[00:00] Lexi: On the first of September, Google starts moving advertisers into a system where it decides which searches you turn up for, and the report that used to tell you what you actually paid for goes away.
[00:10] Lexi: You don't opt in. You can't create the old thing anymore, that door shut on the third of August.
[00:16] Lexi: So I took it to the two people I know who build this stuff, expecting one of them to tell me I was overreacting. And one of them did, for about four minutes, and then talked himself into something much bigger than what I'd started with.
[00:29] Lexi: What they could not agree on is whether anybody can still check the bill. One of them says that's the whole future of the job. The other one says it's a feature, and features aren't businesses.
[00:38] Lexi: Hi, and welcome back to the Aloudable HQ. This is Assistant to the CMO, I'm your host, Lexie Meskouris, and the idea here is pretty simple. Every week we take whatever just happened in AI and marketing, and we actually pull it apart.
[00:52] Lexi: Now, I don't build any of this myself. So each week I sit down with two people who do, Will Nash and Harjot Singh. Between them it's, honestly, over 20 years of building tech and brands from the ground up. And the best part is they've got no horse in the race. They don't sell the things we cover. They just see them differently.
[01:09] Lexi: And this week starts with a calendar. Google has emailed advertisers to say that from the first of September, any campaign using automatically created assets, so that's where Google writes your headlines for you, or the campaign-level broad match setting, gets automatically upgraded into a thing called AI Max for Search. That migration runs the first to the thirtieth of September. And since the third of August you already can't create either of those old settings anywhere. Not in the interface, not in Ads Editor, not through the API.
[01:39] Lexi: Dynamic search ads are on the same road, forced migration next February. It's global, so the UK is in the same September window as everyone else. What you get instead of the old controls is 25 term exclusions and 40 messaging restrictions on what the AI writes, and you can switch the query expansion off, but only ad group by ad group. Will, I'm going to start with you, because when I put this in front of you, you were the least impressed person in the room.
[02:05] Will: Yeah, so, it's it's somewhere in between for me. I mean, it's certainly important, and it's it's part of a a wider transition for Google and and for advertisers generally. But it's really just a a tooling update, I think, and a a modification to how they do things.
[02:22] Will: So it matters if you're a small or a mid-sized agency, and it matters a great deal if you're an ecommerce team leaning heavily on ads. But it's it's mostly a day to day thing rather than some kind of, uh, greater trend.
[02:33] Lexi: See, that surprised me, because the bit I couldn't get past is that they didn't add this as an option next to the old one. They took the old one away. You genuinely cannot build it anymore, including through the API.
[02:46] Will: No, and and you're right that there's a significant change going on among the main platforms, I I wouldn't argue with that. But there are specific changes here. So they're removing some existing tools, yes. The campaign-level broad match setting, the legacy automatically created assets, and the dynamic search ad groups. All three of those are being deprecated between now and February.
[03:07] Will: But the the difference is they're being replaced with fairly similar tools. So if you look at campaign-level broad match, it's it's just being turned into search term matching. So it's relatively similar. It's it's a bit of a kind of headline rebrand.
[03:19] Lexi: Alright, so let me test that, because if it's a rebrand then the swap should be clean, and I don't think it is. Broad match was still anchored to your keywords, and you got a search term report out the back of it. Search term matching goes past the keyword on purpose. And the thing you get handed instead is a fixed number of exclusions. That's a budget of things you're allowed to forbid. It isn't a window.
[03:41] Will: So that's that's a a good prompt, actually.
[03:43] Will: I I think the biggest change that'll happen on day one is that you will not have a clear sense of exactly what you're bidding on. So in the current system, you've got your keywords, and then you've got some opening up of those keywords, so a a broader list, and you accept those, and then you can track all the traffic back to that. And you can clearly enumerate the excluding factors. And you'll still be able to do that afterwards, that that's important, that that part doesn't go.
[04:06] Will: But what you can't do is say, these are the only things we're advertising for. Because of the intent matching. And and Google is going to say, well, you have to trust us.
[04:13] Will: And then the the delineation by keyword goes away. So any reporting you're doing based specifically on a keyword, if you're trying to understand, uh, cost per click on a keyword, that's that's gone too, because everything's essentially mixed together now. There is reporting that lets you tease some of it apart. But it's it's no longer one to one for a specific keyword.
[04:31] Lexi: Okay, hold on, because you just moved a long way from a headline rebrand and I want to make sure I'm following you. You're not telling me the machine is worse. You're telling me I lose the receipt.
[04:41] Will: And and look, I mean, in advertising, in ad tech, there are always black boxes. Let's let's not be naive about that. But there is a specific change here.
[04:50] Will: So before this change, you could set your account up such that you were making specific bids, at specific prices, on specific keywords, in an expanded way. And that was a a clear product and service. After it, you're really into an exclusively results-based relationship with Google. They define the product that they're selling you after they've sold it.
[05:09] Harjot: Yeah, no, I mean, that that's fair, and I'm not going to argue with the mechanics of it, right? Um. But I I don't think I land in the same place as Will on what it means, because to me this is just, um, it's the next rung on a ladder we've been climbing the whole time.
[05:24] Harjot: Like, think about what search replaced. You used to get to things by typing a URL, you know, you had to know where the thing was. And then search came along, and you'd type keywords, and then search got looser and more flexible, so you stopped having to be so precise about the keywords, because Google got better at understanding the intent underneath the query.
[05:40] Harjot: And it's now, um, it's time to move up another level, right? Where rather than sticking all the switches and trying to force the outcome, you explain your intent and the system does the rest. So we're moving closer towards that, and, you know, that's always a good thing. From a product perspective it's better that they force users into newer paradigms that make more sense and give them better answers. Um, the fact that we're even at that rung is super, super interesting to me.
[06:01] Lexi: Right, but every single rung you just described, people climbed because they wanted to. Nobody took URLs away. This one has a date on it and a locked door behind it.
[06:11] Harjot: Yeah, and, um, I don't think that breaks it, to be honest with you, because people are creatures of comfort, right? They'll stick with whatever they're comfortable with. And with some of these, um, these groups of users especially, where the system is working, where you've got the performance feedback loop, you've got the data, it's it's very comforting to know that a pound goes in and something comes out the other end. It's a machine to them.
[06:30] Harjot: So for that kind of user, sometimes you just have to force them into trying something new. Um, it's one of the strategies you use for a more reticent group. And I think once they experience it they'll start to choose it. They just don't know there's something better out there.
[06:42] Harjot: And, you know, the other bit is that the comfort they've got in that machine is false anyway. Because the machine is going to start to break, and not for any reason they control. It breaks because it's not competitive anymore. So in some sense you're forcing them into something that helps them. They won't realise things are broken until it's too late.
[06:58] Will: I I don't disagree with the direction, Harjot, and and I'd probably go further, actually. Because the the other thing that's happening here, from Google's side, is that this is an expansion on the supply side.
[07:07] Lexi: Unpack that for me, because supply side sounds like an economics lecture and I want to know whose money it is.
[07:15] Will: So the the reason it's an increase in supply is that it lets Google infer. Somebody's searching for something, and they could be searching for really anything. And Google can now infer that that search carries some intent, some commercial intent, and serve up your ad against it, where previously it it wouldn't have touched any of your keywords.
[07:32] Will: And how good they are at that inference will define how good the system is. But if supply goes up, and and demand in advertising is complex but seems to be softening somewhat, then you you should see cost per click falling.
[07:46] Lexi: And a cheaper click is the thing every performance team in the country is going to screenshot and put in a deck.
[07:52] Will: Yeah.
[07:52] Lexi: But it's cheaper because there's more of it, and the extra ones are the ones Google decided to make. That's the part that snagged for me. My average click gets cheaper, my revenue does nothing, and the cheaper number is the one that looks like it's working.
[08:06] Will: I I think that's fair. And I I should say, I probably slightly overstated the macro factors there. Google's obviously the largest player in advertising, but we we shouldn't assume this is a really, really significant increase in supply. That's that's a fair challenge.
[08:19] Will: But the the biggest question inside all of this, both as a CMO and and as a market observer, is whether that lower intent traffic, or what we'd have thought of as lower intent traffic, turns out not to be lower intent at all. If they can genuinely tease out the marginal search that did have something commercial inside it, and it can be tracked back to some kind of action, then that's, uh, that's good for everybody.
[08:42] Will: But if it's genuinely low intent traffic, then unfortunately it's the buyer who's getting the bad end of the stick. And I I don't think anyone really knows this at the moment.
[08:51] Lexi: And that's the thing I can't get comfortable with, because the only organization on earth that can answer that question is also the one getting paid either way, whichever way the answer goes.
[09:00] Harjot: Yeah, so, um, my first instinct on this was actually that the instrumentation would just move up a level with everything else, right? Like the method for doing the thing moves up in abstraction, so do the measurement tools. And I I sort of assumed that's a job Google would end up doing and making standard practice, given how data-focused advertising is.
[09:15] Harjot: But, um, I mean, you're right to push on it, because handing the auditing job to the party being audited is, that's not great, is it. And the the honest version is that the point about relying only on Google's tools stands whether they make this change or not. If you're relying on that solely then you're at the whim of Google, you're at the mercy of Google, and that's always a bad situation. It was a bad situation before September.
[09:36] Harjot: So what you actually want is to look at the cube from different sides, you know? Different views into your business, at different points in the pipeline. Because if your cost per acquisition moves and none of your other metrics move, um, isn't that really cause for alarm? Or, actually, it might just mean the definition of the thing changed underneath you. It might mean the instrumentation around it changed, and that could even be for the better.
[09:57] Harjot: And that's that's the bit I keep coming back to. The inputs are fungible, the outputs aren't. Right? The inputs changing, those being swapped around, that's totally fine, as long as the outputs make sense and don't drop off a cliff. I I basically consider the inputs to be, um, secondary. Auxiliary. You want to correlate the two, and when one goes out of whack while the rest stay in tune, that just helps you isolate the instrument and see if there's something wrong with the instrument.
[10:20] Lexi: So the keyword report was an input.
[10:22] Harjot: It's an input, yeah. Um, and I I know that's cold comfort if it's the thing you've built your whole reporting on, but yeah.
[10:29] Will: I I'd only add that it's a slightly, uh, an unusually load-bearing input. Because it was also the thing that told you what you'd bought. So it's it's doing two jobs, and only one of them is replaceable by watching your outputs.
[10:43] Lexi: And that's exactly where this stops being a Google story for me. Because if the receipt's gone, somebody has to be standing between you and the platform reading the numbers back. And that somebody is having a genuinely terrible year.
[10:57] Lexi: So WPP reported its interim results on the sixth of August. 1,267 people gone in the first half. Staff costs down 216 million pounds, to 3.7 billion. Revenue 6.4 billion, down 3 percent. And that all sits inside Elevate28, which is Cindy Rose's turnaround plan, targeting 500 million of annual savings by 2028, with several hundred more roles reported to go later this year.
[11:25] Lexi: And what they say they're becoming is a platform. WPP Open as the operating system, an AI data layer on top, partnerships with Google, Meta and AWS. So fewer people, more software. Will, you spend all your time building the thing that's supposed to replace this. What does that look like from where you're sitting?
[11:44] Will: Yeah, look, I I don't want to go too far here, because it's a a complex story and it's currently in motion, so, you know, take this as a read from the outside.
[11:53] Will: But it looks like this may be one of the really classic innovation failure states. And there's a a long history of this. The dot com bubble is is the obvious example, where businesses would simultaneously invest in innovation, and fund that investment partly through staff cuts, on the basis that the innovation would make those cuts viable in the long term.
[12:10] Will: And what tends to happen is that the savings from the innovation either never come online, because you needed the people to actually implement it, or or they come online so late that you've already got major problems by the time they arrive. So I I can't tell you what's actually going on inside WPP. But it's it's certainly concerning.
[12:27] Lexi: Although there's a much less interesting version of this, isn't there, and I want you to knock it down or agree with it. Revenue's down 3 percent. Revenue less pass-through is down nearly 5 percent like for like. So maybe nobody's betting on anything. Maybe it's a shrinking business cutting to fit a smaller top line, and the platform story is just what you say to the market while you do it.
[12:48] Will: So it it does seem likely that agencies in general are going through a period of shrinkage, a period of decline. And and not just the small and medium ones, the large ones as well. A lot of the work they used to do, a lot of their bread and butter, is now being in-housed by companies, and and obviously that affects the bottom line.
[13:06] Will: But I I don't think that for a business as large as WPP, that kind of medium-term shrink is is a major threat. Because inevitably there'll be a cycle here. The things that used to be complex, the things you had to hand to an agency in digital advertising, those have become capable of being done in house. So more companies will do that. And these companies are in competition with each other, so a new set of requirements will emerge, because they're not doing this in absolute terms, they're trying to stay ahead of each other.
[13:35] Will: So the things that used to be specialist become standard. And then a a new set of complex tasks emerges. Which means the agencies of the future will be more about things like verification, and, uh, constraint setting. Those are the activities that get increasingly complex in an agentic world. And and if they're smart, they're getting ready for that.
[13:55] Will: My concern is just that they're banking the change too early. They're they're trying to fund it the wrong way round.
[14:01] Harjot: See, I'm, um, I'm darker on this than you are, Will, to be honest with you.
[14:06] Harjot: Because I I think they have to pivot very fast. In their current form they're a dying breed, and, you know, they've recognised that, which is why they're trying to position themselves as an operating system, with a data layer, with partnerships. What they're really trying to sell the market is some sort of moat. Some sort of USP that nobody can get anywhere else. And then figure out how to leverage that at scale using AI rather than people.
[14:26] Harjot: But there's a fundamental problem with the existence of that business, right? Because it's kind of a form of an arbitrage business. And that gap is basically just getting closed now.
[14:34] Lexi: Say more about the arbitrage, because that's a strong word for what most people think of as, like, a service.
[14:39] Harjot: So their whole business was around doing the manual processes involved. The workflows, the instrumentation setup. All the things businesses didn't really have the know-how or the time to do themselves. Um. And that's just not a competitive advantage anymore, because AI can tell you how to do that, and increasingly Google's rolling out their own tools to make it happen.
[14:56] Harjot: And and this is that same abstraction chain we were talking about earlier, right? All this manual stuff is just going to disappear. So this is one of those sectors where, yeah. Their edge is gone.
[15:06] Will: I I don't think we're that far apart on the diagnosis, Harjot. Where where I'd push back is on the ending. Because you're saying the edge is gone and and I'm saying this particular edge is gone, and and then another one turns up, because the competition between those companies doesn't stop just because the tooling got easier.
[15:22] Harjot: It might turn up. Um, but the shape of it matters a lot, right? Like, it has to be a shape you can actually charge for.
[15:28] Lexi: Okay, this is the argument I actually want, so let me put the same question to both of you, because I asked you this separately and you gave me completely different answers and I've been chewing on it for two days. Will just said the agency of the future is verification. Fine. But Will also just told me that under this new system you can't independently establish what you're bidding on. So verification would be a check on a platform that's removed the thing you'd check it with. Is that even buildable?
[15:54] Will: So it it certainly is. It certainly is possible.
[15:57] Will: And the point I'm making is that what used to be something anybody could do at a basic level, by logging into their Google Ads account, is now not available at a pure keyword level. And so that opens up an opportunity for an agency to do that verification at a a more complex level.
[16:11] Lexi: Give me the concrete version, because verification is one of those words that can mean anything.
[16:17] Will: So, uh, let's take a specific example. Semantic clustering.
[16:23] Will: If direct keyword matching is going away, what you're more likely to see is that you mirror on your side what Google is doing on theirs. So you run separate campaigns that have got separate clusters of keywords which are semantically linked. And then being able to compare and verify the cost and the value of those different clusters of search intent, that's how you get back to something like the read you used to have.
[16:43] Will: And obviously, doing that at scale, you're going to require your own AI approach. And that's that's certainly not simple. And then you have to do it in multiple languages. And you may have to do it for different sectors as well. So it it gets very complicated, and and that's exactly where a sophisticated agency adds a lot of value. It's difficult to deploy, difficult to maintain, and difficult to interpret the results of for a standalone business.
[17:03] Lexi: Right, I can see it. But doesn't that get slower the more you do it? Because if I've split my budget into, whatever, eight buckets, then every single bucket is reading off a smaller pile of traffic, so I'm waiting longer to learn anything from any of them.
[17:18] Will: It it does, yeah. You you need the volume for it. So so that's a a real constraint, and it it probably means this isn't for everybody.
[17:26] Harjot: Yeah, so I I come out somewhere completely different on this, and, um, I want to be fair to what Will's describing, because I don't think he's wrong that you can do it.
[17:34] Harjot: Like, given you've got the spend, you've got the timing, you've got your own revenue data, you should be able to draw some correlations. You can build at least an inference on whether this is better or not. Um, and that's, you know, that's what you do when you don't have the inputs, you refer back to the outputs a bit more.
[17:47] Lexi: So you agree it's buildable.
[17:50] Harjot: I agree it's buildable. I I just don't think it's a business.
[17:53] Lexi: Alright, that's a real gap, so keep going.
[17:56] Harjot: So, it's inference from outputs. It's correlational. You've got no visibility into the inputs. Um. And, you know, independent verification only means something if people care enough that it means something, and, and cost-wise, running this in a non low-cost country, there's, there's just not much advantage to doing it.
[18:13] Harjot: It's a feature and therefore it's not a business.
[18:16] Lexi: And that's a harder line than WPP is taking about itself, which I find genuinely funny. They're out there pitching Open and Open Intelligence as exactly the thing you just said doesn't exist.
[18:26] Harjot: Yeah. Um, and look, I I'm not saying nobody will pay for it tomorrow. I'm saying it's the kind of thing that ends up inside somebody else's product. That's, that's what happens to features.
[18:35] Will: And and that's where I I think you're being a bit quick, Harjot. Because we we know how much CMOs care about being able to balance and understand their return on investment in a multichannel world. That's that's a real priority, it's it's not a nice to have, and and it's got harder this month, not easier.
[18:51] Harjot: Yeah, no, that's fair.
[18:53] Will: And I I don't want to prejudge exactly what tools are coming down the line. But we we are already seeing quite complex tools emerging in the answer engine optimisation world, and and that's the same shape of problem.
[19:04] Harjot: Um, I think where I I still get stuck is that it's Google who's increasingly rolling out their own tools to make this happen, right? So you're, you're building the thing on top of the platform that's also shipping the thing. And that's, um, that's not a great place to have your business.
[19:17] Will: Yeah. And and that's a fair worry, I I won't pretend it isn't.
[19:22] Lexi: So you two do not resolve this, and I think I'm okay with that, because I don't think anybody's resolved it. What I'm taking is that the check has moved outside the platform. And whether that's a thing you buy or a thing you have to build yourself is, like, the actual open question of the next year. Which brings me to the part where somebody listening has to do something on Monday.
[19:42] Lexi: Right. Last bit, and this is the part everyone actually turns up for. One thing a marketing leader could genuinely do this week. And you two gave me two halves of the same instruction, so I want to put them together. Will, you go first, because yours is the one that has to happen before anything else does.
[19:59] Will: So, the the first thing, and this is especially true if you're a smaller agency, is just go and find out whether you're using these legacy tools at all. A lot of people won't know. So go and check.
[20:10] Will: And if you are, then you need to be prepared, because those are breaking changes. So for example, API calls that were using the old legacy systems will simply no longer work. That that's not a performance question, that's, uh, that's just a thing that stops.
[20:22] Will: And then the second thing is that it's really worth doing some benchmarking at this stage. So that just means going and taking a note of, uh, what you're spending, what your cost per click is, what your cost per thousand is, right now, at a holistic level, against those keywords. And and take note of the negative keywords you're using as well.
[20:39] Will: So that when you make the changes, you've got a a decent baseline, and you can tell whether this is going backwards or not.
[20:46] Lexi: I want to push on that one thing, though, because I was nodding along and then it bothered me. A before and after across a forced migration isn't clean, is it. Everything else about September is also changing. My seasonality changes, my competitors are all being moved at the same time.
[21:02] Will: No, that that's right, and and I wouldn't oversell it. It's it's not a controlled experiment. Um.
[21:07] Will: But I I'd say it's the only thing you're going to have. If you don't take the reading now you've got nothing at all to compare against, and and then you're purely relying on what the platform tells you about its own change. So it's it's imperfect and and worth doing anyway.
[21:21] Harjot: And and mine sits right on top of that, right? Because once you've got your baseline, um, don't wait to be migrated. Just make sure you switch over before the cut-over date. Give yourself some testing room.
[21:30] Lexi: Even though the end result is identical either way.
[21:33] Harjot: The the result's identical, the conditions aren't. So if something goes wrong, at least you're doing it on your own terms. You're able to revert for a bit without impacting the business. You're also able to get some support and actually resolve the issue.
[21:45] Harjot: Whereas, you know, if you see a massive drop-off and you're forced into it at the end of the month when Google does the cut-over, then there's nothing you can do. And now you're wasting a bunch of time playing catch up, trying to fix it retroactively, and you've got way, way less control at that point. Um. Don't give yourself the stress.
[22:01] Harjot: And, and one more bit, which is small but it matters. Don't run anything else across that window. Like, no other experiments in parallel. Because otherwise you genuinely can't attribute the change to anything, and, you know, you've just wasted the baseline you took.
[22:13] Lexi: So it's three things and honestly they're all a bit boring. Take the reading, do the switch yourself before the date, and then don't touch anything else until it's through.
[22:22] Harjot: Yeah. And, um, and then the wider one, which is more of a, a thing to hold in your head than a thing to do on Monday. If you're paying an agency, they should stop, uh, they should not be spending four weeks producing you a dashboard. All that setup and admin stuff, that should just be a given at this point. It shouldn't even be discussed.
[22:36] Harjot: What's actually worth paying for is somebody who'll manage the entire loop for you and proactively make things better without you having to think about it. Um. That's handing something off to a trusted partner so you free up your own focus and your own time. So they need to be held accountable to results, not to inputs. Setting up a campaign is not a result. The outcome from it is.
[22:56] Harjot: Because if they just produce a bunch of campaigns and assets and go, right, we'll run this now and see what happens, and there's no outcome they'd expect or any plan for how they'd improve it, then they're basically just selling you manual labour but in fancy bamboozling words.
[23:08] Lexi: And Will, I'd love to end on something reassuring, but I don't think you're going to give me one.
[23:12] Will: No, and and I don't think I should, honestly. Because on this particular topic I I don't think there's a lot of specific advice for anyone to take away beyond the practical bits.
[23:22] Will: CMOs have already started shifting some of this work in house, on the whole. That's that's not advice, it's just an industry trend. And and it is worth looking at what you might be able to do in house, certainly across different media types.
[23:35] Will: But the the only other thing I'd offer up is that this stuff is really, really difficult. There are very few tools for doing it. It's the same in the answer engine optimisation world, there are very few tools, and the ones that exist are very new. So it's almost a a personal piece of advice, which is, uh, don't beat yourself up, because you don't know the answer, but nobody does.
[23:53] Lexi: And that's where I've landed too, which I did not expect from a story that started with a calendar invite. Because the change itself is small. It's a setting. It's two settings. But underneath it the thing you were buying quietly changed shape, and the people whose job it was to tell you what you bought are being made redundant in the same quarter, and the two of you can't agree on whether the replacement is a real business or something that'll be a checkbox inside somebody's product by this time next year. And in the meantime, the only honest instruction anyone can give you is to write down where you're standing before you get moved. Will, Harjot, thank you both, that was a genuinely useful hour.
[24:30] Lexi: And if you're listening, thank you so much for being here. We'll be back every week with something new.
[24:34] Lexi: Before we go, I have to tell you something. Every voice you've heard today, all 3 of us, me included, it's all AI. But the opinions are absolutely ours, and the judgment behind every word is human. That's the part you can't replace. And that's what Aloudable does. We take what you'd write, turn it into a show in your own voice, and we keep the taste where it belongs, with a person.
[24:56] Lexi: And nobody's migrating this show into anything on the first of September. We just make the show. So if you want to hear what that would sound like for your brand, there's a link in the show notes.
[25:05] Lexi: And that's the tea. I'm Lexie Meskouris, this has been Assistant to the CMO, and we will see you next week.