You Can't Automate Guesswork in AI Marketing

Point AI at your marketing and it'll do an astonishing amount of the work. Research the market, write the emails, build the campaigns, segment the list, run the ads — faster and cheaper than any team you could hire. What it won't do is tell you who you're for. And that's the part that decides whether any of the rest works. Because every AI marketing tool works the same way underneath. It takes your understanding of the customer and acts on it — targets them, writes to them, chases them. It doesn't create that understanding. It inherits it. So if your picture of the customer is sharp, AI makes you faster and sharper. If it's vague, AI makes you faster and vaguer. It scales whatever you give it, and most businesses give it a guess. A sound **AI marketing strategy** starts by fixing that. ## Where the understanding actually lives Here's the uncomfortable bit: the understanding marketing needs doesn't live in marketing. Who's genuinely a good fit? Which accounts are worth chasing, and which quietly drain you? Who do you already serve brilliantly, so you could go and find ten more like them? Which clients actually make you money once delivery's accounted for? None of that sits in the marketing function. It sits in the CRM your sales team didn't fill in. It sits in the account notes nobody wrote down. It sits in the profitability breakdown Finance keeps to itself, and the delivery reality only Operations really knows. Marketing can get at some of it alone — website behaviour, search data, support tickets, what people click and open. That's real, and it's useful. But behaviour tells you what people did, not who's worth pursuing or why. You can watch a thousand people visit your site and still have no idea which three were the kind of client you'd build a business around. So when Sales doesn't profile prospects, doesn't record what happened on the call, doesn't know its ideal client or which lookalike companies to go after — marketing is left working from vague characteristics and hope. Industry, company size, job title, a bit of a hunch. Then you point AI at that, and it does exactly what you asked, brilliantly, often to the wrong people. ## Automating the guess This is the trap, and AI makes it easier to fall into, not harder. A vague strategy done by hand at least moves slowly enough that someone might notice it isn't working. Automate it and it stops looking like a guess at all. The output turns crisp — clean dashboards, confident targeting, personalised-looking emails going out at volume. It feels like a system. But it's the same guess as before, now running on rails and harder to question, because who argues with the system that's clearly Doing Marketing? Fast, confident, at scale, and still often pointed at the wrong people — as well as a few right ones. But how do you know which is which? You haven't fixed the problem. You've given it a bigger engine. ## Why everyone ends up doing content There's a reason so much marketing collapses into content — blog posts, LinkedIn, newsletters, the endless feed. Often it's the only lever left. When you don't know precisely who to target or where they are, you can at least make things, put them out, and hope the right people find them. Content is what marketing does when it's flying blind. And that work is worth doing. Good content builds trust, answers real questions, gives people a reason to take you seriously. AI has made it cheaper and faster to produce than ever. But cheaper and faster is exactly where the new danger sits. It's now possible to generate an enormous amount of content while having no clearer idea of who it's for than you did a year ago. Volume starts to feel like a strategy. It isn't. Content is part of a marketing strategy — a useful, hardworking part — but it was never the whole thing, and making more of it doesn't turn a guess into a plan. ## The unglamorous fix So the honest question isn't "which AI marketing tools should we buy?" It's "does this business actually know who it's for, and does anyone capture the data that proves it?" That's a far less exciting project. Getting Sales to fill in the CRM properly and profile who they're actually talking to. Finance being willing to say which clients are profitable and which just look busy. Operations being honest about who you serve well and who you probably shouldn't take on again. Really it means Sales, Finance, Operations and Marketing in one room, pooling what each of them knows — because the full picture of your customer is split across all four, and no single function holds it. It's the same idea behind [AI Doesn't Create Clarity — It Runs on It](/insights/ai-doesnt-create-clarity/): the tool amplifies clarity you already have. ## Where AI actually earns its place None of this means AI sits the process out. This is exactly where it's useful — once it has something real to work with. Give it your account history, your delivery data and your numbers, and it's very good at the analysis a stretched team never finds time for by hand: spotting what your best clients have in common, ranking accounts by fit rather than gut feel, finding lookalike companies you could serve well because you already serve someone just like them. Build that into a proper discovery process — pull the sources together, share the insight, then let AI do the heavy lifting on the analysis — and you've turned a hunch into something you can act on. That's the same argument we make in [You're Sitting on the Customer Data You Keep Trying to Buy](/insights/customer-data-you-already-have/). And the place to start isn't the cold, expensive end of brand-new prospects. It's what you already have: existing clients you could do more for, lapsed ones who already know you, warm prospects sitting half-worked in the CRM. Winning business there costs a fraction of chasing strangers. [Harvard Business Review](https://hbr.org/2014/10/the-value-of-keeping-the-right-customers) has cited figures ranging from five to twenty-five times more to acquire a new customer than to keep an existing one, depending on the study and the sector — the exact multiple is argued over, but the direction isn't. So the tool was never the strategy. But give AI a business that knows itself — one where Sales, Finance and Operations actually feed it, and where you start with the customers you've already earned — and it stops guessing. That's the whole difference: between automation that scales a hunch, and analysis that finally shows you who you're for. Our [AI in Marketing](/ai-marketing/) work is built around exactly that order. ## Frequently asked questions ### Can AI build my marketing strategy for me? Not the part that matters. AI can research, write, segment and run campaigns brilliantly, but it inherits your understanding of the customer — it doesn't create it. A sound AI marketing strategy still depends on you knowing who you're for; AI just executes that knowledge faster. ### Why does AI make a vague strategy more dangerous? Because it hides the vagueness. Done by hand, a weak strategy moves slowly enough that someone notices it isn't working. Automated, it produces clean dashboards and confident targeting that look like a system — so the same guess runs at scale and goes unquestioned. ### Where should we start if our customer data is a mess? Start with what you already own. Pull your CRM, delivery data, finance numbers and support history into one place, then let AI find the patterns among your best and most profitable clients. Begin with existing and lapsed customers before chasing cold prospects. If you're ready to understand where AI can genuinely save you time, reduce friction and create real capacity in your business — without automating a guess — we're happy to talk it through. [Book a discovery call](/contact/) and we'll look at your situation honestly. *Peter Lowe — AI Consultant and Founder, Smart AI Studio.*

This article was written by Peter Lowe. The ideas and opinions are his own; AI was used to assist with drafting and editing.