Mark Ritson: Digital marketing's endgame was always automation. Strategy matters again

Google's AI Max migration takes another set of controls out of marketers' hands. Mark Ritson thinks that could ultimately be good news. The time of the dashboard marketer is over? (Adobe Stock) On September 1, Google started moving the digital furniture. Every Search campaign still running campaign-level broad match or automatically created assets was upgraded to AI Max, whether the account owner read the email or not. The migration will run all month, and there is no opt-out. The only option to switch these features off before the door closes. And why would you? Anyone who wants to appear in Google's new AI search ad formats, the conversational discovery ads and highlighted answers it unveiled in May, needs AI Max or Performance Max to get to the party. Cue outrage from the usually pliant performance community. But from Mountain View, it makes perfect sense. Google already has hundreds of thousands of advertisers on AI Max. It learned from the Performance Max black box and this time built in query-level reporting, brand controls and text guidelines before it pulled the lever. It has a number, an average of 7% more conversions at a similar CPA, much more modest than the 14% it claimed at launch and therefore more believable. And it has correctly worked out that the median search advertiser is not a digital savant who stays up all night running spreadsheets and hand-tuning match types. Say your keyword is 'running shoes.' Exact match means the ad only shows for searches that are basically 'running shoes' and nothing else. Phrase match means the search has to contain that phrase, so 'best running shoes for flat feet' qualifies. Broad match means Google can show your ad for anything it thinks is related, like 'trainers,' 'marathon gear,' or 'how to stop shin splints.' Advertisement Tighter means less waste, broader means more reach, and managing that trade-off, keyword by keyword, was a big chunk of what a search specialist did all day. AI Max takes the dial out of their hands and turns it itself. To be fair, most of this was highly automated already. AI Max now also writes the headline and picks the landing page. So a whole suite of joysticks and dials is now, essentially, absorbed into the system and run on autopilot because it's a better way to fly the plane. That said, the independent data is messier. It usually is. One analysis of more than 250 retail campaigns found AI Max conversions delivering roughly 35% lower ROAS than traditional match types. Lunio found 72% more invalid traffic on AI Max campaigns. It's all about client averages. So the truth is that automation wins for most and loses for the select few who actually know what they're doing. And it is September 2026, the Jurassic era in terms of where this tech will be by Christmas. Or, God help us, 2028. For a platform selling to the many, that is an obvious and immediate upgrade. Not least because everyone else is doing the same. Microsoft is folding Predictive Matching into its own version of AI Max. Meta has been at it longer and harder. Advantage+ has become increasingly central to its campaign setup, while advertisers have seen more placements and controls absorbed into automated systems. Mark Zuckerberg has been Zuckerberg-clear about the destination: a business should tell Meta its objective and increasingly let the platform handle the rest. Creative. Targeting. Better outcomes. Three platforms, one summer, all removing the same manual dials and levers and describing it as an upgrade. Because it is one. In most senses of the word. The only people likely to dissent are digital marketers. For 20 years, the fastest-growing job in marketing has been the person who sets the bids, picks the placements, builds the match-type architecture and reads the search terms report on a Monday morning. The digital marketing role is now becoming fully digital. Done by the platform, on the platform's terms, with the platform's data. A closed loop that excludes the weakest link in the system: us. Want to go deeper? Ask The Drum Agencies have already baked this into their corporate evolution. At least 18,000 roles have gone across Omnicom, Dentsu and IPG in the past 18 months, while WPP alone has cut thousands more. I have already argued that AI is helping drive a 'Great Stay' in marketing, with nervous employees clinging to roles in an increasingly hostile jobs market. I have also warned that marketers with a decent radar are unlikely to leave their roles voluntarily in 2026. Clients are under growing pressure to move money away from overhead and toward working media. More funds for the platforms that increasingly plan it for them. There has been a lot of unsubstantiated BS and idle estimation about the impact of AI on marketing jobs. Most of it from me. But this is that general story told with a very specific, brutal exemplar. AI is coming for the mid-level digital tactician, the performance lead 12 years into a career whose entire craft was mastery of a dashboard that has just been simplified out of existence. And that hurts more than it should, because for a large chunk of the profession, digital tactics is not part of the marketing job. It is the marketing job. Somewhere in the last two decades, marketing quietly redefined itself as the department that runs ads. Mostly just digital ads. It would be easy to argue this will send marketers back to a broader, more strategic approach. Marketing like it's 1995. But most of the digital marketers losing their roles were all digital and no marketing in reality. Take the ads away and the question is what, exactly, is left. At the disciplinary level, there is some hope that the hinterland of marketing can survive the AI incursion. The coastline has gone. The ports, the harbors, the bits where the money lands, all now belong to the platforms. But the interior, the territory many never bothered to visit because it was harder to reach and slower to pay out, is still ours. These parts are not evenly weighted, though, so let us rank them, from least important to most. At the bottom, digital communications. Bidding, matching, placements, creative rotation, budget pacing. Gone, or going. Fight it if you like. Advertisement One up from that, the strategic inputs to communications. If you believe Zuckerberg, and he will make it true even if it's a lie right now, AI needs an objective to work from, and that's it. But that is the communications version of strategy, the thinnest form. The machine can find the people most likely to click and the content that makes them do it. It cannot tell you which of the four segments in your market to build the company around for the next five years, or who will pay more, or whether this year's objective should be penetration or price, or what the brand should mean in the heads of the people who never click anything. Before your communications strategy, your targeting, messaging and communications objectives, there was always meant to be a higher level of strategy. The overall brand positioning, the targeting strategy for the whole market, the handful of strategic objectives for the year ahead. These boxes aren't doing that. They aren't meant to. We are. Above that, product. Marketing has spent 20 years defaulting product to somebody else. R&D. The founder. The category team. The product management team. Nobody in Mountain View or Menlo Park has an algorithm that decides what to make, at what quality, in what range. The fastest way to amazing search results was always to sell a better thing. That is still true, and the machine has just made this lever more important. But only for marketers who actually know what product marketing is. And I'm not talking about that weird tribe of self-trained newbies who think the 'Ideal Customer Profile' is somehow a good thing. Above that, price. Nothing moves the P&L faster, and nothing in marketing gets less marketing input. Smart bidding does not set your price. It sets the price of a click. This is literally 1% as important as the actual price Marketers who understand elasticity, price architecture and the difference between a discount and a promotion are rarer and worth more than ever, precisely because nobody is automating them. Yes, I know your company does not let you control the price. I agree with that decision. But you should have a seat at the table. Well, again, provided you actually have proper marketing training that includes pricing. Because with that, plus some conjoint or a maxdiff and a plan brewed from proper anchoring theory about what the company should do, you are the most valuable ass in any chair at that meeting. Just because your dumb-ass company has been setting prices on a combination of wet fingers, variable costs and inertia doesn't mean it should. Imagine basing price on what customers are prepared to pay? Mind-blowing stuff. Then there is, whisper this part, all the communications tools that are not 'digital.' I need the quotation marks because everything these days is digital. From outdoor to radio, it's all digitally delivered. But some of it, for now, is not digitally planned the way search is. Television, outdoor, sponsorship, packaging, the store, the sales force, PR. Programmatic is nibbling at the edges. But planning a two-year sponsorship or a pack redesign remains stubbornly manual, requires judgment about things the platforms cannot see, and is where a disproportionate share of the long-term brand effect still comes from. Advertisement And at the top of the post-AI marketing pyramid, we get to managing. The verb, not the noun. Management implies gray-haired women in power suits on the top floor. I mean something more prosaic and quotidian. Someone has to take the objectives, the target, the product, the price, the automated comms and the manual comms and make one plan out of them. Budget them. Present them. Review them against the quarterly numbers. Find slush funds when the shit hits the fan. Or the sales force goes rogue in Taiwan. Platforms have automated the buying. They have not automated the deciding, the doing or the managing, and they have no intention of trying. It's too hard. And there's no money in it. I'd say Google has done the profession a favor, though it may not feel like it for a while. It has taken away the thing marketers were doing instead of marketing. And left us with a job to do. Fortunately, Mark Ritson does not teach search as part of his MiniMBA in Marketing, which kicks off this week for the last time this year. But he does teach targeting, positioning, DBAs, strategy, objective setting, product development, pricing and a host of other shit that is not, currently, being replaced by AI. Not too late to sign up at www.minimba.com. Read more opinion from Mark.