Performance Max: Boost your campaigns with these tips, scripts, and checklists

Episode Description

#Performance Max — the campaign type that nobody seems to stop talking about — has drastically changed Google Ads best practices that PPC experts have been building for several years.

That’s why advertisers are looking to figure out what works now to run their Performance Max campaigns more effectively.

This time, we spoke to Mike Rhodes and Cory Lindholm — two of the best PPC experts and practitioners in the industry. Mike and Cory have run several experiments on Performance Max and are here to show their findings and lessons to you.

Mike reveals some fascinating results from his scripts and Cory shares an in-depth checklist to diagnose your Performance Max campaigns.

Tune into this episode to learn:

- How to run more effective experiments by visualizing asset groups better

- What’s working and what’s not working anymore in Performance Max

- How to effectively diagnose low-performing Performance Max campaigns

and more

Episode Takeaways

Effective Experiments with Asset Groups:

• Mike Rhodes shared insights on using scripts to visualize asset group performance in PPC campaigns, highlighting the dynamic nature of asset allocation based on performance signals.

Current Trends in Performance Max:

• The episode discussed the evolving best practices in Performance Max, emphasizing the shift from traditional strategies to new approaches due to changing algorithms and campaign structures.

Diagnosing Performance Max Campaigns:

• Cory Lindholm provided a comprehensive checklist for diagnosing underperforming Performance Max campaigns, covering aspects from conversion tracking issues to external factors like inventory changes and negative reviews.

Importance of Audience Signals and Creative Assets:

• The discussion highlighted the varying impact of audience signals and creative assets in campaigns, questioning the traditional emphasis on creative inputs when the data shows a predominant expenditure on shopping ads.

Machine Learning in Performance Max:

• There was a critical look at how Performance Max uses machine learning, particularly how it reacts to performance signals rather than proactively predicting outcomes, which could affect the strategic application of these campaigns.

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 CEO and co founder at Optmyzr. So we got to talk about the thing that nobody can stop talking about in PPC, which is Performance Max. And the thing with Performance Max is it’s really drastically changing all the best practices that we’ve been building for the past 20 years of doing PPC.

People are trying to figure out what are these new best practices. And so in an effort to get some of these new best practices, we said, we got to talk to the experts who are actually running these campaigns, who’ve done some really exciting experiments. So we got two guests. We got Mike and Cory joining us to tell us what they’ve done.

We’re going to look at scripts. We’re going to look at audiences. We’re going to look at campaign structure and try to get very hands on. And we’ll leave you at the end with a checklist for how to go through your PMAX campaigns, And make sure they’re performing. So welcome to another episode of PPC Town Hall.

All right. And here are my guests for today. Mike Rhodes and Cory Lindholm. Welcome guys.

CORY LINDHOLM: Hey Fred, thanks for having me.

FREDERICK VALLAEYS: Thanks for having us here. Yeah, it’s Mike, it’s been a few years since I’ve been to Australia and we’ve been able to meet in person. Hopefully that’ll happen again. Hopefully again.

Wait, I seem to remember a lot of glasses on the table that

MIKE RHODES: night.

FREDERICK VALLAEYS: Strangely, I don’t remember that. So Mike, tell us who you are, what you do.

MIKE RHODES: I’m the CEO and founder of Web Savvy. So we’re a done for you agency here in Australia, but we look after brands all over the world. I also teach a bunch of agencies over at agency savvy.

com, which sort of fell out the back of writing the book with Perry Marshall, the ultimate guide to Google ads. We’re on our sixth edition of that. We’re told it’s the best selling book on Google ads in the world, which is kind of crazy. Probably do another edition given the the pace things are changing.

And yeah, I’ve been doing Google ads for 18 years, which makes me a dinosaur. So I’m here to learn from Cory. Here’s. All over it and on top of things, and I’m going to ask as many questions

FREDERICK VALLAEYS: as I answer.

MIKE RHODES: I

FREDERICK VALLAEYS: think there we go. Let’s

CORY LINDHOLM: call you a lab instead of a dinosaur. Well, yeah, definitely. Mike’s Mike’s a legend in my book.

I actually read the ultimate guide to Google ads six, seven years ago, so I’m with him.

FREDERICK VALLAEYS: Yeah. Okay. Cory. So yeah, you’re the young blood in this conversation. Tell folks a little bit who you are and where you’re hanging out these days.

CORY LINDHOLM: Yeah, definitely. So I’m mostly hanging out on LinkedIn, a little bit on Twitter.

I’m a paid search freelancer, mostly specializing in Google and Microsoft ads, so paid search. Nearly a decade now, I’d say particularly in the e com space. I do a lot of account management, a very, very select group of larger clients. I also am sort of the ghost advertiser for a lot of some of the best agencies out there that But I may be that guy that’s actually delivering amazing results for you.

So don’t tell anybody. I’m also doing a lot of consulting and training services, not only for premier partner agencies across the globe, but trading and consulting for business owners in house marketing teams, individual marketers that are just coming up in the field. So very vast experience. And Yeah, it’s people come to me, I think a lot of the times because they, they want to get right, they want to cut right through it and just no BS.

What’s the good information? What’s the real stuff? You know, a lot of people, they don’t necessarily want management. They just want the insights as if someone was managing their account at a professional level. And that’s a lot of what it provides.

FREDERICK VALLAEYS: Right. And you go way beyond the theory, you got the hands on experience.

So say what you think is going to happen and then see how it actually works. And so that’s how we get these best practices. So Cory, you’re a first timer on the show. Really great to have you here for the listeners and the viewers today. We’re going to talk about performance max. There’s sort of two angles on performance max.

I want to be very clear. We’re going to spend most of our time talking about performance max or e commerce. For people who sell products online. That said, we all have a little bit of experience with performance max for allegiance. So we’ll start with that for a couple of minutes, and then we’ll cut off that topic people who don’t do e commerce stick around if you want, but if you’ve got other things to do, we understand.

Cause then we’re going to go dive deep into e comm. But as far as P max for allegiance, is that something you have dabbled in? And what are some thoughts on that?

CORY LINDHOLM: Yeah Mike, I’ll, I’ll jump in real quick. So on, on the Legion side again, most of my specialty is definitely in the e comm space, but I have made a performance max quote unquote work and perform very well for some Legion clients out there, particularly the finance space for whatever reason from what I can tell a big part of making that successful for a lot of these brands is going to be.

how well you’re qualifying what counts as a conversion. Also making sure you’re feeding back any kind of offline CRM conversion data back into the algorithm to make it continue to thrive and scale long term. And the reason I say it’s so important to qualify that traffic from performance maxes, some may know the ones that are a little bit, have been around in the industry a little bit longer is that display the Google ads display network.

There’s a lot of spam on there, unfortunately. There’s a lot of great traffic as well definitely. But in order to help negate some of those maybe click farms or spam type traffic, just putting a caption on the site may not actually do a great deal of help. Really, what I’m seeing is if you have a large, difficult form, it sounds like Totally counterintuitive to the CRO experts out there that conversion rate optimization specialists and best practices, but having a lot of difficult forms to fill out.

Maybe it’s like a solver puzzle that takes a little bit of time for someone to go through that can really help alleviate having to work. accidentally feed back into the algorithm, Google’s algorithm to go find more people that are spam or that are click farms. So if you think it’s just going to be a matter of targeting the right country, know that there’s VPNs out there.

These click farms like to use those. So targeting a location isn’t going to necessarily fix that issue. So yeah. Just to wrap that up, it’s really make your forms very difficult to fill out. If you’re going to run performance max for a lead gen company whether that’s like, for me, the finance thing works really well because they have to do a full on credit check just to finish getting the conversion fired.

So that’s just really important. And then making sure you’re integrating your CRM, your offline conversion data and feeding that back into the algorithm. So you have that is a competitive lever in your account, but also just again, to make these algorithms a little bit smarter for you.

FREDERICK VALLAEYS: Yeah, totally agree with PMAX.

It’s so automated, you have to give it the best information possible to make those decisions for you. And so that either means integrating with a CRM system or making it really hard for a bot to fill out that form. Now, Cory, I’ll stay with you for just one more minute here, but. You said, you said you’ve made it work.

Yes. Do you love the challenge of making it work or do you just flat out prefer to stay with the traditional search and display campaigns and maybe throw in a, a discovery campaign on top of it?

CORY LINDHOLM: Yeah, I more the latter. Yeah. I, I like to use it as a nice test to see if we can use a more full funnel approach outside of what we’re already accomplishing and doing well in the rest of the account.

But preferably, I’d like to see what success we can get with. very targeted campaigns, whether that’s video discovery, et cetera, to your point, display plus search and then test performance max on a light amount of budget. I’d rather go that approach then, you know, putting the majority of the available ad spend budget.

towards performance max and kind of hoping for the best. That’s not not generally the strategy that most of my clients are going to want to be happy to go after, though some are exceptions and they do just want to give it a shot. I do, I will say that if you are one of those businesses that are willing to give it a shot in these sort of early days, you have a large competitive advantage from a CPC standpoint because a lot of these lead gen businesses are hesitant to try PMAX.

So you may have an edge up on the competition if you can make it work by getting in there early.

FREDERICK VALLAEYS: Right. And if you’re going to pay a smart PPC manager, then why use the fully automated thing, right? So that’s kind of the argument. If you’re going to fully automate it, just do some of the basics, right? And maybe don’t invest in an agency.

If you’re going to invest in an agency, know that they come with additional smarts that allow you to do things better in those old school campaigns. Mike what are your thoughts on PMAX for lead gen? I think

MIKE RHODES: Cory’s

FREDERICK VALLAEYS: pretty much nailed it,

MIKE RHODES: mate. Not a lot to add. I think, yeah, that the risk tolerance of the client, of the person running the account, of the person paying the bill is a, is a really important point.

Most businesses have spent a lot of time thinking about conversion rates, optimizing, The hell out of their sites. And so to sort of build a new page off to the side, to make it difficult on purpose, to add all of that friction and to really make sure that the quality of that data is good. There’s not many, certainly not as many of the, of the businesses that we talked to that are willing to sort of go through all of that when things are working really well without PMAX.

So for me, until we have to use PMAX for lead gen, find a test, if the risk tolerance is there, if they’re going to do the work. If that data is high quality, but for most, I think for the average business. It’s a, you’re almost better off staying where you are. If you’re already doing YouTube and discovery really well and getting good results from that and find that helps PMAX along, but that’s almost another reason to kind of stay where you are.

Cause if that’s already working really well, you’ve got that dialed in. Do you really want to risk it? But yeah, I just was just auditing an account yesterday and their CPC, CPA was fantastic relative to their old campaigns, which was bizarre because the guy had no conversion value in there at all. And I don’t quite know what the machine was optimizing for a bit of a setup problem, but the quality of the leads seemed to be there.

I think he had a difficult. Site just by accident and it was working well and the numbers were really good for him. So it can work well, but I think on, on balance sticking for the vast majority of our clients to, we don’t need to do it yet. So we won’t, unless you really, really want to push the envelope and explore and experiment.

FREDERICK VALLAEYS: Right. And I think there’s no question that it can look like it performs well because it’s just sort of takes in everything, including remarketing. And it’s sort of the same problem that advertisers had in the past with smart shopping campaigns that combine all the data and don’t tell you the details.

And sure, remarketing audience is going to perform at a great CPA is going to have a great return on ad spend. But that doesn’t necessarily mean it’s finding you qualified new traffic, right? And so there’s some tricks you need to deploy to steer it towards those new customers. As opposed to our old ones.

MIKE RHODES: And it’s very reactive as we’ll see with some of the other stuff we’re going to chat about today, but I’m finding that PMAX campaigns is sort of like, Oh, and it just squirrel and it runs after that little signal really quickly. And sort of throws everything at it. And if that is a fraudulent signal, because you haven’t done all of the work to sort of stop that getting through.

It can spend a lot of money very quickly crappy traffic

FREDERICK VALLAEYS: So be careful buyer beware. All right Good. So, some of the rest that we’ll talk about e commerce mp max I think some of the things might be relevant to an audience that’s more legion focused But let’s shift to the topic that both of you. I don’t want to say are stronger in because you’re both strong in Legion.

It’s just that you don’t use Pmax as much for Legion. And the reason, of course, that most advertisers want to talk about Pmax for e commerce is that it has fully replaced smart shopping campaigns. And so if you still want to run a smart shopping campaign, then you have to use Pmax. That’s the replacement for it.

And then it comes with all this other baggage of also showing your ads on discovery on video on display. So let’s shift into that and let’s talk about the ideal campaign structure when you have e commerce campaigns. So let’s start with the simple one. Do you put your stuff in one campaign or multiple campaigns?

CORY LINDHOLM: Do you mind if I, I go through, this is a sort of exercise I go through with a lot of my consulting clients and I think it would be really helpful before we get in the weeds Like one versus multiple, just talking about the strategic sort of pillars or maybe like dials in order to decide which of these make sense for you.

Would that be cool if we went down that quick?

FREDERICK VALLAEYS: Okay,

CORY LINDHOLM: awesome. So I say pillars or dials dials in the sense that there are certain aspects that we have to consider before we make a lot of decisions on how we’re going to structure, what products we’re going to target, what bid strategies we’re going to choose.

So it’s going to sound like a long list, but it’s, it’s very, very relevant. And I do this every single day with clients, so it’s going to come very quickly, but available budget margins and profitability in particular, the percentage of your revenue that you’re wanting to allocate to PPC on top of your actual percentage of revenue as a margin, right?

So we need to consider both of those, assign your margins, profitability and structure accordingly. Get into some of the more details later. Sales volume, inventory, if you have any inventory clearing goals other internal goals or focuses, there’s a lot of things out there. Improve the quality over quantity of customers for some businesses.

That’s going to help determine structure and what products we want to go after. Expanding into new marketplaces dominating digital placement inventories for just particular products in a, in a marketplace. Sell that maybe some, you want to sell your company soon and you want to You know, set things up accordingly for that.

Those are all very big pillars and dials that need to be considered. We’ve talked about aversion to risk for experimentation in the account. Your affinity to patience, I think is really important. PMAX is, it’s a large campaign. Let’s just say it’s six to seven campaign types rolled up into one. You’re going to have to be a little bit more patient than maybe you’re used to with Google ads, especially those do it yourself kind of business owners that are looking at their accounts every day.

PMAX will challenge your ability to be patient and give the data some time so that you can make some good decisions with the data that you’re getting in. How quickly as far as a proof of concept is needed in the results your pricing and promotional models, whether it’s subscriptions, no LTV, repeat purchase rate is high.

You have predominantly lost leaders that make up your, your sales. Those are all going to, Determine specific strategies, marketplace economies or economics. I should say the quality of your website, conversion tracking, quality of your product feed, your inventory fulfillment, product availability, buying behavior, reporting capabilities.

So I just want to show, at least to start, there’s a lot of little things that are going to say, okay, we’re going to try in today’s call to go over some of the general best practices, but there are a lot of things to be considered in these decisions of whether you want a single campaign, multiple asset asset groups, et cetera.

So I just want to cover that really quick so that people don’t take it at face value and say, oh, I should only do this. Yeah.

FREDERICK VALLAEYS: No. So the answer is it’s really simple. No, but the one thing that I took away from this, I mean, I took a lot away from this, but sort of Google will tell you one campaign, keep it simple.

Don’t worry. We’ll figure it out. What I’m hearing from you is, well, you’ve got a business to run. And when you run that business, you have sort of an overarching goal, but then you have different strategies to achieve that goal. And those strategies will equate to different campaign structures, asset group structures.

So Your business is usually not that simple and straight forward. Likewise, your campaign structure probably is also not that simple and straight forward. It’s going to kind of mimic how you run your business. Correct. Based on all the dials and pillars that you talked about.

CORY LINDHOLM: Yes. So rather than us saying, well, it depends a hundred times and annoying the entire audience.

I’ll just cover all of that and then we can move forward. Okay.

FREDERICK VALLAEYS: Mike, what’s sort of your high level on how you approach the whole structure problem for p max? What you want

MIKE RHODES: me to add to that amazing list?

FREDERICK VALLAEYS: Here’s the one thing, Cory forgot. .

MIKE RHODES: Yeah. Do you have, well, whether you have dogs or cats.

Money. Money or patients, you know, growth or profit. I think that’s a, that’s a really big one. What your, what are your business goals starting there and coming down from the business? What other marketing you’re doing? It’s going to help this and drilling down into digital marketing and then drilling down into, okay, now on Google, what are your goals?

What evidence do we have? What previous data do we have? What sophistication level of the client what spread products, most of your products around about the same sort of price level and AOV, or if there are really widespread there and different amounts of margin across your various products.

FREDERICK VALLAEYS: Now let’s maybe give people a simple answer.

Would you ever recommend a single campaign for PMAX?

MIKE RHODES: Yeah, I, I, I would for a, for a smaller business that’s. Wanting just to get off the ground, wanting to run the things themselves. You’ve just got to, got to know if you’re doing that, that you’re putting a lot of trust in the machine. And if you’re really leaning into that, know that it’s not always going to work well, but if you minimize the risk, lower budget, you’ve got some evidence that things could work.

I was gonna say should work, could work, should work, you know, but if you’ve, if you’ve have no idea what your product feed, if it’s good or bad, if you’ve never run any kind of Google shopping in the past, then I would, it’s difficult. If you’re trying to do it yourself, I would want to sort of start with legacy shopping.

Get a bit more control, get a bit more data, have the ability to have insight. You know, Andrew Locke, I think had one of the most profound sentences that anybody on this show has said. It’s not the control that we miss. It’s the insight it’s being able to get insight from the data. And that’s why we miss having data.

It’s not because, well, we are all control freaks, but it’s not the, the fiddling around that we miss. It’s, it’s being able to understand what’s actually happening. It’s, it’s three, but if it’s a small business, they’re just trying Get some extra sales in the door, then, then yes start with that. Keep things simple.

FREDERICK VALLAEYS: The small business with a simple product line or a simple kind of one service type thing for one location. Yeah. A few SKUs. Exactly. Not

MIKE RHODES: tens of thousands.

CORY LINDHOLM: Yeah, from a starting point I like to say, you know, keep it consolidated to start with the expectation that you, you might expand from that. Right.

And, and that’ll help people that are just getting started with it, but also even the more advanced advertisers. I think Mike Ryan was talking about it with you, Fred, the last time he was on is, you know, you want to start more consolidated, get the insights that you have available. get that data to then later make some segmentation decisions.

I like that approach as well. When you start hyper consolidated, hyper segmented, it, the risk is, is higher. You may see the same search terms competing across multiple campaigns, for example, for different products. How do you deal with that? Then you have to be more reactive. Whereas if it was consolidated, you can make those decisions on segmentation around what the data provides.

So I like to say start consolidated, Then start creating complex, add complexity as you get data to make good decisions with. But if you don’t need things to be separated because of different product feeds by country is a big example or language targeting, any kind of campaign setting, like you don’t need different budgets for different products.

You don’t need different, at least to start different bid strategies for different products, customer acquisition tools, et cetera. Then keep it consolidated, keep it simple and then add complexity as you get some data in.

FREDERICK VALLAEYS: So Mike, I’m going to go to you, but I want to show viewers a really good example of that.

And I think that’s what you brought to the session today is how do you start high level and then segment? So you’ve done this with audiences, right?

MIKE RHODES: So yeah starting a campaign we’ve tried a few different things, starting with just one asset group and putting all of your audiences in there, starting with loads of asset groups we’ve sort of come on balance to a nice happy Goldilocks middle point, which is starting with a few asset groups.

So typically that will be custom match and remarketing often will lump those together because they’re kind of similar. If the customer or the client has a large amount of traffic. Or a particularly big list. We might separate those out. Next one is custom segments. Cory had a great trick for custom segments that I heard on a different video.

I’ll let him explain that. I’m not taking any credit there. But all of your different custom segments in another asset group, and then another asset group for your in market and in market other, which are amazingly precise and specific these days. And maybe if you’ve got some evidence from Google analytics, maybe an affinity category or two going in there and then what we’ve seen, and we’ll talk about the script later on and being able to chart things is sort of where those impressions go, which is leading me up all kinds of crazy thoughts about what we might do, but that’s a sort of typical starting point, the full effectiveness of that.

That’s one thing I’m really keen to chat about maybe later on when we show some, some data of

FREDERICK VALLAEYS: what I think might

MIKE RHODES: be happening.

FREDERICK VALLAEYS: Yeah, exactly. So the, the other topic then, I mean, we’ve talked a little bit about campaigns. We’ve talked about how many asset groups to start with. I think it all goes back to the question of how do you run your business?

Let’s talk about cannibalization because that is significant, is a significant concern that especially if you run a regular shopping campaign and a PMAX with a shopping feed, what’s going to happen? Cory.

CORY LINDHOLM: Yeah, so Mike, you mentioned earlier the comment about control freaks, which has sort of been an internal joke, right?

Forever in the PPC space. I think. I have maybe a different view than a lot of people on this. Now, if you’re a major enterprise brand that is spending millions, I get the huge concern with cannibalization across your campaigns because your messaging needs to be hyper specific for that audience. It’s going to make a dramatic messaging is going to make a dramatic effect on quality scores in a very competitive auction space, et cetera, et cetera.

But. I will personally say there’s going to be two, two groups here. One is how do you avoid the cannibalization with shopping campaigns and performance max? So if you truly want to avoid the cannibalization generally speaking, you’re going to just have two options there. You have listing group exclusions in the performance max campaign or product group product filters in the standard shopping campaign.

But then you also, and I see this happen a lot, you have to remember if you’re using URL expansion setting for performance max with those in that campaign, even if you’re not targeting those products and the listing group side need to make sure to disable URL expansion or. Probably more preferably just make sure to add URL exclusions for those products.

You don’t want to show up in that performance max campaign. So from a standard shopping for performance max if, if you want to control it, that’s your best way of doing it. Search is a lot more tricky and goes into my main point about my opinion may being a little bit different than most with the cannibalization.

I, if I am making my clients more money, it’s more profitable. And. Google has so much more data than it’s ever going to give to me and things are doing really well. I could care less whether that search term or that product shows up in performance max Versus my search ad as long as the messaging is relevant as long as we are increasing profit of profits at scale there’s not much that I can honestly do to make sure that It shows up in performance max versus search other other than just not using performance max.

Our, our controls are a little bit more limited, but personally speaking I’m not hugely worried about it showing up in multiple campaigns, as long as I am looking at things on a media efficiency ratio outside of Google ads, and it’s adding up and it’s what we are doing and testing is, is working and I’m seeing those products show up and be generated and attributed for both search campaigns and performance max.

And we are growing. I’m happy. And the client’s happy. So that’s, that’s kind of my two cents on

FREDERICK VALLAEYS: it. That makes sense. If the client’s happy, then yeah. We’re all happy, right? Mike, what’s your take on that cannibalization?

MIKE RHODES: In the early days, I thought we were going to go back to the old days of, of, yeah, having 10, 000 exact match keywords in our campaigns in order to stop cannibalization and that is clearly a step backwards, clearly not what Google intend, you know, they unwound broad match or modified broad match because people weren’t using it the way it was intended to be used.

So clearly that wasn’t the answer. So. It requires a level of trust in the numbers, which I think many in the industry find difficult. Especially with the increase of modeling of conversions and PMAX being somewhat reportedly greedy in some instances. And and I know Google are never going to invent a conversion and obviously not in e com they can’t, but, oh, this conversion rhymes with this one over here.

We know what happened over here. We don’t know what happened over here. Yeah. We’re just going to say PMAX cause that one. That’s only going to increase the amount of modeling I think in the future with, with all of the issues that we have around data. So if you trust Google markets own homework, then it sounds like a great plan.

There’s I’m Eileen more that way. But I’ve also got plenty of mates in the industry with tin foil hats firmly planted on their heads. And so it’s interesting to listen to their views as well. I definitely lean more towards Cory, but I’m probably not at the extreme of yes. It’ll be fine. Yeah.

CORY LINDHOLM: I should add a quick disclaimer because this is really important.

Mike, my clients, and I’m always on a soapbox about this with every new client is if you can invest in an, in a third party unbiased attribution tool, that is a sophisticated machine learning modeling based attribution tool. It’s its own rabbit hole of a conversation, but I can’t tell you how important that is to be able as a business to get.

Again, an unbiased view. Every channel is going to want to take credit for that sale, and you’re going to ask that question at a certain level of scale of where should we put more budget towards. You can’t just trust Google. You can’t just trust Facebook. They’re, they’re all going to try to get, take credit for that sale.

So. The sooner you can get in on the game of a third party attribution tools, I think the better in most cases. So a lot of my clients and the people I work with are going to already have those tools in place. So also know that my trust for Google is that I don’t really trust them that much, and I have tools to help support my trust in the, in the data.

So.

FREDERICK VALLAEYS: Yeah, and I think the crypto world is teaching us a lot about trust and the fact that you need third party oversight if you don’t want something that looks like a bank to not be a bank and then lose all your money. And then also with crypto contracts some of those are based on the code and sometimes the code is written wrong.

And so a hacker is able to go in and legitimately take your money because, hey, it was the code that was doing what it was supposed to. Right. But the code is. Often wrong due to no malice on Google or Microsoft’s part. It’s just mistakes happen. So you want to monitor things. And that’s by the way, also where Optmyzr plays.

I mean, we have audits, we have automatic alerts to keep you informed when all of a sudden smart bidding goes kind of haywire because maybe it’s getting a bad signal from somewhere. I also have some thoughts on the cannibalization really quickly. So. Fundamentally, if you’re going to run a shopping campaign and a PMAX campaign, do you have to separate products at the query level basically, right?

So you can’t just say, Oh, I’m going to put these products in PMAX and these other products in regular shopping. If there’s any potential overlap on one query, kind of matching to either campaign, it is going to go to PMAX. So if you sell bicycles and you sell candles, great. You can put candles in one bicycles in the other, but if you sell Mountain bikes and road bikes.

That’s not going to work because some of these queries, Google’s going to say, eh, it’s close enough. We’ll just show the one from PMAX, right? So that, that’s one point on that. The other point that I’ve heard at SMX, which just happened was use what Cory was saying, which is the URL exclusion to basically get your brand out of the equation of PMAX.

And so what you do is you take your main domain, your main homepage, you make that the excluded URL, and you just advertise. The subpages of that domain and that tends to, to some degree, take brand out of the equation because brand is usually most aligned with your homepage. And then of course, there’s going to your rep and asking them to exclude brand keywords, which gives you back control, but you do have to ask for it.

CORY LINDHOLM: And I will say as a claim from Google, they wrote to me on this and I posted about it that. They did say with both dynamic search ads and exclusions and the URL rules for performance max for URL exclusions, they are synonymously buggy. So just know that if you set up a URL exclusion as a rule, It might not actually accomplish the goal.

I’ve had it many times and that’s why I contacted Google on this was to say, I set up these rules to not show, I don’t want any clicks for performance packs going to these URLs. I set up the rules, but clearly it’s still sending traffic there. And they said, Oh yeah, it’s, it’s, it’s kind of buggy. It doesn’t always work as intended.

So just make sure to use the full URL. I’m like. Do you realize that some, some brands have thousands of pages and products? There’s no way, like realistically, you’re going to get all the individual URLs. So just a quick reminder that you can use URL rules, but they may not

FREDERICK VALLAEYS: always work. Use scripts and tools, right?

If you have thousands upon thousands to exclude, that’s a really good tip, Cory. Final question on structure before we move on. Is there a benefit to running An asset group with no assets in it to force it to be a purely e com smart shopping campaign replacements

MIKE RHODES: while you can. I’ve heard from a couple of sources inside Google this week that engineers are all over this and they’re working on it and it’s going to go away.

At some point fairly soon.

FREDERICK VALLAEYS: And so to clarify, so the workaround that exists as of this recording is you can no longer do it on a new PMAX campaign, but if you create the campaign, you pause it and you unpause it, then you can go in. And I think remove the assets or make new asset groups that have no assets, but in the new flow they already no longer allow it.

Okay, so I think that’s enough said on that, Cory. I’m sure you agree with that whole point. Yeah, we can. Mike, let’s take a moment and look at this amazing script that you’ve written. So tell us what you did and why you think it’s important.

MIKE RHODES: Well, a couple of sort of use cases to begin with. So we’ve got this script that will put out a couple of charts for you.

It’ll show up in the top left, the percentage of your spend that’s going to shopping. So therefore we have product ads. The red line is the percentage of spend going to YouTube. And then the gray line is everything else. And as far as I know, unless you’ve got some genius ideas, there is no way to pull apart this everything else buckets.

I’ve just called that other. So that’s that DSA portion. It’s the remarketing. It’s the dynamic remarketing. I’ve looked at impressions. I’ve looked at click through rates of interactions. I can’t find a way to tease that

FREDERICK VALLAEYS: apart. If we’re up for people listening to the podcast. So Mike is showing some stuff on screen here.

So go to the show notes and we’ll put that in the resources so you can find this. But he’s showing us a dashboard right now.

MIKE RHODES: And more than happy. I like everybody watching the show. I can get a copy of this script and go install it in their accounts and get the same output. So up in the top, right.

Then we’ve just got a list of our. Pmax campaigns in a particular account. So this is a single account script that works on one account at a time. And then we’ve just got the little spark lines there. So you can see the actual spend over time. So on the left is the percentage on the right is the actual spend over time.

You can see a few, a few spikes in the YouTube, in the red lines there in the bottom left. We then have the percentage for each campaign. So up in the top left, that was for an, a, an individual campaign so that you can see the trends over time for an individual campaign bottom left. It’s all of your campaigns so that you can see really what we’re looking there is sort of generally how much blue.

So like roughly what’s going on in this. Account for all of these campaigns, roughly how much is going to shopping. Are there any campaigns that are showing a ton of YouTube? We’ve sort of seen these. Spikes in different accounts since we started doing this. And it was a bit of a pain before you could do this in the interface, but in order to turn on the columns for the number of views and the cost per view, you had to have a dummy YouTube campaign.

You had to have had a YouTube campaign in the account at some point. Otherwise. That data was just hidden from view. It was all available via the script, but if you weren’t using the script, you had to, it was a pain in the bum to set up a dummy campaign just to see that. And then you had to do the mental maths of multiplying number of views, average cost per view.

Okay. So that’s how much I’ve spent on YouTube. Now, how much have I spent over here? And then to find the, I’m out of spend on shopping. Again, you had to do a bit of mental maths, particularly if you had lots of listing groups, you pretty much had to go through there and add them all up. There’s no export button on the listing group view.

There’s no filter on there to say, show me just the actual listing groups with spend, not all of the subdivisions, all products, and then all of those little bits. And then down in the bottom right here, we’ve just got the, the, the spend, the actual spend per campaign. But the interesting bit, so one of the use cases with this was things are going well, ROAS is there, and we want to up the budgets.

One of our fears was if we crank budgets too hard, that the machine would just take a lot of that and go and waste it on YouTube or other, where the performance maybe wasn’t going to be there. So that was one of the, Use cases for this. I wanted to see if there was a spike in YouTube, but the other insights have come from, let me just jump forward.

I’d also asset groups. So it lists all of your asset groups with all of the data per asset group over the last 30 days. Lots of people have enjoyed that view because you just can’t get that table in the interface for whatever reason, but having these individual asset group charts. So what we’re looking at now is we’ve got.

It’s gone inside of one campaign, and there are three of these different tabs in the sheet that you’ll get when you get the script. One of these charts impressions over time. One of them charts cost over the time. And then one of them charts revenue over time. And this is what happened. So you’ll see there’s almost sort of three buckets.

We’ve got the total in black across the top. We’ve got the blue line. So that’s an asset group that’s getting lots of revenue. We’ve got the, what is it? Green, yellow, red lines in the middle. Getting a decent chunk. And then there’s all of the other asset groups sitting down at the bottom. But what I wanted to show you was this story of what happened when we split out.

If this was a campaign that had just a single asset group to start. And then we launched three new asset groups. As we were talking about earlier, that was a fairly simple, we didn’t want to make too many assumptions. So we had our custom match and remarketing, we had our custom segments, and then we had our in market.

And those three asset groups, you can’t even see that there are three lines. There’s a red and a green and something else sitting underneath that yellow line. So the yellow line was the single asset group. And then we launched these three. And what I found fascinating with this is that the number of impressions, which is what we’re looking at here, the number of impressions for the first few days, all of those new asset groups was almost identical.

And we’re talking tens of thousands of impressions here and almost exactly the same number. And then that incumbent, let’s call it the incumbent asset group, which is the yellow line. Again, the black line is the total. That incumbent is getting almost exactly the same Delta each day compared to all these new groups.

So it sort of feels to me that doesn’t feel like machine learning. I’m not a machine learning engineer. I don’t play one on the internet. You are. So I would love to get your input on this of that just doesn’t feel that feels like a deterministic rule that some engineer has created. All right. We’ve got this thing.

We know this performs. We’ve got a whole bunch of new things, new asset groups here. So let’s split the traffic pretty evenly, but sort of back to rotate evenly that old setting and. Let’s give all of these new asset groups the same, but let’s give that incumbent that we, that we know that we trust a little bit more.

Then what happened next? We got this big spike in one asset group, the red line, loads more impressions. Why? Well, if we look at one of the other tabs, the other chart, the day before the revenue in this asset group shot up more than the other groups. And so it’s shot up impressions. It went chasing after that revenue.

FREDERICK VALLAEYS: This is chasing the twirl basically.

MIKE RHODES: Yeah, Jason squirrel, but that revenue didn’t last that red line comes back down. And so the impressions come back down, but, and this was the bit that really blew my brain. Look at where that red line comes back to it comes back to exactly where the yellow line is.

That was our incumbent, you know, the, the trusted asset group. So it feels like there’s almost two buckets going on here, or actually maybe three. We’ve got trusted asset group, we’ve got new unproven asset group, and then we’ve got squirrel asset group. Let’s go chase this thing because we think it’s really, Oh no, it didn’t work out.

But because it sort of has worked a bit in the past, it’s almost like it fell back into the proven market, which gets again, that same sort of Delta, that same sort of a little bit more impressions than the asset groups that we don’t just yet trust. And then the green one starts to take off because a bit of revenue happened there.

So it starts chasing after that squirrel and those impressions jump up. And then if we play this out over time, over the, over time, 30 days, you can sort of see everything starting to converge the blue line shot up in the middle there. Cause that got a bunch of revenue. So that shot up and then fell back down.

And we’ve seen this sort of patent over and over again in a ton of different accounts now. So I just was really curious. One, are you guys seeing the same things? Why do you think this might be? It’s meant to all be machine learning and a lot of that, but that just does not feel like a machine learning kind of a thing.

Revenue is a lot messier. You can see that blue line sort of, yeah, maybe it’s above the other lines on some days, but. It all sort of converges. And this goes back to the question you were asking before about which asset groups, how many asset groups, as I’ve seen more and more of these charts. I’ve started to wonder just how important the asset group and the audience signal, and certainly the creative actually is if I show you.

So this is not all of the campaigns, but this is a good selection of campaigns. I haven’t cherry picked this. This is just a random chunk of campaigns. So it’s a second script that I will get up on GitHub as soon as possible before this comes out so that everybody has this as well. So this is a script that you can run at the MCC level and now see all of the P max campaigns across your whole MCC in one place.

I know this is going to do some people’s heads in, but essentially we’re looking at the same. But just for everything. So the amount of blue there is the shopping and we’re averaging about 86 percent cost goes to shopping. And there’s the odd campaign there that lights up with red. So that’s spending a bunch on YouTube, but you can see the vast majority of campaigns are hardly any YouTube if any at all.

And then the gray is other, so if 80.

FREDERICK VALLAEYS: Go on. No, no,

MIKE RHODES: go ahead. Just to wrap it up. If 86 percent is shopping, then how important are these other assets? I’m actually setting up with our account managers, a bit of a debate internally. Cause we’ve got a few people that are really like creative. So important.

And there’s a couple on my side. Well, I know it will be, but it doesn’t seem to be yet in the early days of a max, we were saying to clients, look, I know it’s a pain, but please, we really need a YouTube video from you. Otherwise the machine’s going to create something that looks like my daughter’s PowerPoint from 2007.

And we really don’t want to show that video. So we really need a video from you. But then we look at this and go, well, how important is YouTube? How important are all of those texts and image assets that we’re all adding to our PMAX and being forced to add to our PMAX campaigns now, if 86 percent of the time our shopping ad is being shown.

So if creative isn’t that important, if impressions are being sort of randomly assigned. To all of these asset groups and the impressions per asset group is converging to roughly the same. How important are those audience signals? It seems to me that the product is potentially the most important signal.

He you know, the product image we know is important. We know the title is really important. Your price relative to your competitors, hugely important. Andreas from creative prove that years ago. So I’ve got more questions than answers at this point about where we go next with this.

FREDERICK VALLAEYS: So one important question is what is the value of putting in these assets?

And it’s an important question because like you said, it takes a tremendous amount of time compared to just splitting out your product feed and maybe adding different audiences to asset groups. So is that investment of time into getting the videos, the images, is that worth it? Yeah. Or do you just flat out take one set of standard images and videos and put that across everything?

So it sounds like that may be good enough I think the other important question that you then raised is is this really machine learning? Because oftentimes when we think about machine learning from google, it’s about making predictions that help us Whereas here it’s not about predicting that there will be a squirrel It’s about seeing the squirrel and chasing it until it’s gone in the tree And then it’s waiting

MIKE RHODES: for squirrels to appear.

Yeah,

FREDERICK VALLAEYS: I love this whole new analogy of squirrels here. One question I do have though, Mike once it sees a squirrel, which is kind of, so that squirrel is basically one audience. That’s all of a sudden performing better. Do you see any impact on that same audience in different campaigns?

MIKE RHODES: That’s a good question.

I haven’t looked closely enough at that. I’m Leaning towards no, but I haven’t looked closely enough to know. Right.

FREDERICK VALLAEYS: So folks, everyone’s going to get this script. So please download it. And we’d love to have a discussion around whatever else is seeing.

CORY LINDHOLM: And someone please create a site where we can all just like provide hashed data so we can get insights across thousands of accounts.

That would be so awesome. Let’s share. Sharing is caring.

FREDERICK VALLAEYS: Let’s see if Optmyzr can help with that one. We can take my script and we’re already working on building some of the functionality he showed us. He’s been very generous letting us see the script and we want to make it easier for everyone to use it.

MIKE RHODES: Everything I learned about scripts. I learned from you. So thank you.

FREDERICK VALLAEYS: Thank you All right, so super interesting. So again download the script Let us know how it goes, but really interesting insights there. Mike Cory, what about let’s talk maybe briefly about audiences. I think Mike was saying you have a great hack of some sort for doing audience related.

Tell us about that.

CORY LINDHOLM: Yeah. So this is a fun subject because I think you can get a lot of insights for, for the business as a whole. It’s kind of like PPC professionals for years have been trying to preach from the mountaintops of use your search term data for more than Google ads. Like there is, this is valuable marketplace information in a lot of, in a lot of cases, if the targeting is fairly relevant and your conversion tracking is correct and all these other things.

But when you go through and you are looking at setting up audience signals, for example, in performance max the hack I think Mike’s referring to is it, Mike, correct me if I’m wrong, but are you talking about like the in market other. No,

MIKE RHODES: all of the buying words, your big mix and match.

CORY LINDHOLM: That one. Yeah.

So I have this massive spreadsheet that takes forever to load, even in Excel, because I’ve just continued to build on it. So basically what I do is there is a, an ability when you’re creating an audience signal for performance, max asset group, where you can choose custom segment, and then you can put in.

search for keywords. Basically, they’re display search keywords, technically speaking, and you can load that up in the Google ads interface with as many keywords is as you want. And because audience signals are just seed audiences, we know that they’re going to expand if it can’t find success with that initial seed audience, or even if it does, it may still expand from that.

My thought process was, well, what if I just Cherry pick every commercial intent variation of product related or product category related search keywords. So that’s what I ended up doing was just creating this massive sort of, you know, if this then that concatenated spreadsheet where it’s, you know, you basically enter the product, the product category.

And then it’s going to pair that with a long list of variations, like buy, you know, buy gold leggings or, you know purchase gold leggings or gold legging sales or anything that just shows some sort of commercial intent. And then I load that list up and load it in there as a seed audience for Google ads.

And a lot of cases it’s like the system like wants to just crash on me because it’s. Thousands. In a lot of cases, it averages like 6, 000 different keywords, but they’re all highly cherry picked commercial intent. Sometimes I’m taking it from past converting search keywords. If those products are still sold for that business and yeah, just loading them up in there and saying, all right, Google, here’s the product.

Here is literally of all the searches that are out there. This is like the best of the best start here and then expand. Note that there’s two ways to do it. There’s two options. There’s one that says purchase intent, and then the other one is basically people who have recently searched XYZ. I like to personally load up both either in a single audience signal or separately.

But yeah, I think that’s a, that’s a really fun approach. It’s definitely an audience that I see consistently work really well, which kind of makes logical sense, right? It’s really just cherry picking on people who are in the market and they’re ready to buy, as opposed to leaning on Google to decide who’s in market for something, which I also use, and I really love in market audiences, especially the in market, other audiences, but this custom segment approach works very, very well.

MIKE RHODES: Can I jump in please? What I would, what I would love to see. Is the difference between adding those 6, 000 all in one asset group and adding it to what’s there versus six lots of a thousand and adding those as six new asset groups, because based on those charts that I was just showing, it seems to me that if you add six new asset groups, a chunk of impressions get taken from the incumbent and shared evenly.

So in other words, it’s going to end up those six together are going to get a lot more impressions to them than just one new one. In market other, we took our in market asset group, split out the top, you know, based on your insights, split out the top in market, put that in a separate asset group. And instantly that new asset group gets the same number of impressions as those other asset groups.

The those, those lines on the chart converge really quickly. They’re all getting the same impression. So it’s a kind of a way to go, Oh, this one is getting this 21. 7 X. It gives you that index of 21. 7. Times more likely to buy than the average bear and instantly throws more impressions at that asset group.

So to, when that works really well, I wonder what would happen of lots of smaller asset groups rather than one big one.

FREDERICK VALLAEYS: I love it. Put Cory’s brain and Mike’s brain together, and you get this hybrid solution of overloading with custom intensive notes and splitting it six different ways to get even more volume on it.

So yeah, no, this is great stuff, guys. This is hopefully why people are watching to get some of these insights from you. All right. So now that we got these PMAX campaigns running with all the best practices that you’ve laid out there what happens in terms of all the things, these things, if maybe performance isn’t going the way you expect it, Cory, I think you have a checklist, right?

CORY LINDHOLM: Yes, I do. Let me, then I’ll load it up here. So as my fan base will know, I am really big into like lists and I go super deep on everything. I have simplified this diagnosis list. So basically. People love checklists, as do I, especially big agencies that have a lot of work to do every day. And when you often go in your auditing account or you’re trying to figure out what is going on with your Performance Max campaign, it can be a headache to remember all of the different checklists.

For what might be going on. So I’ve created this I would say a pretty massive list. I think it’s over 40 different items that you can cover. Of course, not every business is going to need to go through every single one of these. But I’ll just cover the first couple of slides. And if you want the full 10 slides and all 40, some odd check, the full checklist we’ll probably provide that as like a downloadable or something for you guys to access later.

So first one to go over. It’s just going to be sort of the high level things that we first need to look at, right? So what is the estimated conversion reporting delay from the first interaction with an ad versus so someone first interacts with the ad and then when they actually convert, we need to first have a solid understanding of how much time is going to take place between those two touch points, as well as just the overall estimated conversion reporting delay.

There’s going to be Not real time reporting that’s been at the bottom of the Google ads interface for since it all was the green ugly green Google ads interface that’s always been around. It’s not real time reporting that needs to be considered before you freak out. So that’s just the 1st thing to say it may be that people you warmed up with the strategy change are just going through the funnel and you might need to give it some more time.

So 1st, just make sure you’re aware and you’re familiar. with the conversion reporting delays and any kind of average days of conversion from first ad interaction. You can look at this in an account wide status in the attribution section of your section of your account, or you can look at a campaign specific by adding a segment to your campaign tables.

Obviously conversion tracking, I put that really high up on the list, but Because, well, more often than not, if things tanked from my experience, the first thing you should be looking at is conversion tracking, as opposed to going down rabbit holes of trying to diagnose individual elements. It may just be that something broke with conversion tracking, or maybe the client or yourself were messing with something in the checkout flow.

And then that. Unintentionally really screwed up something with the conversion tracking pixel. So make sure you check that before you go down this list. Also make sure you’re familiar with what normal fluctuation looks like for the account. You may just have launched PMAX and it, you know, it’s it, this isn’t going to be applicable for you, but if you’ve had performance max for a few months, Familiarize yourself with what normal fluctuation looks like for clicks, impression impressions, conversions, conversion rates.

Maybe every five days, isn’t really going to be a good time period to look at, look at different time periods, get familiar with what a weekly up and down ebb and flow looks like in terms of a sort of normal fluctuation. And then go from there again, the whole idea, you may not have to freak out. So figure that out first and then kind of go from there.

After that just recent changes. This seems really obvious, but it’s amazing how many people go down rabbit holes of analysis before just checking on, well, what was the last thing I did to this Performance Max campaign? Biggest things are usually going to be budget changes, bid strategy changes asset group changes, listing group changes.

There’s other settings that may have major impacts, but these are the sort of more common ones are going to be changing. On a more regular basis. Second slide. It’s going to be any kind of Google Merchant Center things, right? So disapprovals warnings that may be affecting the products that are being that should be being served in that performance Max campaign.

Any account issues or feed issues? You’ll find that in the diagnostic diagnostic section. Of your Google Merchant Center, if that’s what you use any changes to the site, this goes back to conversion tracking, but it could just be maybe another agency or your agency is doing CRO, and we’re testing different landing pages, or we’ve recently updated some of the keywords on a product page for a bestseller, those things all need to be taken into account, and unfortunately, Those things aren’t always communicated across staff members or even with a team inside of a business.

So just check to see, has there been any major changes to the site? If it’s a large site, check and start on the bestsellers and see if there’s any major changes there. In stock products, is there any inventory changes? Again, particularly bestsellers. Any changes to pricing or promotions that’s happened pre or post any kind of major performance fluctuations.

This next one gets missed a lot, but is there any extremely negative reviews? If you are a smaller, lesser known brand and you’ve only, maybe you’re selling just on Amazon and your website. Go ahead and check Amazon. A lot of PPC professionals forget, I mean, a lot of consumers are going on Amazon before making their purchase decision.

They might buy, they might not be buying from your site because they saw a very recent, very maybe inaccurate negative review. Don’t forget that. That can really affect what you’re going to be seeing in terms of your results for Performance Max. Obviously, if you have a lot of products, it might not change things dramatically, but definitely important to keep that in mind.

And then the last one I’ll share today is gonna be Google Search Console for failing URLs. If you’re not doing a product feed only type of campaign and you you are using URL expansion, or even if you’re not, it’s important to make sure to keep tabs on Google Search Console. This gets missed quite often.

Check the core website vitals section of Google Search Console to see if there’s any quote unquote failing URLs. This is gonna affect the performance of Performance Max, in particular, the The dynamic search ads, a portion of performance max. So just a few things there’s a lot more to go into again.

We’ll be sharing that really soon here, but

FREDERICK VALLAEYS: yeah. Thank you, Cory. I mean, I, you’ve been very generous in sharing that very extensive checklist. It’s a no freak out P max checklist and avoid, let’s not call them rabbit holes, let’s call them squirrels, squirrels, right? Okay, so this has been an amazing episode.

I hope everyone’s enjoyed it. And we’ve seen some of the ways that Cory looks at analyzing these accounts and the way that Mike looks at sort of going into a fairly flat structure and getting a bit more segmented and figuring out what’s going on with the machine learning. So we’re going to continue this discussion on performance Mac.

So please contribute in the comments. Let us know what you’ve seen. We’re also putting the script URL and a link to the checklist in the show notes, or you can look for those at ppctownhall. com. With that, if you’d like this episode, make sure to check out Cory at ads by Cory. And follow Mike at agency savvy.

com. au. And if you like the show, subscribe to it on YouTube, we’ll be back again with another episode soon. We also have Optmyzr. That’s the software where I’m co founder and CEO. We have a two week free trial. Many of the things we’ve talked about today, we hop make them a little bit more doable, do a more easily and less time.

Thanks to automation and software. So check it out. Thanks for watching and we’ll see you for the next episode. Thanks

guys.