make any drastic changes to the target. But I always like to start at actuals and then make small step changes from there to get to ultimately where you want your goal to be.

How do you think about the learning

Frederick Vallaeys: period? I mean, does that depend on the size of the client for you? Do you always run at the same amount? Talk about that a bit. It

Laurel Galloway: varies a lot. I’ve seen it take two weeks or longer, and I’ve seen it take just a few days. I think it comes down to how much volume the campaign has.

Do you

Frederick Vallaeys: kind of rely on Google’s indication of what state you’re in?

Laurel Galloway: I mean, there is an alert right there besides your campaign, whether it’s in learning or it’s finished.

Frederick Vallaeys: Yeah. And Peter, any thoughts on that whole learning mechanism? I mean, I have some thoughts on it too, right? Like, it

Well, so what I’ve heard from Google, and you can agree or disagree or just no comment, but basically that there’s a trigger in the system. So if you change your bids or your targets by a certain amount, then it, by default, goes back into what it calls learning mode. Even though the way that machine learning works, There’s no really such a thing as needing to learn.

It’s just, it’s always learning with every single event that happens. It’s just that you may have introduced a bigger variable into the system. And so, because you’ve really increased your target by quite a bit, it’s able to go after a whole bunch of new traffic that it’s never seen. All the traffic that it’s seen before.

It knows what’s going to happen. It’s still going to bid reliably. It’s just that small new bucket of opportunity. Or if you lower your bids and all of a sudden you don’t qualify for some of the auctions that you had traditionally shown up for. So that’s what introduces the uncertainty, but it’s not like machine learning is just completely offline for a while.

So I think there’s a grain of salt that people need to talk to take with that whole, what does it mean to be in learning stage?

So it depends. For those of us, those of you listening and not watching, Peter is nodding his head to what I just said.

Peter Oliveira: I think you hit on it pretty well. I think our machine learning, Obviously it smart bidding is going to learn a lot about like the current traffic that it’s on. And when it’s exposed to new traffic, it’s going to be exposed to things that it hasn’t necessarily experienced before.

Right. But when it’s on that new traffic that it hasn’t necessarily been learning on, it doesn’t forget everything that it learns before. Right. It obviously, there’s a bit of a curve. There’s a bit of. Adjustments that it needs to make since it’s dealing with traffic that it hasn’t dealt with on your account before.

But at the same time, we still, it still remembers what’s been happening in your other auctions. And it can use a lot of that information to continue to bid on this new traffic. Yeah, it’s not as it’s not as scary as some as it may seem.

Frederick Vallaeys: Hey, Peter, people are saying YouTube is still spinning. Can you tell someone over at Google to go and take care of that, please?

You must know.

Peter Oliveira: I can go try.

Frederick Vallaeys: So apologies for the the streaming issues here. We can still see each other. So I think this is all gonna be recorded. And we’ll share the whole recording efforts. I did see people missed some of my, my points, so to, to kind of bring it back to basics. The more that you can use value reporting, that is your dial to tell the system what to prioritize, right?

How you decide what value to report. That’s up to you. That’s the optimization. And we have a couple of examples of that. So if we go back to the slides, I’ll show you a few things you could do. Okay. So consider offline conversion import, right? So you have a scenario where you do lead gen. You have two examples, one on the left and the right.

So the user comes to your page, fills out the lead gen form, and you report that a conversion happened and that conversion has say a value of one. That’s usually where most advertisers will stop. They don’t do offline conversion import. Offline conversion import is a methodology that you have with Google to later on restate the value of those conversions or to report them after the fact.

And so in case one, if after 90 days, which is the maximum amount of duration of reporting the conversion, if after 90 days that customer who filled out a lead form, You haven’t been able to reach them on the phone. They haven’t responded. They haven’t done a trial. You might want to go to Google and restate that as a value of 0.

  1. So an indicator that, you know, it looked like a lead, but it wasn’t really a lead. Whereas on the other side, if that customer gets on the phone with you, and after 40 days, they decide to buy your product, go ahead and boost that value, right? That is the right type of lead. Now, what was it that made one customer or one lead better than the other?

We don’t know. But the beauty of the machine learning system is that it can start looking for common attributes based on a single conversion. It’s not going to figure out what was the difference. But based on tens, hundreds, thousands of conversions, it might start to pick up on the fact that there’s a geographic difference, the behavioral difference, some difference in audiences.

And so now we can start to find more of those conversions that you actually care about. Whereas if you hadn’t restated the conversions, then it would think that those people who didn’t actually buy from you were just as qualified as elite as those people who did buy from you. So this is a simple way to restate values.

Laurel, do you guys use OCI or any other examples of how this might work?

Laurel Galloway: Yes, for some clients, we have this going. Yeah, and it would, it’s not exactly this example. So I don’t necessarily want to speak to this example to confuse the audience. But we do have OCI being implemented for some clients. Yeah,

Frederick Vallaeys: I’d actually love to hear the example, right?

So because I find that some customers they were like, Well, I don’t do exactly Legion, like, give me other exam. So So

Laurel Galloway: we have a client that has a lot of conversions that they’re tracking, probably 15 things and five or six of them are coming in via offline conversion tracking to Google. So matching back to that G Clinton that we have relative values for each of those that are dynamically being passed back in.

We’re using that for T Row as bidig.

Frederick Vallaeys: Yeah, and actually, let me show a slide here which is kind of to your example. So let’s bring this all back to microconversions, which you had mentioned earlier. And I’ll let you talk about this after I give my slide back. Basically, micro conversions, they’ve long been a thing in Google tracking, right, so there’s micro conversions, which are kind of the little things users do along the path of doing a macro conversion.

A macro conversion being the thing that you really truly want, so someone signing up, someone buying something, that’s a macro conversion. Bye. To get them convinced to buy something, maybe they have to sign up for the newsletter. Maybe they have to watch a certain video on your page. Maybe they have to visit your pricing page.

These can be counted as micro conversions because you can have some sense that someone who visits your pricing page or looks at the shipping rates page. It’s probably in a state of mind, more likely to want to buy. So you still want to value that you want to do something with that. Now, what’s interesting is that in a smart bidding world, it’s no longer about just having these three types of micro conversions, like in this example, but it’s also about giving them some level of weight which you can do through value.

So the thing that I’ve heard though, is some companies, they struggle to say, well, what is the value of someone who registers for a webinar? And listen, we’re doing these town hall sessions and I don’t know the value of each of you watching this and perhaps at some point becoming my customer or staying my customer and the same for my co panelists.

They have no idea, but we do have an idea that if you’ve engaged with us, you’ve heard us talk. It probably makes you a little bit more likely to know who we are, what we stand for and do business with us in the future, right? So. That’s a little bit more value. Requesting a demo. That’s a whole lot more value because now we actually get to talk to you.

We don’t know if you’re going to buy, but there’s a lot more value. So it’s all about relative values. You don’t have to put an exact number on it. You just have to say this thing we think is more likely to correspond to the outcome that we want than this thing. And that way by reporting these values, the smart bidding system now sees, Oh, someone who requested a demo, Fred seems to prefer that.

So let me get Fred more of those types of conversions. Laura, I know you talked about microconversions. So any other thoughts on this?

Laurel Galloway: I agree. I think clients can get really hung up on exactly what that number is. And it’s more important to come up with a relative proxy to give the system versus exact calculation of exactly what that revenue number is because that’s very difficult for a lot of clients to get to.

But if you could work together to just theorize it. This is smaller. This is medium. This is bigger and come up with some values. It’s it’s going to be enough for the system to work off of to prioritize the larger revenue actions.

Frederick Vallaeys: Yeah, and then we have a really good question from vince and it’s directed to peter but what about offline sales delays, right?

So if you use offline conversion imports but it takes five days before something converts Does that impact smart bidding in a negative way?

Peter Oliveira: Personally, I think that should be fine we can double check and get an answer to Optmyzr support, but I think if it’s like within, within a week, at least I’ve seen some solid results with smart fitting.

I can’t speak for how the product we treat it.

Frederick Vallaeys: Yeah, I mean, I totally agree on that. So if you just think about how machine learning in general works, like it’ll recognize that those conversions still happen to just take a little bit longer. Ultimately, it doesn’t, it doesn’t care that there’s five days to the conversion or 10 days to the conversion because there’s still that value that eventually comes out.

And it knows what those attributes were that were in common between those people converting after five days or ten days. Now, if the machine learning was really good, maybe it says, well, if we can choose, I’d rather have someone who converts in a day, right? There’s less money floating around that I have to wait for it to be paid back by a customer. So the machine learning was really smart and it look at that as a factor and say, if I have a limited budget, I’m going to prioritize it to the people who convert fast. And then if I still have money left, then I’m going to spend it on money who convert slowly. But if all your conversions tend to take five days, I don’t think there’s going to be any issue with that.

And that’s regardless of whether you use OCI to report those conversions, or you use general conversion tracking, which has the same issue, right? If it takes five days for someone to convert. There’s that five days of nothing happening and smart bidding is still able to deal with that just fine. I think

Andrew Lolk: just a last note on the micro conversions.

I personally, we were very, very bullish on, on micro conversions with, with all the tracking challenges that have been there for the last year or so. And the tracking challenges that will continue to happen moving forward. I think it’s really opening the eye, people’s eyes for what you’re seeing in analytics and in Google ads, Facebook ads, et cetera, that it’s not really like true anyways, what you’re seeing.

So the more, especially a longer customer journeys happen, then the more you have of those cases, the less you will actually be able to track the final sale back to that initial keyword or that initial activity. So having those micro conversions happening and giving them some kind of value, like Laurel said, Is I think it’s going to be key in scaling online ad spend and online results moving forward I think it will be the next big thing like you’ve seen a lot of people do in the last three four years But I think it will it’ll become a huge thing to actually scale Accounts moving forward because it’s just it’s that like we all know it if you’re selling a bed online It’s not the, it’s not the final keyword that made somebody buy a bed.

Like it never happens. Just got the survey results. 50 percent takes more than a month for one of our clients that sells beds online. So why do we have any value on anything less? So I just like, we’re very bullish on the micro conversions. I think it’s going to be key moving forward.

Peter Oliveira: I totally

Frederick Vallaeys: agree.

Yeah. I mean, the micro conversions they did back so far, but I love that they’re coming back and being more important.

Andrew Lolk: I see. I remember five years ago, I saw somebody added it to a, to an account and it was like, add to cart. And I was like, and then like after a month I was like, Oh wait, this actually makes sense.

So yeah, you learn something new

Frederick Vallaeys: all the time. Hey. And as far as implementation, right. So I think people. You’re very we’re all enthusiastic about micro conversions. How do you do them? And how do you do these offline conversion imports peter? I know there’s some material on the google help centers what should people look for to restate values for example?

Peter Oliveira: Yeah looking at offline conversion Conversion tracking help centers that we have they should detail the process. I think it can be A bit more complex than what we can run through quickly, but I would highly recommend looking into that. And on the broader micro conversion point, I just want to call out to that.

I think some of the issues and some of the struggles that people have had in the past is that if you’re adding a bunch of other conversions and they all have, you know, Different values and matter to you differently. We need to communicate that to Google ads, right? So if you have like an actual purchase in there and you have a product view and we’re not assigning value specifically to those two, we will add, like we’ll treat them as being equally as valuable to you.

And it may not be going after the purchases as much as it does for the product views. So make sure that we’re like incorporating the values for each of those. And if we don’t have precise values, like the panelists have said before include estimates and ratios. If you think. One in every 10 product views turned into a conversion.

Maybe that’s what you want to use, right? I think it’s just helpful to start thinking about things on what type of business impact they’ll have and making sure that those values are there. Because if they’re not, I think you can see some of the issues that people have had with these in the past, and also.

Why people may have started shifting away from them a few years ago. But as long as the values are accurate in there, or at least close, I think you should see pretty good results with them.

Frederick Vallaeys: Yeah, Peter, you reminded me a couple of days ago about one of the other Google tenets of innovation. Don’t let perfect be the enemy of good.

And I think this is very applicable to micro conversions, right? Don’t look for the perfect values. Just put something in that’s relative and then work from that. Start tweaking it over time and get better as you go.

Peter Oliveira: Yeah, absolutely. I think even if we start, if you’re like a lead gen advertiser that has never used conversion values, if you have three different conversion types, let’s say you have like Phone calls, leave form ads, or like phone calls, leave forms, or like booking an appointment, right?

See how much each of those value to you. If one of them is worth three times as more to you, maybe assign a value that’s three times as more, right? And then as we get over time, we might assign a value that’s more representative of the actual business impact that it has to you. But just getting those relative values in there is a good start.

Frederick Vallaeys: And then Peter’s asking good attribution account for micro conversions? I mean, I might be answering this question wrong, but so from my perspective is you still have to have the micro conversions because the machine learning models that to the bidding. They may recognize certain common behaviors across users, but you can’t necessarily rely on them looking at that.

So the more that you can actually specify the value you derive or the relative importance of something, the more you communicate that to Google, the less there’s uncertainty that they may not be considering that. So I would absolutely set up these micro conversions. Give them a value and then let the machine learning loose on top of that better data to do its job for you.

And then like all the panelists, I want to monitor it. Yeah.

Andrew Lolk: And I think if you, if you consider the opposite, let’s say that, that Google did account for micro conversions that you, that you hadn’t specified you, we would all be losing our, our shirts. Like we would go crazy. Why are you doing that?

Like, so I think the opposite of this, like, It just makes like, so it makes no sense that they Google account conversions, that we have not added into the system, no matter if we call them micro conversions or not.

Frederick Vallaeys: Yeah. And let’s see here. So another so the question is what if you have conversions with sales calls, leads, forms, and emails.

And again, so this is a really good thing to maybe investigate a little bit offline, there’s many videos on this, but but that’s exactly why OCI offline conversion import is so nice because it can help you bridge the gap. Between those online events that happen and then your sales team picking that up and moving it through the pipeline of sales with phone calls and maybe in person meetings.

But basically if you have a CRM customer management system, then you’ll be able to find out, okay, that sale happened, it corresponded to a GCLID. GCLID is a really important term to know G C L I D. Go and look that up. That’s Google’s way of reporting which click led to something happening. You can carry that forward in your own system to make those to connect the dots basically.

And I just come

Andrew Lolk: with a super simple thing there because like, I think, I think we’re, I think many of the suggestions that we’re doing are really high level. And I know people have trouble with the offline conversion import, but like if, for instance, if you have call lead an email and you know, to college convert at a 50 percent rate.

Email convert at 25% rate. And the lead form convert at, let’s say 10%, then associate a, a, a value in the conversion value column that’s associated to that. So a call at 50% rate gets maybe a, a five the lead form at 25% gets 2.5 and the email at 10% get, get a one. So you use that in your overall value when you set up your conversions really low.

Like very practical way to just get started on assigning values to these micro conversions or the various conversions that you’re having, that you are tracking but it doesn’t have to go into the whole offline import offline conversion import.

Frederick Vallaeys: Yeah. Makes a lot of sense. Thanks for Bringing it down to a level that everyone here can understand.

So another good question. So from Ben, so we did in fact say that we use offline conversion import to restate the values of either conversions or micro conversions after the fact. So in my example, it was the conversion regardless of micro or macro. But the fact that we’re restating the value, it actually took me a little bit of digging to figure this out.

But you can, in fact, restate the value of an existing conversion after you’ve already reported it. So you just got to look for the right Google help materials or ask your rep. But that is possible. Oftentimes when we talk about O. C. I can. The simpler way to do it is to just add a new conversion, right?

So you could have the form fill is one conversion, but because you have to click I. D. When somebody purchases it, You can have a different conversion, which is the actual sale that you now report back to that click You include both of these conversion types in count towards conversions what google calls it and then that feeds both of those Into the bid management system, right?

So the The conversion that was just a form fill By default now doesn’t have as much value as the same click that had a form filled plus the sale as a separate conversion, but both associated to the same click. So these are some of the more advanced tactics to to make this all work. Okay. So thanks Vince for saying this is a useful webinar.

Unfortunately, there’s a lot more content but we’re very close to the end of our session here. So why don’t we start wrapping it up? And I want to bring it to all of our panelists for any final closing thoughts. And then again, I know we missed a lot of stuff. We’re going to have to do this again, but that’s why we come live every two weeks.

So who wants to say bye bye first and final thoughts?

Andrew Lolk: I’m not muted. So I can, I can start. I forgot to mute. So I’ll, I’ll start. I think. I think with with PPC management and smart bidding it’s I can, I can speak on our behalf. We are as an agency, we’re really technical. We have primarily B2C e commerce clients at high volume and direct to consumer brands with high volume.

So we get a bunch of data and we’ve always prided ourselves on, on manual bidding or Third party tools. That’s kind of man, kind of manual that we have some kind of impact on. And the more we see, the more we’ve just said, like. In 2021, we are moving to default smart bidding, not that we use it in all cases we use it maybe in 50 percent of the cases we have currently, but just because we’ve seen so many times where we just defaulted to manual and just spend way too much time on that side.

So I just, I strongly recommend anybody who sits in house and spend a lot of time on bid management today. Or not spend any time for that matter to, to really try to explore smart bidding, because I really think that for, especially in house, if you, if you’re not that into bid management, it works really, really well.

The only caveat I’ll put to all that is that just. That you’re using smart bidding or automated bidding doesn’t mean you ignore everything else. You still have to optimize your, you still have to look at negatives. You still have to look at the ads. You still have to look at all the different aspects, product exclusions, priority levels, all the other things that go into actually making Google ads work for you.

So smart bidding isn’t the end all solution to everything, but it can really help you with a lot of the time consuming PPC these days.

Laurel Galloway: Totally agree. I remember five or six years ago when I don’t know how many years ago when ECPC first came out and we were like, I don’t know about this. I don’t know about giving Google a little bit more control over bidding.

And that was way before TROAS, TCPA or any of those strategies came out. But my confidence in the system has grown a lot. Over the years Google’s algorithm can get so much more granular with a bid adjustment with all of the triggers and inputs that they know compared to what we can manually adjust ourselves.

So I’ve been a huge fan of automated bidding and completely agree that it doesn’t mean that you don’t have that human touch. Negatives still need to be optimized. Landing pages still need to be tested. But. I would say, I always try to keep in mind what is the business reason that I have for manually intervening in a campaign if The automation is working how it should be.

So sometimes I think it’s hard not to, or it’s hard to hold back from like going in and overriding things, but the systems really work well. If you could kind of find that balance of going in and wanting the control versus giving it up to Google. So,

Peter Oliveira: yeah. And I’ll just add on, like, just really like embrace like the process and like potentially like having to take baby steps with all of this. Right. If you’re using smart bidding today, maybe just start, look at the conversions that you have in there and like. Ask yourself, should smart bidding be treating everything equally or not?

Right. If you have a bunch of different conversion types and they’re not actually the same value to you, that’s a starting point to change. If you’re able to just use relative values to start, if that’s all, you know, just a guesstimate, maybe we start with that, if you actually know the true value that they bring and you have like some bigger benchmarks, maybe that’s the way you want to go.

And if you’re, if you want to set up offline conversion tracking and actually dynamically update the conversion values every time a sale happens, then you can look into that. Right. But don’t get intimidated by the whole process. See what you have right now. Anything that’s in the old conversions column is what smart bidding is optimizing towards.

So ask yourself, should it be optimizing towards that? And if not, think about ways that you can start changing it.

Frederick Vallaeys: Great advice. So thanks to all the panelists for sharing your opinions and views, which are not necessarily those of the companies that you represent. Thanks everyone for watching.

We’re going to be back in three weeks. We do these roughly we do them twice a month, PPC townhouse. So subscribe on YouTube. If you want to find out about the next one. We also have learn with Optmyzr . That’s on the every other week, so not the same week as town hall, and that’s much more in depth about how you can use Optmyzr to do some of these things.

Next session we have is going to be on April 14th Wednesday. It’s going to be landing page optimization and CRO. So it should be a good session then. Thank you for tuning in today. Have a great rest of your day and we’ll see you for the next episode.

More Episodes