Google is bidding on the WRONG products: The ‘Clicked vs. Bought’ dilemma explained

Episode Transcript

Mike 00:00:57
Welcome to another episode of Growing E-Commerce. I am one of your hosts, Mike Ryan. And here’s Chris, one of our other hosts. Actually, there’s just the two of us. That’s the gameplay. Chris, ask me what I did this weekend.

Chris 00:00:53
What did you do this weekend?

Mike 00:00:57
I saved this. This is just some rage bait. On Sunday morning, from four until seven, I mixed and poured over 1,500 kilos of concrete.

Chris 00:00:53
What? Why? By the way, how is your construction site doing in general?

Mike 00:00:57
I had to pour a slab at the endless construction site. Well, it’s endless. If this were taxpayer-funded, we’d be all good.

Chris 00:01:03
But mate, I told you before. Hats off to you, because you are doing a fantastic job at smec, as people know. And you do this in parallel to building a wonderful house. It’s crazy. I would never build a house on my own because it would lead to massive friction internally and externally with my wife. Shout out again to you, but we wouldn’t survive it. It’s the ultimate test. What is easier: to run a highly complex PMax campaign or building a house?

Mike 00:01:48
Building a house. Either way, you are covered in dirt. There’s a lot of unbounds. So what’s up for today, Mike? Let’s get into it.

Following through on some open threads from the last couple of episodes, we were talking about which themes are hot with our clients. We discussed high CPCs and I want to connect the dots to the intersection of two things. One thing we hear a lot from clients is concern about winning new customers and having a high share of new customers. They look at their campaigns and tactics, asking: what is the share of new customers on PMax? Is it higher on Shopping?

PMax has a higher share of returning traffic, and this could be viewed as less incremental. To me, that’s logical because Performance Max has remarketing baked in. Both of these campaign types are actually agnostic to who they are targeting, but I think that is not a bad thing. Most advertisers desperately need this returning traffic and this remarketing, even if it costs them money. That’s my short statement.

Chris 00:03:18
I’m in the Mike camp. We have had discussions about that a couple of times already. My first question would be: is PMax even the right campaign type to really create new customers, or are there better campaigns out there, like in the Google ecosystem, which are better designed by default to create this new demand for you?

Mike 00:03:24
I’ll let you cook.

Mike 00:04:17
The simple answer is no, it’s not the right type because it’s a bottom-funnel campaign. These are all people who are somehow down-funnel, which we can tell from their activity and the searches they do. Google knows things about them and if they’re in-market. At that point, it’s still binary; they could be existing or new customers. If you really want to focus on building new customers, that’s branding activity. That is your mid- and upper-funnel activity. The proper place to be concerned about that is with Demand Gen campaigns, display campaigns, and across all your other channels and platforms besides Google.

Chris 00:05:20
I love that we are so consistent because that’s what we have been talking about for the last couple of episodes—that Demand Gen is actually the type of campaign which is potentially creating this new demand. I want to come up with a real-life example. I’m not a super loyalist; I know my brands, but most of the time when I’m buying online, I start my research on Google. I’m a good Google client.

Say I bought something at Zalando a month ago. I’m happy with the experience. But for my next purchase, I start my research on Google looking for an Adidas sneaker. For Zalando and Google, I’m probably an existing client. If Zalando is now focusing only on new clients, the probability that a Zalando ad gets shown to me as an existing client is close to zero. I would probably go somewhere else that has the sneaker I’m looking for. Whenever I start my research and don’t go directly to Zalando, I’m potentially a new customer because if they don’t show me the ads, I will buy somewhere else.

Mike 00:06:31
Yeah.

Mike 00:07:13
Let’s unpack that. You’re on Google search starting your shopping journey. If Zalando wants to focus on new customers, they can activate a new customer-only bidding mode in PMax. Google has its own system for detecting new versus returning customers. You can add your customer match list and say, “These are my existing customers.” Assuming you are logged into Google across your devices, you should not see those Zalando ads in Shopping PLAs or elsewhere.

Zalando has just gone dark. They are depending on their organic visibility and your loyalty to go directly to them. Just because you are an existing customer of Zalando does not mean that you belong to them.

Chris 00:08:36
That’s a smart quote. This idea of not running ads for existing customers is a pretty substantial fallacy. Where is the arrogance coming from to assume that I’m always buying there whenever I have a new demand for a product? On theory, it’s a good thing to focus on new customers, but what is really an existing customer? Just because I’m an existing customer doesn’t mean anything in terms of my buying behavior. In the case I brought up, I started my research on Google and I ended up buying somewhere other than Zalando.

Mike 00:09:25
Yeah.

Mike 00:09:38
New customer campaigns are a tactic. You have to think: “This is just one tool in the toolbox. If I’m going to use this, what can I do to make it special?” Which products do I want to advertise? Are there ones that make sense to advertise exclusively to new customers? You can have tailored offers and a tailored experience.

But to close the circle back to high CPCs, we talked last episode about the painful upward trend. That’s a market effect. In general, high CPCs correlate with high conversion rates and high order values—a high value per click. Google is very good at matching value per click and cost per click. The problem is that everyone is using that technology. Smart bidding adoption must be upward of 90% or 95%. Every advertiser is using this technology. If Google is so sure about a conversion, they must know that person. If you’re not bidding on them, everyone else will be. It’s a red ocean.

Chris 00:12:11
This technology intensifies how cutthroat it is. As an online shop owner where it’s all about incrementality, you just don’t have the luxury to say you will only focus on new customers and catch existing customers somewhere else. The lines between existing and new customers are so blurred. The only thing I care about is revenue. When someone starts research on Google, the theory about an existing customer goes out of the window. I should have gone to Zalando directly, but I didn’t. From that point on, I’m not an existing customer anymore.

Mike 00:13:27
That’s right. They may be in the consideration set, but they’ll have diminished visibility while others will have promoted visibility. This is an order acquisition or revenue acquisition channel. Customer acquisition happens and is a part of it, but it’s not primarily a customer acquisition channel, especially for standardized products like Adidas sneakers.

Chris 00:14:13
Meta or Google Demand Gen are more like customer acquisition channels. There are different campaign types for a reason. PMax is a revenue acquisition channel. Whenever you try to focus on this theoretically existing versus new customer tactic, you’re leaving out a lot of market. This existing customer is then acquired by someone else. I don’t like this theory at all; it doesn’t make sense to me.

Mike 00:14:56
The bidding strategies are called “Maximize conversions” or “Maximize conversion value.” There is no strategy called “Maximize new customers.”

Moving on to our second topic: Performance Max is a keywordless campaign type. Even if you add your customer match list, it’s not for targeting purposes per se; it’s a Google signal. It’s like a recommendation engine in Netflix or Spotify where you say, “These are movies I like,” and it finds similar ones without necessarily knowing what you liked about them.

Chris 00:16:18
For the sake of comprehensiveness, you can also add themes for positive targeting that might have been missed in your feed or landing pages.

Mike 00:16:47
Fundamentally, when we look at targeting, there is no keyword or audience targeting. The core thing left in e-commerce is product targeting. The problem is that product targeting is often quite inaccurate.

Chris 00:16:53
Whoa, what’s the topic now? Why is it inaccurate?

Mike 00:18:10
Say you’re advertising a camera lens and Shoppers A and B enter your web shop. People are still browsing, which is your chance to upsell. In the end, the revenue in the platform is attributed back to the product that was clicked. But what happens next is anyone’s guess. They could buy that product, or buy a tripod as an upsell, or go for a camera bag instead. While you can target a product for advertising, you cannot control what happens next. That’s human nature.

Chris 00:19:10
We saw online shops where this deviation from clicked versus bought is highly significant. Let’s walk in the shoes of an online retailer. What is the underlying issue from your perspective?

Mike 00:20:26
It’s about understanding what’s going on with the platform and about optimization. Product targeting has become increasingly popular as other promotional levers fell away with automated campaign types. The platform is going to collect that order value and consider it in their optimization. If one product is consistently generating outsized average order values, that is reflected in the ROAS of that product, and Google will know that.

Chris 00:21:09
This inaccuracy is not negatively impacting Google’s decisions. Google has the order value and pushes the product. But Google doesn’t give a flying fuck about the composition of the basket.

Mike 00:21:15
No, they don’t care—or they don’t know.

Chris 00:21:18
But you should care. Why?

Mike 00:22:51
Where this really becomes important is in the profitability of the order. Google doesn’t see this layer at all. It’s fine if there are clicked versus bought effects if they are positive, like upsells or increased margin. There are also downsells and cross-sells. If it’s a cross-sell into a different category or brand, do I care? Not necessarily, if the profit is intact.

But you might have a concern if it’s a co-funded campaign and you’re trying to promote a certain brand, but a different brand is being bought. The vendor brand wouldn’t be happy. Brand substitutions are probably the most sensitive topic among cross-sells. Google doesn’t care about these things which can be important to your bottom line or revenue mix.

Chris 00:23:02
Google doesn’t care because they don’t know that one brand is more important to you than another.

Mike 00:23:10
We’ve said this on stage a million times: Google doesn’t know what they don’t know, and that’s not their fault.

Chris 00:23:45
So, this inaccuracy might hurt your business objectives if the composition of your basket isn’t in favor of your margin. What is the way out?

Mike 00:24:54
Google does have a tool here. They have “conversions with cart data,” which is a small snippet of code. You add this to your existing conversion tracking and you can then know your basket composition. This unlocks reporting attributes like clicked brand versus sold brand and clicked item versus sold item. Additionally, you can add your cost of goods sold, if you dare to share that with Google.

Chris 00:25:20
Let’s make a heartbreak here. A lot of clients are hesitant to share their cost of goods sold with Google. What is your take on it?

Mike 00:26:56
I used to work in purchasing and handled cost of goods sold. In a red ocean category, cost of goods sold is one of your most important levers and it is very sensitive information. People might not want this information to fall into the wrong hands. There’s also a theory that if Google knows your margin, they can tighten the screws. I think that’s tinfoil hat stuff; I don’t buy it. Google has the “shopping graph” where they are building data points about all products. If enough people share their data, Google can reasonably infer what your margin will be anyway. What’s your take, Chris?

Chris 00:28:23
It is highly sensitive information, but at the end of the day, Google is one of the most successful companies of the last 100 years. Do you really think they are super interested in your specific cost of goods sold? The trade-off is that if you don’t share it, you might be optimizing on the wrong data set. The fear is unnecessary. I would go all in because it’s about letting Google campaigns work the smartest for me.

Mike 00:28:32
I agree the opportunity outweighs the risk. I would be more scared of sharing that with Amazon.

Chris 00:28:53
Fair point. Amazon has a different business model. I would be hesitant to even advertise on Amazon in the first place, but that’s a different case.

Mike 00:29:40
If you integrate that, you can bid toward profit. Google has a GitHub repository called Soteria for profit bidding, which is secure. If you do it this way, you get the profit dimension in your reporting. It becomes a formidable data set.

Chris 00:30:25
If you empower PMax to take decisions based on true profit, it seems like a no-brainer. One question: can you modify the cost of goods sold by putting a factor behind it to cipher it for reporting purposes?

Mike 00:30:57
You could do that and then decode it. For reporting, you could definitely cipher it a bit. All right, man. Time flies.

Chris 00:31:41
I enjoyed it as always. Don’t ask me again if I’ll do the outro; it’s not time yet. The Christmas episode will be the time. I’m already nervous because the intro and outro are next level. I always feel like I’m talking to a person who is smarter than me.

Mike 00:31:44
No, Chris, come on. I’m not smarter than that.

Chris 00:31:51
I just want to share my feelings, man. It was a good episode.

Mike 00:31:51
Well, your feelings are valid but wrong. And on that note, we’re bringing another thrilling episode to a conclusion. This has been Growing E-Commerce, brought to you by Smarter Ecommerce. To learn more, visit smarter-ecommerce.com. If you enjoy this podcast, share it with your friends or colleagues and give us a rating. Thanks very much. We’ll catch you next time.