← DTC Podcast
DTC Podcast
Ep 534: Google’s 90% “New Customer” Illusion: How To See What Your Ads Are Really Doing | AKNF
00:23:47
Loading summary
Transcript48 lines
- [00:00]
A
The native build in Google Ads we've found to be fraught with errors and inaccuracy. Google generally, in our experience, has a 90% rate on just about every customer being identified as net new, which we just know to be false. You should have different goals associated with repeat purchasers when compared to net new acquisition and be treating acquisition of net new customers differently from returning a common occurrence in lead gen is okay. On the advertising platform you're driving a ton of leads. Great. We've got super low CPLs. It's exactly what we want and you look on the back end and they're all crap quality. So over time with our lead gen clients we found. Okay, once we have enough data on lower funnel quality touch points on these leads, let's feed that information back into the platform.
- [00:53]
B
It's all killer, no filler. I'm Eric here with Pilothouse Google Lead. Dougie, welcome. I know we have an exciting topic today around how we need to be feeding the right data to Google to optimize towards the right goal. What have you brought for us today, Douglas?
- [01:10]
A
Yeah, so we're talking a little bit about something that's natively built into Google Ads and focusing on new customer acquisition. But the native build in Google Ads we've found to be fraught with errors and inaccuracy. So we've been working through it in a slightly unique and more nuanced way, which I'm excited to dig into.
- [01:33]
B
And it's the most pivotal thing in modern digital marketing which is really being able to understand your incrementality, understanding who you're bringing in, who are new buyers and who you are maybe retargeting and bringing back in. And those users are quite often from external platforms a little less valuable because you might have got them anyway with email, you might be counting them as email revenue or some other traffic source.
- [01:58]
A
Right? Yeah. And at the very least they should be. You should have different goals associated with repeat purchasers when compared to net new acquisition and be treating acquisition of net new customers differently from returning. And Google's tried to do this. You can upload customer match lists and try to differentiate new versus returning customers, but it's all still based on cookie data. Right. And so using a platform, a server side tool like Elevar allows us to get a bit more accuracy on okay, who are truly the net new versus the returning. And especially for brands that have a high level of returning traffic. So CPG is one that jumps out apparel even at times, you definitely want to value a new customer differently. Than a returning customer. And Google generally, in our experience, has about a 90% rate on just about every customer being identified as net new, which we just know to be false. We've applied this to a multitude of brands and found that it's pretty much even across the board, even when we're pulling direct CMLs from Shopify or from Klaviyo, that we're very confident in the segmentation of Google's still aligning these users as being net new when for a fact we know that they are not.
- [03:16]
B
So how are they doing this? Is this because cookie data is only lasting a certain amount of time? How long are cookies lasting at this point?
- [03:22]
A
Yeah, so it depends on the browser you're using, but there are definitely limited time frames now. Right. And especially if you're using an imported conversion goal from GA4, for example, you have the same issue where GA4, like Google have a difficult time if they're not being fed server side information. They have a difficult time attributing net new versus returning. And so yeah, it's really integral for a funnel marketing strategy applied to Google that we have that interpretation of net new versus returning.
- [03:57]
B
I can imagine almost every marketer would look at that if they saw a 90% new customer rate from a platform like Google. They probably would be smart to doubt that. That seems extremely like you guys just knew it was wrong, you backed it up with data.
- [04:11]
A
Yeah. And I think most people would align to that, that kind of thought process of hey, this seems off. I think you've got some people who are maybe false actors in the space where they'll take that information, they'll go to a client or go to their boss and say, hey, look at all the net new. I'm driving through Google. Right. Whereas we're trying to really unpack, okay, is this truly incremental? Is this truly what we think, what we're trying to do? Is it truly what we're doing? And it's always kind of trying to peel back the hood and understand what we're actually targeting. And it actually initially came from a bit more of a lead gen thought process. So a common occurrence in lead gen is okay, on the advertising platform you're driving a ton of leads. Great. We've got super low CPLs, it's exactly what we want. And you look on the back end and they're all crap quality. Right. So over time with our lead gen clients we found, okay, once we have enough data on lower funnel quality touch points on these leads, let's feed that information back into the platform. So that's operating off of a qualified user as opposed to just any lead that came in the door. And you are able to go further and further down provided you have enough data to feed back into the platform. So we took that kind of thought process and applied it to dtc and we think generally what we're trying to do here on Google is acquire net new customers. Right. That's our goal. If we're targeting a ton of brand and we're a CPG client and we're. We're naturally getting a lot of repeat buyers, we don't want to be spending 2, 3, 4, 5, whatever amount per click on these users that are naturally coming back and purchasing. We want to make sure our efforts are focused on. NET new. So optimizing to a. Net new conversion goal as opposed to segmenting at the audience level, we're actually picking a conversion goal that is aligned to. Net new and that allows the algorithm to focus in on that as the performance target as opposed to targeting any purchase. And then we're using a poor segmentation of that new returning. Exactly.
- [06:16]
B
That just is okay. So you're seeing. So it's one of those, those changes that has ongoing incremental benefits because you're actually putting good data in versus bad data, which we've talked, I know in the past about. You can get locked in these. You know, your performance can dwindle in a sort of downward spiral if it continues to cycle in bad data.
- [06:35]
A
Yeah, 100%. It's like tying this to AI. Right. If you give it a bad prompt, you're probably not going to. Bad inputs lead to bad outputs. Right. Bad prompts probably don't lead to the optimal result you're looking for when using AI or what have you. And all this algorithmic bidding is automated. It is AI. Right. So we have to feed in the right input so that we get the right output. The business aligned output of more net new acquisition. It comes down to that conversion level going deeper as opposed to that upper level audience that we're trying to segment.
- [07:11]
B
We're being told that these are converters versus our data, which is actually converters.
- [07:15]
A
Exactly.
- [07:16]
B
And so how are we doing it?
- [07:18]
A
Yeah, so Elevar has a functionality. So server side tracking first and foremost, generally it's more accurate. We have the server being the middleman and first party cookies as opposed to third party. So we just generally get more accuracy out of server side tracking. And then the server is able to or the LVAR is able to determine a new versus returning customer more accurately. And we are pulling that conversion goal from Elevar into our ad campaigns, provided they have enough conversion data behind it. You don't want to do this if you've got one conversion a month or what have you. You're just not going to get a high enough volume of data to feedback into the algorithm. But for most CPG brands that are of a relatively decent size, you'll get enough data back where you can focus in on. Net new exclusively because you're getting the conversion volume to provide the Google algorithm with enough learnings behind who these customers are.
- [08:16]
B
Super cool. Are there other. Is Elevar the only platform that does this or are there other platforms that that do this? Is with Triple Whale. Do it.
- [08:23]
A
I don't believe Triple Whale has a native like they're more MTA than kind of a server side tracking solution. There are definitely others out there, but Elevar's generally the one. We've preferred to partner with Elevar.
- [08:40]
B
So if you're listening to this and you do this and you work with Elevar, tell them that DTC sent you and that they should sponsor the newsletter or come on the podcast.
- [08:49]
A
Yeah, exactly.
- [08:49]
B
Because I feel like we talk about Elevar quite often on the Google side. It must be absolutely indispensable. What kind of results, what kind of increase in the results did we see after setting up this more accurate server side conversion tracking?
- [09:04]
A
Yeah, we're definitely starting to see it's a slow pivot. Right. Because we're using 30 day look back windows and you're shocking the system to change from all conversions that hey, maybe 50% of these conversions weren't net new and you're essentially cutting some of that volume. So there's that initial shock to the system to overcome. But generally we're starting to see an uptick in new customer purchases overall coming through Google, as well as a decrease in customer acquisition costs. So it's still something we're testing and trying to find the right fit for the right brands. This shouldn't necessarily be applied across the board for every single client, but the general philosophy of aligning your conversion action to the ultimate goal is what I'm trying to get across here. And most of the time for DTC brands, we're trying to get Net and new acquisition through Google and leaving returning customer purchases for other platforms. You mentioned email being the primary one there, or at least having separate goals aligned to those audience segments as well. Right. It might be valuable to get a returning purchase, but we have a different Level of value assigned to that compared to a net new.
- [10:15]
B
Give me an idea of how you would score that in a campaign or.
- [10:20]
A
In an overall account. Yeah, so I think it depends on your repeat rate, your ltv. Right. So you have to take those things into account. One thing that I've been discussing recently is as well, to kind of get really in the weeds is like product entry point for brands that have apparel brands, for example, brands that have quite a few different SKUs we're trying to look at, okay, how much is a customer worth on average? So ltv. Right. How much is it costing us to acquire that new customer? And then on the repeat purchases, what are they buying? How much are they spending? What does the second purchase look like typically compared to the first? These are all things that we have to unpack at the business level to make sure our marketing efforts are reflecting the overall business goals. And I think that oftentimes we just look at an account, we look at, okay, what's our ROAS look like? If we just look at ROAS that has 50% returning customers, it's a bunch of brand awareness, like all stuff that you and I have talked about in the past on these podcasts. But we're probably not being particularly incremental to growing the business. So that's why we really want to unpack and make sure that at the high level, our strategy is aligned to what we're actually seeing in platform. And we're not just kind of taking that at face value. We're looking under the hood to say, okay, what are the conversions looking like? What do the search terms look like? What are the placements looking like? Is this aligned to our higher level strategy? Getting kind of zoomed out there. But all that to say there are quite a few inputs you want to look into that determine how much you should value a returning customer versus a net new. And for each brand, that's going to be slightly different based on those inputs.
- [12:01]
B
Are there any other aspects of the conversion journey that could be improved if they were used, you know, this enhanced. If they used an enhanced data set versus maybe what the platform says it can do? Or is this sort of a. Is this just a new one off instance?
- [12:18]
A
Yeah. Do you mind unpacking that a little bit? A bit more for. I don't know.
- [12:21]
B
I'm just trying to. I'm just trying to think of other areas where a system like Google is telling you something and you're building assumptions or campaigns off that, where if you were to, you know, place an elevar Instance and it would, it was able. I'm just trying to think if there were other aspects of the conversion journey that would benefit by using enhanced data versus on platform data.
- [12:45]
A
Yeah, I guess I would say that again, like the more quality data you provide the platform, the more quality results you're going to get. Again, back into good inputs lead to good outputs, bad inputs lead to bad outputs. So the parallel all maybe draw, which isn't necessarily about taking a new data source into account. But you say, okay, if we provide more quality data or we're analyzing data that's of quality, we'll be able to drive more performance. The most common example, when I look at accounts, when I'm auditing accounts, I see, okay, we're trying to target a lot of specific terms on the keyword side of things. But keywords are just signals nowadays. They're not, they're not especially broad match. Right. And then you look at the search terms, they're all over the place. Nothing remotely close to what you were actually trying to target. So again, coming back to looking under the hood, making sure we're aligning to good quality data. So we might have the intention of, hey, we're doing the right work and targeting the right keywords and getting net new. But if we don't pull that back and trying to unpack is that what's happening in actuality, then we're having a, we're giving a bad input in terms of a irrelevant targeting sequence leading to a bad output which results in either no conversions or poor quality traffic or what have you.
- [14:08]
B
How do you account for LTV in. Because you want to train the algorithms on finding high LTV customers. So are you also. Are we. I remember I used to do this back in the day on Meta where we'd have people who are multiple repeat purchasers and we'd be using them for lookalike, building them into lookalike audiences. That goes on with Google as well.
- [14:27]
A
Yeah. So when it comes to lookalikes, the only about a year, year and a half ago, Google walked back their ability.
- [14:35]
B
Oh, I remember that.
- [14:36]
A
Yeah. So now you can only use lookalike audiences on demand gen, which encompasses YouTube. So look like audiences can be deployed there. But in terms of how we're using LTV and trying to provide like high quality customer signals back to Google, it's not as prevalent as it used to be, partially because they've walked that part of things back. It's still useful to have a klaviyo integration, make sure that you're uploading those CMLs to continue lookalikes for the whales of your consumer base or what have you. But generally it comes down to making that LTV analysis and aligning that to customer acquisition costs which yeah, I candidly don't think enough brands are doing.
- [15:23]
B
That's great. Super interesting. Everyone in the audience needs to take a look at this and audit at the very least. Like what's the simplest way to go about auditing whether this is happening? Say look under the hood. Give me the step by step of how you actually audit this for brands.
- [15:38]
A
That want to know. Yeah, so for one it's whatever CRM you're using or Shopify I don't really have high confidence in as mentioned Google Ads and analytics ability to distinguish new versus returning for reasons we got into around first versus third party cookies, what have you. Really the easiest way I'm trying to find a way to present this without just straight up pitching Elevar but get Elevar try to get server side tracking set up. It isn't that expensive to get sorted. There are trusted partners out there like Elevar that you can work with and try and get a an accurate picture of your data set. When it comes to new versus returning, don't rely on just the platform itself. And then once you've got either server side tracking or alternative solutions to diagnosing your new versus returning from paid ad platforms, then you can make determinations on whether you should make the pivot to optimizing towards net new or continuing on with the current conversion goals that you're operating with.
- [16:42]
B
Do you have any data on how? Because I remember server server is such an old technology, right? Like when I was running web display ads In 2009 or so we were on server server tracking. Do you have. Because but there's and I know that the cookie side tracking has sort of been deprecated in a number of ways. Do you have any data or anecdotes on why everyone really needs to be on server to server? Is it like a good chunk percentage better? Even though we're talking about how to improve server side even further with Elevar, but just already that switch from cookie to server side tracking, is it a big improvement in your data fidelity just with that?
- [17:18]
A
It is. And scale is a factor too, right? If you're a small brand driving like a few conversions a month like then it doesn't really make sense. Right? There's the concept of scale having a significant impact on your data set, right? Like a 5% to 10% whatever increase on three sales a month doesn't really matter for what you're feeding back into Google. That's not even, that's 0.1 of a sale or what have you. Whereas when you get up to a certain level of volume then it definitely makes a significant impact. And I'm throwing out 5 to 10%. I think in certain cases depending on your cookie consent banner implementation on site, especially if you're a brand operating in California, like the law there and in the UK around privacy is only going to continue to like those are going to be the leading GEOs in law around privacy for advertisers. So I think if you're in those two key areas, you're probably already seeing maybe lawsuits come into play or you're being dinged by certain policy violations from the governing parties in those GEOs. And I think that watching those GEOs specifically over time, like it's all like the cookie list future we've been talking about for forever. Right. And it is eventually going to show up. We're already seeing ad platforms use more and more modeled conversions compared to actual cookie data. So yeah, I think that if you're of a high enough scale and if you have enough data flowing in, you should absolutely be adopting server side tracking because the benefit, even if it is 5% is pretty significant at scale. And my argument would be that I don't want to just throw out okay it's 25%, it's 30% because I don't have the number offhand. But in my experience with these brands that we've integrated Elevar on, it has been significant. The trackable conversion, growth in platform and.
- [19:22]
B
It'S again it's about rolling Thunder as well. Right. It's about improving the model over time which the more fidelity your data has, the better it's going to be long term. Super cool. Look forward to this. What are your vibes this year on Q4 on the Google side of things?
- [19:40]
A
Excited, always excited for Q4. Yeah, I think we haven't had any big roadblocks come our way so we're pretty excited to have another dialed Q4. I think there's some small new product rollouts that Google's been working with that I think probably won't have completely been fleshed out by Q4 thinking of things like AI Max and what have you. But yeah, I think we're feeling pretty ready to grab onto the opportunity for our brands.
- [20:13]
B
Probably still room for a few brands out there who want to get on join the pilot house team. Oh I also, but I Wanted to ask you what, what did you think of that headline, Google's search revenue being slashed by 40% by their AI summaries, which are very useful.
- [20:29]
A
Yeah. So I actually had an internal discussion about this and if you think about the search process, if you're searching something that doesn't have a binary answer, you're more commonly going to be served these AI overviews. Right. They're meant to answer more upper funnel type queries. And I was having this discussion with a colleague and we were talking about how this more significantly impacts people outside of the DTC space in the content ecosystem. Exactly. The content aggregators, those are the people being hurt the most because they're just pulling content and advertising that content and charging revenue on that content. Whereas for DTC brands, like if someone searches yellow Nike shoe size 11, there's not going to be an AI overview talking about the benefits of a yellow Nike shoe that's size 11. It's going to still serve you your shopping listings. And long term, I think there's going to continue to be a short term rocky road as Google tries to figure out, okay, how am I going to integrate with AI? I think it's going to do it slowly but thoughtfully. I have another podcast where we can talk about the difference in how Gemini is using, I guess their structured information to influence discoverability versus other LLMs.
- [21:58]
B
But yeah, let's do that. It'd be fun actually to. I know that ChatGPT, we're getting leads for the agency like every week. We're getting a couple leads now from Chat chatgpt, which is.
- [22:10]
A
Well.
- [22:10]
B
And that must be happening for businesses all over the place, optimizing for that. In some ways I'm sure it's akin to SEO in some way. So let's definitely put a pin in that and come back to it.
- [22:20]
A
Yeah, I've got a cool tidbit for you there from. There's lots of good content out there. But Fred from Optimizor used to work at Google and now launched his own. Or now multiple years ago, now launched his own PPC management software. Has some. Yeah. Unique takes on that specifically.
- [22:40]
B
Oh, I love rogue ex Googlers.
- [22:42]
A
Yeah. And he still obviously works in close concourse with Google. But yeah. Anyhow, I think we've strayed away from your initial ask around.
- [22:55]
B
No, this is great. We always like to end with a ramble. We don't have a Jeff Bezos to point to like we do on the Amazon team, so. And I think. And we'll just save the hockey conversations for. For slack because this has been great. Thanks for coming on the all killer no filler D2C podcast today, Doug.
- [23:10]
A
Yeah, of course, anytime. And look forward to our chat on maybe a bit more AI focus on the next one.
- [23:22]
B
Thanks for listening to today's episode. If you're not getting the DTC newsletter, you can subscribe for free at directtoconsumer Co. And if you want to learn more about Pilothouse's All Killer no Filler services, take off to Pilothouse Co. I'm Eric Dick and this has been the D2C pod podcast. We'll see you next time.