Tom Haslam - (hostaside:
Welcome to Marketing Blabs. This podcast is brought to you by Marketing Labs, an expert digital marketing agency based in Nottinghamshire. If you're a business owner or marketing professional looking for straightforward, non-salesy tips and advice to help grow your business online, then this is the podcast for you. Strap in because we're about to reveal the things that other agencies would rather you didn't know.
Hello, listeners. Welcome back to another episode of Marketing Blabs. Today, we're diving into the world of programmatic advertising, particularly focusing on its implementation within Google Ads. If you've ever been curious about the automated real-time auction system behind those digital ads, then you're in for a treat. We're going to unravel the intricacies of this dynamic advertising landscape together. And by together, joining me on today's Blab, from the ML team, is Matt Janaway, Founder and CEO of Marketing Labs.
Hi, Tom. You all right?
How are you?
Yeah, I'm okay, thank you, yeah.
You got some new golf clubs yesterday.
I did.
And I'm jealous.
I know.
So I'm not going to talk to you nicely on this podcast.
That's fine.
But I did want to give Nick a special intro because he beat me at golf yesterday.
Yeah, he did.
Are you feeling proud of your achievement?
It's quite easy, isn't it, so no.
How many points by?
Two. Anyway, we're digressing, and we're here to talk about programmatic advertising. Do you guys want to give a definition of what that actually is? How does it differ from traditional models of advertising? And, obviously, by programmatic we mean automated. So can you put this into context?
I think you should for us, Tom.
I have no idea what it is. I'll be completely honest and tell our listeners that I have no clue.
Okay, well, I'll tell you then.
Thank you.
Yeah, so automation essentially means it's connecting the various different parts of the supply chain, from your ability to serve an ad, right the way through to placing it against a customer who matches your target audience on the screen that they're looking at for the website or environment that they're looking in. And, historically, there's quite a few sequences that need to occur in order to deliver that. There's lots of different ways of buying that historically, especially through display, that's not automated and is quite manual and takes a lot of time to achieve that, and it's less precise. But, essentially, programmatic means you have the capacity to serve that in an automated fashion for the target audience that you're looking to serve an ad for.
Right, okay.
And most people might not realise this, but if they're running ads, especially in Google Ads, there are quite a few elements of programmatic decision making that happen on everyone's account. So if you go back years and years, you would mostly control your bids manually. But then Google introduced enhanced CPC, which gave some automation to how much that cost per click would be, based on programmatic decisions. So, for example, if the bid you were almost there, you're giving Google a little bit of freedom to automate a little bit of extra spend on the bid.
But are you not losing quite a lot of control over it though?
Yeah, yeah. So we can come onto that a little bit later on in terms of we want to talk about the kind of concerns that-
Challenges.
Exactly, the kind of challenges it can bring. But there's a lot of automation in most Google Ads accounts. And also a lot of that automation is controlled by machine learning and AI, which comes onto another concern about whether you actually have access to data to make the decisions on that data, or whether it's data that isn't actually available and readily available for you to consume, if you like. But Facebook use it to a certain degree. There's a lot of automation inside there. And it's hidden in the platform, taking ownership of a little bit of decision making, which benefits them, doesn't it, of course. Because Google have control over every account type to a certain degree for every product range or every niche or every service. If there's five people competing, for example, on digital marketing services in our area in an ad, Google's got control over all five. So if you allow even an element of automation on all five accounts, they've got levers that they can increase and decrease, based on their automations. And, yes, you have ever so slightly less control in that. But on the opposite side, there's also some advantages.
Yeah, I was going to say, from what I gather of this system, I don't know about programmatic advertising. I have very little understanding of Google Ads as a whole. But the main benefit, I'm guessing, is that it will save the person advertising, save them time. But are there any other advantages of that?
Yeah, so it depends on what platform and what provider you're purchasing that through as well. So Matt was just mentioning there, Google and Facebook, who will have different systems, and within their own ecosystem. You can also extend it across the whole internet as well. So there'll be a variety of different partners that you could partner with in order to serve ads and display ads or video ads to your audience, depending on where they exist, and what they're looking at, and what websites they visit, and how frequently they visit them. So within Google's ecosystem, and within Facebook's ecosystem, you can target them that way. But, also, you can open up your network, or you can open up to the wider network via different platforms as well programmatically.
So, essentially, what it does, and the benefits of doing this, are that you can more precisely target your target audience. So you can define them quite well, in theory. We'll come onto some of the pros and cons about that later. But you can define your audience pretty well. And then you can have quite a tight view of how you target that audience across a variety of different platforms that they might exist. So, in theory, you might have the same audience on Facebook as you do on Google, as you do across online video, for example. And you can target them with either the same message in different formats or different messages in different formats, depending on how you want to deliver that and what your message execution is as well.
There are lots of benefits to targeting your customers in different ways across the sites that they visit. And you can either have a sequential targeting and messaging, so you could target them with a sequence of messaging. So it might be top of funnel, middle funnel, lower funnel, and obviously, hopefully, they convert. It might be a different range of products that you might offer or services that you might offer. Or it might just be the same message that you want to get in front of them with a higher frequency and a high reach against your target audience. So there's a variety of different ways of doing that. That'll largely come down to your strategy and how you approach that. But the execution can be quite varied, depending on what your strategy is.
Interesting. And is the automation applied to things like budget allocations as well?
To a degree, yeah. So if you're just taking Google as an example, you could just give Google a budget and say, "I want to reach this audience," however I'm defining it, "across these ad types and these asset types." And it will spend your budget, depending on how much you give them, it'll spend it in a way that it believes is most effective, in terms of reach and whatever goals you've set up for that campaign.
And you can edit these goals and it'll use different automations. So, for example, one campaign might have a goal for a certain target of return on advertising spend, so you might set that to 300%, for example. But you might have another campaign with a separate goal for maximised conversions, for example. So what you're doing is you're buying customers instead of trying to achieve a specific return on advertising spend. Another one might be about gaining traffic, maximising traffic instead of maximising customers. And all of those are automations really, because you're automatically, if you like, handling some of those decisions over to Google's machine learning, using Google as the example. But most platforms have similar things. Okay, maybe not as sophisticated as Google's. Google probably has more data than any other tech platform that has advertising, I would say. And they also, I think, lean on machine learning heavier than others. But, clearly, as well, it's a direction that Google are heading in. It is only going to get more... I guess manual control over these things is going to become a thing of the past, bit by bit, over the next few years.
Would you say that this type of advertising, programmatic advertising, is beneficial to startup businesses who need Google's, let's say, automations to take over and help them? Or is it more of a specialist subject?
In some ways, in some ways. There are pros and cons to this. So certainly one of the cons is you need data for things to figure... The machine learning, you've got to feed it, you've got to feed it with data. Now, if you've got low budgets, the chances of getting the data for it to make great decisions is slim. So that's certainly one of the cons. But at the same point, some of the advantages, so Google has so much data. We're using Google here quite heavily, but I think it's probably a good example. Google has a huge amount of user behaviour data, so it knows when somebody is at the bottom of the funnel, ready to convert, compared to somebody who's maybe in their research phase. So aligning keyword data, along with Google's user behaviour data from... They use machine learning to make decisions on. Can be a winning formula. It's just the data has to be readily available. It's difficult to get that with low budgets.
Yeah, makes sense.
Other issue, I think, with people who are starting out is that they don't necessarily know, typically, who they're targeting or who they need to target. They might have a view of who their target audience is, but it's not necessarily who they need to target, in terms of performance or generating revenue.
Actually, getting those conversions and-
Tail wagging the dog again, isn't it, a little bit. When anybody starts a business, they should have an idea of who their customer is, but it doesn't necessarily mean that's correct. In fact, to be honest, more often than not, it's incorrect. And, over time, the more data that you can gather on that, if you learn from that data and you hone who your actual customer is, you can benefit from that. But you see a lot of this does it automatically. So it tries to target your ideal customer instead of you trying to define it.
Interesting. Talk to me about best practises then. Are there any tips for businesses and digital marketers out there who want to leverage this advertising effectively?
Yeah, so, again, it will depend on your strategy and your objectives, and also how and who you're looking to target and what format types you are. But, essentially, a lot of the success of this activity will come down to how well you define your target audience. And the volume and quality of data that you can access them via. As an example, Google and Facebook's ecosystem are relatively well resourced and huge scale, so buying audiences across their platforms is quite easy. And you have access to certain targeting methodologies and definitions, in terms of how you want to build your audience up and the profiles that they look like and the attributes that you can assign to them. So you can get a fairly detailed view of that audience. If you're looking to do online video, for example, or looking across the network of other websites that you might want to advertise across, you may need a partner in order to find your audiences more successfully.
So the biggest issue with audience definition, by the way, traditionally, it used to work across third-party cookies, which are almost non-existent now, especially across Apple devices and a lot of browsers. So Safari, Firefox, and pretty much all browsers apart from Chrome don't really allow third-party cookies anymore. And, historically, that's how the entire industry works. So there's been a big shift over time, and especially more recently to how they target customers and audiences. So you might need to partner with somebody like LiveRamp, as an example, who can have a better methodology and authentication process, in order to identify who is actually your person, based on first-party cookies and who's authenticated themselves on one of their partners. So it might be that they get data from Facebook, for example, or whoever, Amazon maybe. So they'll be able to match an actual email address, or first-party data, with a first-party cookie and, therefore, allow you to target them that way.
So that can be crucial in terms of the success of programmatic across the wider internet. Identifying your audience and making sure that you've got a good view of who they are and what they're interested in can be crucial. Whereas, actually, across Google's network and Facebook's network and maybe Amazon's network as well, they will typically have that data and they'll know who their audiences is because they'll be logged in, for example. So that's not so much of an issue across some of it. So all of that will feed into your strategy, how you achieve something based on your objectives, we'll need to factor that in.
And those objectives, again, that's really important. Because if you don't define those objectives, you can't feed the correct output, if you like, which, all of a sudden, means that actually the performance probably won't be there. So earlier on, we were talking about just a few examples of how Google would, I guess, determine how it would automate certain decisions, would be based on campaign objectives. So your campaign objectives, if you wanted to buy in more customers because you knew you had long and high valued lifetime value from a customer, you're actually less worried about the cost of that first order because you might get two, three, four, five more orders. Whereas, if your customers only order once and they mostly never come back, then your objectives are going to be slightly different. So you might actually, at that point, be wanting to target a particular minimum return on advertising spend. So, again, these objectives are really important, so that, for sure, is a best practise. You need to understand what you want out of this in order to make it achievable.
I think there's something else I'll throw in there as well is to monitor absolutely everything you can. So when machine learning makes a lot of decisions, generally, you don't get to see what those decisions are or a lot of the insights from that. But what that means is you should try really hard to monitor everything you can have access to. Because if you only have access to a limited amount of insights and data, you really need to know what they are. Otherwise, you're going to struggle to make decisions. So monitoring everything is really, really important.
Also, another thing is when... Again, I'm talking specifically here Google but, again, other platforms do do this to a certain degree. Google have a lot of networks that they can advertise on, so you've got YouTube, for example. You've got search. You've got their display networks, which is websites that allow advertising on behalf of Google partners. So you've got a lot of mediums where your adverts can show. By default, if what you're doing is... If your objective is to get more eyeballs on your advert, they'll generally use those networks. But that doesn't necessarily mean that it's right for performance, if your objective, for example, is to try to find people at the bottom of the funnel who are ready to convert, because a display ad is not that. It's more about awareness. So, again, objectives, it comes back to what those objectives are. But good practise would be to monitor what those networks are, because if those networks don't match your objectives, you need to be making sure that you're not advertising on those networks.
Interesting. Some good insights there. What about the challenges and concerns then of the platform? I know we briefly touched on this earlier in the conversation, but just want to address some maybe common criticisms of the platform or the advertising for programmatic. Are there any concerns or challenges that you find?
Yeah, the big one's data, really, and we touched on it a second ago in how it used to work.
With the tracking of cookies and things like that.
Tracking of cookies, yeah. That's fundamentally changed since GDPR and Intelligent Tracking Prevention from Apple. So the vast majority of that data's no longer available for most people. And if it is available, you may mainly only have access to it for 24 hours. So how you target customers changes, like I mentioned for Google and Facebook, they have regular people that will log in pretty much on a daily basis, or remain logged in permanently or persistently. So it's not so much of an issue for them, but what they do on your website, assuming it's web-based, may or may not be allowed to be collected in terms of the data. So GDPR, obviously, if you don't have your cookie consent policy set up correctly, your tags may not fire, therefore, Google may not be able to collect data in the way it needs to. The same for Facebook.
So data really is fundamental to the system working, so do all that you can. And make sure you have everything set up legally, obviously to protect yourself and also your customers. But make sure that, obviously, when they do accept and they do consent, that allows you then to track them ethically and correctly. And they allow you to say and suggest what data you are taking from them and how that's being used and what systems that's being passed into. And that then allows all of these partners that you're choosing to advertise with the correct consent in order to use their data correctly. And that, ultimately, will benefit you because the quality of data will be much higher than if you don't have that. So, obviously, a lot of those systems now are fundamentally required to work based on acceptance because GDPR dictates that. And that's a big part of whether it's going to be successful or not, and will certainly be more so the case in the next year or two.
Interesting.
And just to touch on that, if somebody declines your cookie policy, if your pop-up says, "Please accept cookies," and someone hits, "No," that's it, you can't track their data. And machine learning can't use their data to make decisions. But, likewise, even if somebody doesn't click decline, you shouldn't be tracking them until they click accept. So, again, you don't have their data to learn from. So, actually, as Nick said there perfectly, the more data you can gather and input into a system, the better decisions it can make. So yeah, that's a real concern.
Something else that, more of a psychological concern here, but you've got to put a lot of trust into the hands of these tech giants that you're advertising with, not only your data, but also that they're making decisions that are beneficial for you. They are hit-and-miss sometimes these machine learning campaigns and these automated campaigns. Sometimes, they can work incredibly well, really, really well. And then other times, less so. And it is mostly because they don't have access to enough data, but also, sometimes, humans can make better decisions. We know this. It's just humans can't process the amount of data that a computer can. But, also, it doesn't have access to the data that a computer has access to. So I would still argue humans can make better decisions, but they can't make them in the same timeframe and they don't have access to the data to make those decisions. So you're handing your car keys over to Google, basically. But the problem is, if everybody else is handing their car keys to Google, Google's in charge of the race. So there's a trust element there for me.
Well, this is a big thing. I've just done a quick search, and obviously looked for some general average consent rates for cookie policies. And just quickly, just the top Google search comes in, the average consent rate is 31%. That in itself is quite low, isn't it?
Pretty low that, yeah.
So that makes it challenging in itself, doesn't it?
It does. Nick's been working on this recently, actually. And we were looking at some, I think we're seeing about 70%, 75%.
Yeah. To be honest, I've not seen it as low as 30%. But, obviously, that will take into a huge spectrum of different websites and whether they're trustworthy or not.
Yeah, I think it's quite broad. It was just a very quick search.
Yeah, yeah.
But it does show you the issue. Even if you were to just be quite broad and say, "Okay, well, it's between 30% and 80%." Even 80%, at the highest level, you're missing out on fifth of all your data. Now, if you're relying on machine learning to automate making decisions based on that, well, they're missing out on important data. So that is a risk, that is a concern. And throwing all your trust in Google when they've not only got your car keys, but the car keys of your competitors, if they just drove five mile an hour faster, if everybody drove five mile an hour faster, you're not doing any better. But it's costing you more, in the race that is. So you're not winning the race. Everybody's driving 5% faster, five mile an hour faster, but you're still not winning the race. You've got to get to 6% or 7%, but then if they go to 6% or 7%, you've got to go to 8% or 9%.
The problem is if they're in control of all the cars, for me, that's problematic. So there's definitely a trust issue there. That's one, for me, I know it's a psychological one, but, for me, that's probably the biggest challenge. And things are going to become more and more automated as time goes by. And manual intervention will become more difficult.
So I think that leads on quite nicely to the next talking point really, which is the future of programmatic advertising. So with the evolution of AI and machine learning, where do we think that all this is going to head? Where's it headed with-
It's definitely going to continue. It'll keep going, and manual intervention will become less and less. That's for certain. I agree with what Nick was saying earlier in terms of user behaviour. So the more you can understand your users, the more you can control your own data, the more you can understand who your customers are and what your objectives are, that's going to have a big impact on performance. But, for me, it's absolutely heading in that direction at some point soon.
Google, not too long ago, they changed their keyword match types. They loosened phrase match. So what that means is phrase match, historically, Google would try to match the keywords that you're entering pretty closely with what people are searching for. So if somebody searches for something ever so slightly different, they might still see the ad. Well, phrase match is getting closer to broad match, which means, actually, if it's loosely connected, your ads can still show. So that adds another element of intervention that you need to make. But that's going to become more and more challenging. And the more Google users machine learning, the more they can implement things like that, which takes the control out of your hands to a certain degree.
I think the big thing for me, this is more specific to Google, but it will impact the industry, no doubt, I think what comes out of this antitrust case in America at the moment will be massive, in terms of how, not just programmatic, but marketing generally speaking, will work. And I guess similar to how Microsoft was broken up 20-odd years ago, if that happens to Google, what's the impact of that? How's it going to work? Are systems fundamentally going to change? Does that mean more automation? Does it mean less? How do the systems talk to each other? That could have a significant shake up of how the industry works. And what that means, I guess nobody knows at the moment. It would probably be broadly similar to how it works now, but just more segmented and siloed, and less controlled from one business, which is probably a good thing, ultimately.
But, obviously, we don't know the ramifications of that in terms of how data's transferred, and how you target, and how effective it's going to be. And there's a lot of considerations, I suppose, that will come out of that that we won't be aware of at the moment. But that could also happen, potentially, across other massive businesses, Facebook, Amazon, or Meta, I should say. And Amazon and others I'm sure might be concerned about that as well. So I think it's not necessarily a specific answer for you, but that could have a massive impact next year or the year after. It could be quite a different environment than we're-
It's going to reshape the whole landscape, really.
Yeah, for sure. And Google's an obvious one for that, in fairness. Facebook's slightly more difficult because there's basically two sides to Facebook, if you forget about the Oculus side because it's so tiny in comparison. But Facebook really has two parts, doesn't it? It has advertising and then the actual platform, if you like. Google has many moving parts. They've got product, they've got Android, they've got Chrome, they've got YouTube, they've got search. Then inside search, they've got ads. They've got Google Assistant, Google Home, all of these different things. So there's more obvious lines, if you like, between the services, and there's many more services. So depending on how that case goes, it could break up in certain ways. Who knows what that would be. Obviously, the vast majority of its revenue comes from search ads. And that also dictates how search results look. Because we see quite regularly now, no organic search results above the fold, for example. It's mostly ads. That wasn't a thing too long ago, but that's the pressure, I guess, from shareholders to ensure that they continually grow.
But when you've hit peak audience and peak usage, where's that growth going to come from? What they've got to do is just control more real estate on the search results. So yeah, I agree with Nick. I think if that breakup happens, who knows. What does that look like? Nobody really has any idea at the moment, do they? If it doesn't happen, I can only see that divide between organic and ads getting more controlled towards ads. Because that's the only obvious way Google can grow their ad revenue is by having more real estate, or by controlling the race and everybody's cost increases. But at certain points, some advertisers will possibly just say, "Well, this isn't working for me, so I'll pull out." So that's a very dangerous, long-term-
Risky game.
Yeah, very risky long-term game, even medium term game. Short-term, fine, you could probably increase your revenues. You do that for long enough, you'll lose advertisers. So yeah, tricky one. But I think, for certain, there's commercial pressures for that advertising revenue to increase. How they go about that, who knows. I imagine automation and taking control away from advertisers and pushing it towards machine learning will be a part of that. It's just an educated guess.
It's clear, isn't it, that's what Google wants. You look at PMax at the moment, and they're heavily pushing PMax, just because that gives them control over where to serve your assets and who to serve your assets to. So that can go across their entirety of their network if it thinks the audience matches a person that you should be targeting. And like I said, I think the only challenge to that will come if Google does face issues with this antitrust case.
It's not necessarily a bad thing, by the way. Some PMax campaigns can work so incredibly well because they have access to data that we don't and that nobody does. They understand, to a certain degree, at what part of the journey a purchaser is on. If they understand that and nobody else does, that click is worth more, let's say, than somebody at the research phase, because that's more about awareness of you or your product or your service. So there are lots of advantages as well. It's just the risks are also... They're pretty big risks.
I guess it's getting a good balance of everything, really, isn't it, in all of your campaigns, whether that's manual setups to the automated setups and PMax. As long as you're filtering through and getting your campaign set up that works for you, not solely relying on programmatic, for example, which would pose all its risks. And making sure that you've got all your data. So there's lots to consider, really, isn't it?
I think the key to all of this... And this always has been the key, by the way, nothing new here at all. The key to all of this is not to have all your eggs in one basket. You need multiple streams of revenue if you're any business type. If you're relying on one or two streams of revenue, that might go absolutely perfect forever. But at the same point, if something were to happen, you could be in bother. So it's about diversifying. And there's a reason the most sustainable businesses diversify their revenue streams.
So I think, ultimately, it comes back to good strategy. And Matt said there, you're just not having all your eggs in one basket. Making sure that you've got a good view of your strategic thinking and decision making, and what that means and the performance and the impact of that over time. And making sure that you optimise towards the best performing strands of that strategy. But also not being afraid to test new things, because it's equally important, especially as and when new platforms come along like TikTok. And two or three years ago, most people probably wouldn't have had a TikTok budget. And, now, lots of people do have a quite hefty TikTok advertising budget. So things change, and it's being open to that change, and understanding where your audience is, and how you're targeting them, and what you need to say to them. And that will remain the case, I think, in marketing for a long time.
We say this a lot, but test, test, test. You've got to test things. And you should see your test spend as part of an investment, as opposed to a budget or an overhead. Because every platform will have a different return on advertising spend, but they all form part of a jigsaw. It's a customer jigsaw, attaining new customers. And the thing is, you've got to match all of that perfectly. As Nick said, it's the strategy. If you're missing pieces of the puzzle, your performance is going to be lower. So even though certain platforms might have a lower return on investment or return on advertising spend, they're no less important sometimes. So you've got to diversify, and you've got to, in your head, have a contingency for spend to, at the very least, do your very best in areas where you're not currently capitalising on. Because the moment you can start doing that, the more sustainable you are as a business.
Awesome. Thank you both for coming on. Have you enjoyed talking about it?
Yeah, a little bit.
Would you rather be playing golf?
Based on yesterday's performance, no. Based on five or six weeks ago, yeah.
Awesome. And there you have it, folks, an expansive look into programmatic advertising within the realm of Google Ads. From its automation wonders to the challenges it poses, it's evident that this programmatic advertising isn't just the future, it's the present. As advertisers and businesses, understanding its nuances can be the key to unlocking the treasure trove of digital marketing opportunities. Remember, in the rapidly changing digital landscape, staying updated is not just beneficial, it's going to be essential for your business. Thank you for joining us today on Marketing Blabs. And until next time, keep experimenting, keep innovating, goodbye, and happy marketing.
