How to Know When to Scale Your Google Ads: A 3-Step Framework for Trusting Your Data

On-platform data doesn't always reflect reality, so how do you actually know when to scale your Google Ads? 

In this episode of the Google Ads Unleashed Podcast, host Jeremy Young shares a practical three-step framework for making confident budget and scaling decisions, covering post-purchase surveys, first-party tracking and attribution models, and on-platform metrics. 

Learn why incrementality matters more than reported ROAS, how to choose the right attribution model for your business, and why channels like YouTube and branded search can wildly distort your picture of what's actually driving revenue. If your on-platform numbers and your bottom line never quite seem to match, this episode will change how you interpret your data for good.

Get your free 30 minute strategy session with Jeremy here: https://www.younganddigital.marketing/

Scale your store with 1:1 coaching: https://www.younganddigital.marketing/1-2-1-coaching

Transcript

A client's asked me recently, how do I actually trust my own data? When do I actually scale? What do I do to interpret the results that I'm getting? And since this is such an interesting question, I thought I'll do some ramblings about this on this episode. Welcome back to Google Ads Unleashed, guys. Hope everyone is doing fabulously. this Monday. So, last week I had a client, actually several clients, come up with as simple as a complex question to me and they said, "How do I actually know when to push and to interpret the data that I'm seeing?" Okay? Because they intuitively know one thing that onplatform data doesn't actually always equate what is happening on the bottom line. Okay? And And I wouldn't say I have mastered this. No one has mastered this, right? Because the the the real world is too complex to understand how everything kind of works together. Which is why humans always try to abstract things, right? We always use models to explain try and explain real life behavior, right? The idea of a theory or model is to explain very complex sort of processes. is to the best degree possible. Okay, so you can think about this of everything, right? Like the theory of relativity is a set of couple of equations that explains a very very complex physical phenomenon, right? Or I don't know a map on your phone isn't actually the real world of course it's just a very uh simplified abstraction of what is happening in the real world. And we do exactly The same in marketing, right? We use different attribution models to um explain what is likely happening in the real world. We use metrics that sort of try and measure some real life impact that is actually happening but isn't really the reality, right? Uh I don't know, for instance, view time or something like that, right? It's just an it's it's an abstraction of of a of a real life thing that happened. And in order to find out what to push in your ads, for instance, which channels to push or which campaigns to push, you have to over time find out what is the best route to choose a level of abstraction for how to explain what is happening in real life with your business. Okay? And the way I go about this I just want to offer one kind of viewpoint on this today. And this podcast is meant to be just more of a sort of philosophical thought of how to arrive at this conclusion. Okay? And in my opinion, in order to understand what is probably reality, um you have to do three things. So maybe four things. The fourth one I'm going to say straight away because The issue that I've had or that we have in digital marketing is that what we attribute to or attribute rather to uh various marketing campaigns or touch points is not what equates to reality. So I've done a full podcast a couple of podcasts about this. Um this is the concept of incrementality. Right? So for instance a rorowass of five on platform could mean that you in reality have a rorowass of eight or um or no row whatsoever, right? Um, what I mean by that, for instance, a branded campaign will always show great rowass, but the reality is it doesn't really give you a pound for or£10 for each pound you spend. That pound was earned in a meta ad or a YouTube ad that has generated the brand search, right? So, you have very very low incrementality but very very high attribution. Whereas on the other hand, you sometimes have extreme poor attribution such as YouTube. YouTube ads and as a result uh or or sometimes shopping ads can look quite poorly but in reality you have um very very high incrementality. Many people are buying. We see this even sometimes with a store that we um that we support. It generates 35 to 40 sales a day when Google ads is on and it's absolutely flying. And on days like yesterday was a bank holiday which typically very quiet but you'd expect even a few sales come through organically. No. Um uh when Google Ads is off, even on a weekday when it's randomly off, when the payment methods like gone down, this has happened a few times. Um then just no sales come in, but it's not as many as it shows on on the dashboard, which makes you wonder, you know, is Google more incremental than than than it actually shows on the platform. So it's so understanding incrementality is extremely important in understanding the concept really. already helps with your budget decisions but it ultimately really helps you to choose the right attribution targets in order to achieve real life outcomes. But this is a completely different sort of uh debate. I don't want to open that up today. I have a simple three-step framework which is the first three steps through which I understand how to make budget decisions, how to make rorowass decisions, how to make um decisions to push and that is number one understanding your post-purchases. survey arguably I find it is one of of course even that can be skewed because people can give wrong answers or they uh don't understand the post-purchase survey sometimes but inaccuracies however it is I mean you can always debate about that the reality is people who tell you where you know you're from or where they know your brand from is a massive indicator of if anything's working right let's say you run no radio ads and you have radio as as a as an option in the post-p purchase survey, absolutely no one is going to um likely click radio ads, right? But what will happen is once you start radio ads, more people will start clicking it. It gives you an indication of how well your ads are working. Let's say you made 50,000 in a week and suddenly in that week you've uh you've started on radio ads and then 10% of people then say, "Oh, I've hit this um uh of this brand on on the radio." And you spend £1,000 on all of the spots on the radio spots. Well, you have an ROI of about five, right? So, it gives you a good indicator of where to push. The same applies to YouTube ads. If you have YouTube as an option um in uh in the post-p purchase survey and then suddenly you get an incremental uplift uh once you actually run ads, let's say only 5% or something tick YouTube because well, they probably have seen a YouTube video about this product or about the idea of this product and then searched for it and then bought not yours. So that's what I mean with inaccuracies and post-purchase survey data, right? So um some people may see a product on Tik Tok Google then and then purchase because they found it on Google, but then they will actually still take Tik Tok in the in the thing because they saw an organic video, right? It doesn't mean that uh um Google wasn't important in the customer journey, but it's just a very very good indicator still um where people come from, right? And if people organic say YouTube let's say 5% in this example and then suddenly you run YouTube ads and it goes up to 10% you know it is working okay the same if you remove a channel and it goes down then you know that channel has been working right so that's usually an indicator that gives me a right idea of sort of overall channel return on ad spend and how much budget I should spend between different channels okay uh we have one client where basically Google ads and uh meta ads is always 50/50 50/50 every single month. So the budget decision is super easy between me and the meta ads guy, right? Every month we know how much you want to spend this month, how much is okay, great, done, right? Sometimes it's more complex, sometimes it's less complex. That's just the reality of the situation. Uh but so yeah, post-purchase survey gives you a really good indicator if something is generally working and how much budget you should be spending roughly and ballpark figure and uh what kind of return on ad spend you're actually getting right now. Okay. Then the second consideration that I usually have is I then look at um some form of uh uh business intelligence and uh and tracking. Okay. So overall business intelligence something I always look at is ME and CM3 right or EIDA mega important to just understand the overall health of the business and uh if our marketing spend overall is sufficient but what I then dive down usually this exists with high ros triple whale north beam whatever is some form of first party tracking right so because the issue that you have with Google ads and with meta ads is that they work on two independent pixels that basically don't talk to each other as you probably know and when you have something like triple whale or whatever it actually has just one pixel so if someone comes via meta ad and then uh searches for um uh for for for the product on Google and then buys on Google, it actually understands to not so Google will then track a sale and meta ads will track a sale but triple way would of course only track one sale because it only was one sale that happened and will attribute it to the original source more properly. But then the challenge that you have here is that you usually have multiple attribution systems. And of course I could probably do like another podcast on this but broken down I want to get back to the abstraction thought, right? An attribution model, whether it's first click, last click, positionbased, whatever the f*** it is, is an abstraction that ex tries to explain the reality as best as possible. And the reality is that for any kind of business, right, there are typical paths of how people find out about the business, right? So there are some businesses that have a lot of search volume. So typically um people are very often find the business via Google and then convert. Then there are other businesses where there's no search volume for the product. So they can only convert via push channels realistically, right? Um and then a plumber will need completely different marketing than a skinincare brand and so on. So there's typical funnels and behaviors associated with each industry and your job is to find out over time what is kind of the common pictures that I'm seeing of uh understanding how to attribute um um like what kind of attribution model I should use in order to find out what is supporting my theory of how people navigate my e-commerce shop and the marketing channels that I use in order to end up with a purchase. What do I mean by that? Let's say I've had this example the other day. We I had client they got they got met ads running it's working they got Google ads running it's working and we've verified this through post purchase survey as well and they asked me what is the best attribution model on triple whale to understand how to uh how to optimize each and I saw zero overlap right doesn't didn't matter which attribution model I I used there was no overlap between the channels and I said to them listen the reality is you should then use the attribution model on triple whale which most closely match is what how the attribution models on the platforms work which would be triple attribution right because it doesn't matter there there's no interaction right so the system that is closest to that on the platform actually explains very well what the behavior is on the platform okay um and of the marketing channel then I had another client that more or less uh Google uh is sort of 80% caused by meta ads well attribution is then terrible of course because you would give Google much too much credit as it doesn't operate in its silo rather you would want to probably analyze Google on either total impact attribution which is uh takes into account for instance post purchase survey data or something like first click right where uh the idea of the channel is to um to to give credit when it generates the first interaction right so which is why first click data would then of course attribute much more towards meta ads which is the introductory channel rather than Google. Um or for instance you even within Google ads the certain channels work differently right for instance uh triple attribution or last click would be a terrible model to judge meta uh to judge demand gen. But first click is a much better model to judge demand gen because that is typically the consumer journey someone has is a push channel is the first touch point that someone has and further down the line they purchase right. So the challenge in step two is to understand which attribution model supports my theory of how people very likely buy with me and then I can base the data on that. So let's say you use triple attribution in here in this one client example with the two different silos. I saw in the post purchase survey data both channels work. I I saw um on the first first party uh in the first party data that there's very little um overlap. So they in work in a in a silo. So triple attribution is great and based on that I can actually optimize the ads. And then the third thing that I'd look at is then of course on platform metrics and attribution. Okay. Does that support what I'm seeing? And very often if you get the first three things right, the last thing will be kind of right as well uh with a few nuances. If there's massive mismatch typically you have probably some form of um attribution or tracking issue, right? Like there's then an issue that you're over or under reporting conversions. Probably something then to look at. But let's say for instance in this uh one client that I've just mentioned, we saw post-purchase survey data. Google is driving 60 to 80% of the uh 60 or 70% of the revenue um of new customer revenue and meta ads only sort of to 30% or so. So we've uh based the budgets based on that. Then in triple attribution and other attribution methods, we've verified they work completely independent of each other. So there's very very little overlap in the customer journey. And then we looked at on platform metrics and he uses uh profit metrics. So pas is of course the northstar metric to a certain degree. Uh we haven't done an incrementality test yet but we're quite conservative on pas targets and now we can actually say confidently when we increase budget and put more towards like this uh shopping campaign which does better than Pmax or whatever it is. we're going to actually increase revenue, right? And that is how you can actually judge what you're doing and make uh uh informed decisions or for instance demand gen right another client of mine who asked me how do I scale demand genen well first of all we looked at the post-purchase survey data we see an incremental uplift then by first click we can see couple of creatives are working better than others so we can push those into a scaling campaign because when we looked on uh the platform the sort of core metrics that we'd look at. So, for instance, uh click-through rate and view rate um support that people actually generate a lot of clicks and and come off that uh um uh sorry, I just knocked something down. Come off that uh um channel. Yeah. Um but uh yeah uh but and and not viewing too much of the video and wasting ad spend. Okay. So, we basically aligned them three things to then make an informed decision of what we do because that is what refle is reflecting reality that as best as possible in an abstract model and it takes practice to do so, right? You have to try and engage uh as much as you can with with that sort of way of thinking in order to make the right decisions. And um that is just a really important takeaway from all of this. Okay, so if this was helpful then please like and subscribe. If um you still need help with it, just get in touch, Jeremy on Google Ads on LinkedIn. You'll also find me uh on the website youngigital.marketing. You can book a one-to-one call there with me. You can also send me an email at jeremyyou youngdigital.marketing and I'm more than uh happy to help you there as well. This has been Google Ads Unleashed uh with your personal Google Ads expert Jeremy Young and I wish you a happy and productive week ahead.