And demand gen and YouTube. Basically, they're using search, maps and YouTube signals to predict who wants your brand before they even search.
What does that actually mean? Deciphered into plain English.
They're trying to basically gather together all the signals and then show people the right ads, even if they haven't searched for the thing. If they're on YouTube and they've also been doing other things, then these ads are going to be more relevant because we can kind of work out that this is what they're eventually going to do, or what they're eventually going to ask for.
So it sounds like remarketing on stilts really, isn't it? Which is either going to be great or it's just going to freak people out even more.
It's remarketing without them having to have clicked anything. Just done some.
Premarketing.
Yeah, it kind of is. That's kind of what it is. It's premarketing.
Welcome back to Digital Marketing From The Coalface. And um, to start off, I've got a blank.
Blank, blank. You've got a blank page in front of you, but you have something to talk about.
I have got something to talk about. Yeah, it's a blank page. I'm getting some curious looks from people walking past.
It's hot. We've got the front door of the office open. Two people are staring at us like, what are they doing?
So, um, I went out to get some drinks — for those that wanted before we started recording because it's a lovely hot day for once, office door open and everything else. And you said, oh, get me something spiky. And I thought, Sprite. Well, that's lemonade. Didn't have any Sprite. I saw this one. And I wasn't sure whether it was full of chemicals and crap.
Yeah.
Or whether it was an okay lemonade. And it's an R White's lemonade.
Yeah.
So R White's. I'm a secret lemonade drinker. Do you remember that advert from years ago?
I know about the advert because it's well known. I don't know if it was one that only showed in England. I don't know if R White's were in Scotland — I think they're English. I don't remember seeing it.
One hundred and seventy years of making it. The secret lemonade drinker — it was a guy who gets up in the middle of the night, opens the fridge: "Secret lemonade drinker, and I'm trying to give it up." And I could remember it as clear as day. Advertising works, doesn't it? Because that made me think — I remember the advert from when I was probably a kid, years and years and years ago. And I thought, oh right, I didn't even know it was still a thing. So I'll just get the R White's lemonade. That advert from fifty years ago just made me buy a can of juice now.
I genuinely didn't ask for lemonade. I asked for Sprite or 7Up, which are really strong brands. So I'm not even thinking lemonade as a category — I'm thinking Sprite. Which actually is similar to one of the things I've written down.
You've written down Sprite?
No, no, but it's weird because — one of the things I thought. It's probably not worth talking about because it's not really relevant to everything else, but it turns out it is relevant. I was listening to a podcast the other day which I'll plug because it was actually very, very good — the Uncensored CMO. It's excellent. He had Rory Sutherland and Tom Goodwin on as special guests, and it was incredible. Rather than regurgitating everything they said, just go and listen to it. It's very, very good.
But one of the things they said, which leads back to the branding point — why I said Sprite rather than lemonade — is: if you're famous, people go and look for you. If you're not famous, you have to go and look for people.
Okay. Yeah.
So rather than somebody trying to sell me lemonade, I just said Sprite because I'm asking for a Sprite. Because that's a famous brand. And the branding thing is — you ask for the brand rather than the product because it's so well known, and it's a way of helping you make a decision. If you said lemonade to me I'd be like, well there's all these lemonades, did she mean Sprite? But if it's a strong brand name, you feel safe buying it. You think the product's decent because you've heard of them and they do good ads.
But you think that's like one of the challenges of B2B? Just to make this episode of the podcast vaguely on topic. If I'm thinking of buying a car, there are certain brands I will buy and certain brands I won't buy. If I'm thinking of buying a pair of trainers, I've got a few in mind. I'd never heard of Hoka when I bought them and I quite liked them — I would now buy them again. But when it comes to B2B, people are, you know — if I need to build a tether management system for an ROV, I don't go, "Oh yeah, Acme, they're the people." More likely people are searching for the thing rather than the brand. They're looking for a tether management system, or somebody who can build heat exchangers.
Yeah. So if your brand isn't strong enough for people to go, "I want an X, Y, Z heat exchanger," then you've got to be a lot more convincing and have information there that makes people feel safe in the way that a brand makes them feel safe. You've got to have certain elements that make them think, yeah, this one looks better than that one.
We always say that when we're talking to customers — we've got to make people feel safe. We've got to give them some social proof because they've never heard of you. But if they see that other people have used you, that's a good thing. Certifications make sense because they expect to see this and that certification. And then some case studies to show them that you've done it before. Adidas don't have to do that if they're trying to sell me a pair of trainers.
Because they're famous.
Yeah. You don't need any of that because it's like, well, they're Adidas.
Exactly. And that actually links to one of the other things they said. They mentioned engineering-led companies — people like Dyson — and talking about how you've got these product variants, the ZX3 Point Six and the ZX355 Point Four Two. And you've got to explain what the difference is, because it means absolutely nothing if you don't name your products in a way that makes sense to people. Otherwise people get confused and they walk away.
Done that recently — I'm struggling to remember what it was I was thinking of buying. I finished up being so confused I just abandoned it. This version did this and this, the other version did that, and this version did all of that and probably — I was just totally confused.
And they say that — people do get overwhelmed by the amount of choice. And actually less choice is sometimes good. The other thing is you can use pricing to show the difference. If one thing has X and Y and the other has A and B, make a difference in the pricing so people kind of know which is better or worse. They gave the example of a Nespresso coffee machine — they brought in a different model that did something different but it was exactly the same price. So you didn't know if that was a good thing or a bad thing. But if it said "this does this and it's more expensive," you'd think, okay, well that's a fancy thing — do I want the fancy thing? You can use pricing to almost help the choice process and help explain the differences.
I think that's one thing that Apple get right — you never buy on price. You buy what you need and you don't worry about the price because the price is what it is.
Well, because I'm wanting a new Apple Watch. My Apple Watch does everything but it doesn't connect to some of the running apps. So there's the Apple Watch and then there's the Apple Watch SE — the SE is way, way cheaper. And they're very clear, they tell you the things it does and the things it doesn't do. And for what I need, it does everything I need. It's only a couple of hundred quid. For the extra five hundred quid I do not need all these extra bells and whistles. You can make a decision really quite easily on that.
In the second hand car market — you can see what look to be exactly the same car with five grand difference in price. But then when you dig into it, it's like, yeah, but this one has got the panoramic roof and the leather seats.
And then you go, how badly do I want these things? Most of the time it's like, yeah, they're kind of extras. But pricing is a great way of figuring that out.
We worked for a company some years ago based down in London — they were pricing consultants. Huge company, fourteen hundred of them, offices all over the world, primarily helping businesses with pricing.
Pricing is complicated and quite difficult to get right. It's really hard to put a price on things.
Yeah. Because in our line of work, if there are any other agency people listening to this, they'll know that traditionally people sell time. Selling time is fine, but it penalises people who are good — efficient. People who can do something really well and do it really quickly make less money than people who are slow.
Yeah. So obviously that model is a bit rubbish.
It is, but it's the one that people understand. We tend to price on deliverables — value pricing using a deliverable model. We work out internally how long it's going to take, but it's also the value that you're adding and what the thing is worth in the end. Remember there was — and there may still be — a cohort of agency owners, mostly Americans, who had adopted a points-based system.
Yeah, which we tried. And people got kind of settled down and got bogged down in "why are you charging me seven points for that and five points for that?" It actually made it more contentious.
If you've got a nice round number — say a retainer of five thousand quid a month that buys you fifty points — people immediately think every point is a hundred quid. So straight away they're going, that's going to be five points, that's five hundred quid. Why not just say, that thing we're going to do for you is five hundred quid? Pricing is — I don't know how we got onto pricing.
We got there via a roundabout route about engineers. But yeah — it's about how engineers describe their products. And if you have a fancy name and don't use some mechanism to explain the difference between two things, pricing would be one of the mechanisms you can use.
In the podcast — the Uncensored CMO — they used Dyson as an example. What was the context of that?
You're going to buy a Dyson and you have all these different models with weird names — one's got a so-and-so filter and one hasn't, one's the V9, one's the V12 — and none of it makes any sense. You really don't know what the difference is between them.
How is Dyson a hugely successful company? Are we saying they've got it wrong?
They were saying that the company is led by engineers, not marketers. And therefore the approach to naming products is a very engineering approach. The model numbers and the naming of the models is very engineering-led and not very consumer friendly — that was what they were saying.
Yeah, it's a problem from a consumer's point of view as well. When you come to buy new filters — I bought a box of aftermarket filters, I thought I'd bought the right ones, and I finished up throwing a box of three filters in the bin because they were the wrong ones. Oh no, yours is the V10 XP or whatever. Plain English, keep it simple, stupid.
Keep it simple and don't provide too much choice. And if there is choice, make what the difference is between them obvious. Anything with Rory Sutherland is good.
Yeah. I will get around to doing his mad Masters course at some point. Right — all the Google announcements last week.
Yeah. Basically a big announcement fest — a video that involved people standing on ladders. It was very fast moving and they packed a ton into an hour, which meant they skimmed over quite a lot of stuff. But I tried to summarise the stuff that's probably relevant.
Some of it is very e-commerce-y and stuff we don't really need to worry about, but they're going to put ads in AI mode. So Google Ads is basically changing.
So obviously they've got ads either side of and in amongst AI snippets, but now they're going to put it into AI mode proper.
So you would ask AI mode, "which running shoes should I buy?" And it would be, well, here are some running shoes — conversational and contextual — and there's a special offer. They did a little mockup and it is going to say "sponsored," but it's part of the AI answer. It's kind of contextual, answering your question but also advertising the thing.
I'm saying, so the advert will look like it's just a response and it won't even look like an advert — they've been trying to hide adverts for a long time. If you were to get people together who are not necessarily Google PhDs and say, right, we're starting to lose advertising market share because of AI mode, what shall we do? The obvious answer would be: just weave adverts into the AI responses. And it seems to me that's exactly what they've done. That's not clever, is it? They haven't come up with a clever solution. They've just done the obvious thing.
But I wonder if more people will just say, well, I'll go and use Claude because they're not putting adverts in the responses. I'm just getting an actual response. ChatGPT are doing it. I don't know about Perplexity and the others.
Perplexity said they never will. You probably will at some point. So yeah, that's one thing. In terms of creating ads, there's going to be a thing called Ask Advisor — rather than just being an AI assistant in ads, it's an agent that's going to have memory. So if you ask it to do something one time, you'll come back another time and it'll remember what you told it before. And it'll go and do stuff rather than just telling you what to do. It's supposed to be helping with planning, asset creation, measurement and optimisation. So everything, basically.
Getting rolled out in North America first, did you say?
Some things are. I'm not sure about all this stuff — I think it's all just getting rolled out and I don't know when. AI Max is taking over. At the moment you've got dynamic search ads where you give them a URL and let them get on with it. They're getting replaced by AI Max because they both kind of do the same thing anyway, but they do look like they're putting a bit more control into AI Max. So you'll be able to tell them certain things they can't do, you'll be able to put in disclaimers, get previews of the assets they're creating, and set certain do's and don'ts. So they have listened to people going, yeah, but what it comes out with is rubbish.
If ever there was a candidate for AI, then it is programmatic advertising, isn't it? If you can educate the AI to what you're trying to achieve and it can go off and do it for you and make the best use of your budget, I can understand why that might work.
You use the AI in HubSpot reports. I've been working on HubSpot reports — it was Friday morning and my brain wasn't working. I said I need a report that does this and this split by this — oh my God. It just does it. It's really good. Very, very happy with that.
I think that's one of the areas of AI. When people talk about AI, they talk a lot about generative AI — it can create this photograph, it can create this video of somebody doing something that person never did, and all that crap. But the way that you can crunch data with it is off the scale. It's phenomenal. You can chuck hundreds of thousands of lines of data at it and say, this is what I'm really interested in, and it'll just come back and give you answers. It's like having your own private geek — somebody that understands Power BI, understands SQL, understands complex queries, and you can talk to them in natural language.
Show all the deals in the dashboard split by the source of the leads, trying to figure out which pulling properties and deals — it's actually quite complicated because there are two completely different parts of HubSpot. And it just did it. And then when it wasn't quite what I wanted, I said, can you change it to do this? And it did. It was taking ages before and I probably wouldn't have got it right anyway. Really smart. Things involving data and reporting — the speed of change is quite staggering.
I was working on something this morning with Claude — it's making me more productive. Not because it's doing loads of crap for me that I didn't want to do, but because it's actually helping me to decipher things and understand things and take things back to first principles. It's no secret that we're big fans of AI. Though I'm not blind to the challenges it's bringing.
No, because I was reading something today — there was something like, if you ask AI a question, it's something like fifty percent more likely to agree with you than your friend would. It doesn't push back as much as a human would.
You can tell it not to do that though. You can set it up to always challenge your thinking. I got it to do something for me last week — a quite nice enquiry came in and I wanted some help to make sure I covered all the points. So I gave it the enquiry and said I want to respond to this, blah blah blah. And the stuff I got back was utter garbage. Complete and utter. I just put it all in the bin and wrote it myself. It was terrible.
I did actually say to it, you know what, forget it, I'm going to write it myself. And it came back and said, oh, can you share it with me once you've done it so I can compare what you did to what I did? So I did. And it basically said, yeah, I was just waffling. You said in three sentences what it took me a whole page to write. So it kind of acknowledged that.
It's quite nice to know that it's not always right. So one of the other things in this Google announcement was they were talking about data strength. You can connect the various data sources direct from Google Ads rather than having to go into Tag Manager. So there's something called Data Manager in Google Ads — you can connect it up with your sources rather than having to go externally. It's trying to bring everything together to keep you in one place.
And then there's Google Tag Gateway, which just seems like another word for server-side tagging. They're like, "oh, we're going to make it simpler and put everything in one place, and we've introduced a new place you've got to go." Great, thanks so much.
But the interesting thing is about causality and attribution. They're introducing something called Attributed Branded Searches — recognising that branded search is actually somebody who found you somewhere else and then searched for you by name. And Qualified Future Conversions, which recognises things like B2B with really long sales cycles. So they're trying to find a way to say, look, if somebody downloads this thing, that's a qualified future conversion — in about eighteen months that's got more chance of turning into a lead than someone who doesn't do this. They're trying to build in predictive data, not just based on conversions now.
And they haven't either. They've got an enterprise version — Analytics 360 or something — with a thing called Meridian. Which is really, really spend that kind of money on analytics territory.
But I think if they're looking at that, it might filter through. It's good that they're thinking about things like qualified future conversions and long sales cycles and trying to go, this is a better lead than that because this is more likely to convert.
And then they're talking about Google Analytics being your data command center, being able to connect with all your different channels — pulling in data from different social media platforms so you can see all your conversions and all your data from within Google Analytics.
We could already do that with Looker Studio because you could connect all the different data sources and display it there. And they're now saying, well, we can connect them to Google Analytics instead. But trying to stay relevant, I guess.
Yeah. They're trying to make Google Analytics the go-to place because there are so many other people offering reporting dashboards and pulling everything together. So I guess they're seeing that they need to compete by getting more and more data in.
And then there's stuff on YouTube. In demand gen and YouTube, they're using search, maps and YouTube signals to predict who wants your brand before they even search.
What does that actually mean? Deciphered into plain English.
That's what I'm trying to figure out. I guess with maps and things, if somebody is in the area and they're looking for something, then they kind of know that they're more likely to buy from you. But yeah, they're trying to gather together all the signals and show people the right ads even if they haven't searched for the thing. If they're on YouTube and they've also been doing other things, these ads are going to be more relevant because they can work out that this is what they're eventually going to do or ask for.
So it sounds like remarketing on stilts, really, isn't it? Which is either going to be great or it's just going to freak people out even more. Showing adverts based on stuff you've looked at and where you are and what it knows you're interested in. Going back to AI — if you pick a particular LLM you use a lot, like Claude, the more you do with it, the more it learns about you. And you can be doing something completely random and it'll say, yeah, and that's particularly relevant for you living in Deeside. So if they're saying the YouTube platform can build up this memory — say Dave's watching a video about Honda Pan European motorbikes. He's just looked on the map for somewhere in France. So maybe he's looking for how to take your motorbike to France. He might need a satnav.
Yeah. So yeah, it's stuff like that. There's also — this is the thing that Leslie was speaking about — in the US they're rolling out Ask YouTube. It's basically a conversational search in YouTube. So you can ask a question and then refine it with follow-up questions to get exactly what you're looking for. It's like an LLM within YouTube. And the amount of natural language data that sits in YouTube is off the scale. They'll be using the transcripts to find the exact answer to someone's question. Mind boggling.
This is a little bit mind boggling. But when it comes down to it, it's still just people trying to get you to buy stuff. Which is, you know, what the internet runs on.
Yeah, basically. And it's just getting more and more complicated. But that comes back to that podcast again — they were talking about how if you're in a large corporation and you come up with an idea, you've got to justify it within an inch of its existence. Whereas in a smaller organisation you can give things a whirl. In a big organisation you've got to show the numbers and prove that if you do this thing it's going to increase sales by at least naught point one percent. Whereas the best ideas are just crazy and random — sometimes the most counterintuitive things work.
And they were talking about things being surprising and standing out. You can kill an idea by using data too much because you're just trying to make it fit into the same box as everyone else. And then everything sounds the same.
Yeah. We had that recently with a new customer where we came up with some nice ideas around a landing page and they just pushed back — oh no, it's against this and we won't be able to do that. And it was all built using their brand guidelines. We weren't doing anything wacky. It's harder to do anything particularly creative if you've got to justify everything with a whole series of numbers. You can't try anything different, which makes it hard to move forward sometimes.
Another Google-y thing — how people use AI mode. This is a US report. AI mode has over one billion monthly users, so it's being adopted. Queries are doubling every quarter. Searches are three times as long as traditional ones. And the number of follow-up queries is increasing by forty percent every month. So people are getting the hang of asking follow-up questions. They're not just using it as a search engine — they're having conversations with it.
My list is to listen to Uncensored CMO. There you go. Third plug for the podcast. And that's what I'm going to do. I do tend to come with stuff I've been doing rather than written-down notes, and I haven't written down any notes from the stuff I've been doing. So I've got nothing to offer.
Fine. It's too hot anyway.
It is a bit. As soon as we finish this, everyone's going home. The sun is literally splitting the trees. I don't think it's as nice the next couple of days, then it's nice again on Thursday and okay at the weekend. Yeah, today is particularly nice.
It's a bank holiday in England as well.
It is a bank holiday. Quite a lot of Scotland take it as well. Kind of mainstream bank holiday but yeah. Okay. Right. We'll leave it there for now. We've got nothing else to say of interest. It's sunny and — Digital Marketing From The Coalface. I promise to try harder next time.
Thank you.