Bait
There were warning signs, I suppose, in the job title. AI and Growth Marketing Manager, which is two disciplines in one, though I didn’t think much of it at the time because the description sounded familiar: integrate language models into the CRM stack; build test-and-learn workflows; own the logic for real-time content assembly; act as the internal SME for a team that hadn’t done any of this before. Four responsibilities, each of which I’ve done somewhere, which is the feeling these descriptions are engineered to produce.
The recruiter, warm and efficient and just back from Ibiza, described the person they wanted as a marketing engineer, and I thought yes, that’s a real thing, I’ve been that. The salary came at the end of the call, as it always does, and it was eighty to ninety thousand pounds for a role in London, barely above the average for London, requiring two days a week in the office; and when I gave her my preferred range she said she’d check whether there was flexibility. What arrived instead, the very next day, was an email claiming the role actually required someone who “independently navigates Snowflake and BigQuery, writes complex SQL, manages production language model edge cases, and builds data-piping solutions for real-time product engines”, basically word salad describing stuff I’ve already done, none of which appeared in the JD I’d applied to or in the forty minutes we’d spent on the phone.
Cycles
I don’t say this to complain about one company, which would be tedious, and anyway this isn’t new. It ran in 2020, when I’d open a listing and find four jobs stapled together with a salary that wouldn’t have covered one of them in the city they expected you to commute into daily (and then work remotely from). It ran again in 2023, and now again in 2026 with the AI craze, and the three-year rhythm is regular enough that I’m curious about the mechanics. Earlier this year I came across a listing for technical writing, UX writing, and localisation, three crafts with their own literature and professional bodies and conference circuits, offered at thirty thousand pounds in London, which is less than a single one of those roles commands and roughly what you’d need to not quite afford a flat in zone four.
The cruelty of it, and I use that word deliberately, is in who it catches. Someone underqualified reads a JD like that and closes the tab, because they know they can’t do it and the filter has done its job. But someone who’s spent fifteen years accumulating precisely that combination of skills, who studied computer science and informatics and taught database courses and wrote for engineers and then spent a decade in growth marketing and learned to orchestrate language models because the tools arrived and seemed obviously useful, that person reads the same four-job description and thinks, with something close to relief, that finally here’s a role that wants all of it. Pick me. That flare of recognition is what the description is fishing for, and it most reliably snags the ones most capable of doing the work and therefore most valuable to underpay.
Everybody loses
What happens next is predictable. The person takes the job, because they’ve been looking for months and the description spoke to them, and for the first quarter they’re brilliant, but then they start context-switching between four disciplines every day and getting good at none of them, and quality declines, and sick days accumulate, and eventually another company with a properly scoped role and salary makes a call and they take it, because of course they do, and the requisition reopens and all the institutional knowledge walks out the door. Everyone else on the team has spent nine months watching their colleague drown in a badly scoped job and has drawn their own conclusions about how the place treats people. The company has spent that time paying for a solution it didn’t get, and now it’s paying agency fees again, and the entire arrangement has cost more than simply hiring the right number of people at the right price in the first place.
Unicorns
What seems to me more fundamental than the economics is that unicorns shouldn’t need to exist. We have horses, we have rhinos, we have narwhals, which are magnificent and strange and perfectly adapted to swimming under Arctic ice. A creature combining all three isn’t an upgrade on any of them, and not even coherent as a design, because consider the practical difficulty of a horned horse lowering its head to graze: the horn goes straight into the dirt. You’ve taken three functional animals and produced one that can’t feed itself. The reason unicorns are mythological is because myth is the only place they work.
Companies build job descriptions the same way, taking three coherent roles with individual shapes and market rates and fusing them into something that reads impressively but can’t be performed by a single person. Then they price the composite below the market rate for its cheapest component, and then, when a candidate points out the mismatch, they discover they wanted something else entirely all along, something they’d neglected to mention, which is a graceful way of saying they never scoped the job and were hoping you’d show up and solve that problem and not ask too many questions.
So why do they keep doing it? Because it works, sometimes, not always, and not for long, but long enough that a manager can get through a budget cycle, finance can maintain a low salary line, and the business can call an underfunded backlog a hiring plan. Some companies do this because they haven’t thought hard enough about the work, but others know precisely what they’re doing. They know it damages morale, slows hiring, raises replacement costs, and burns through good people, but they also know the model can absorb the damage. Agencies, consultancies, and high-churn operators have built whole machines around this math of extracting nine months of output from one unusually capable person, replacing them when they tire, and keeping the margin. They don’t always need the arrangement to be healthy.
So here’s some advice for anyone reading a description and feeling the role is asking for a rare combination you just so happen to be: that feeling is a warning. You shouldn’t have to be a statistical anomaly to be employable, and if a company has written a role only an anomaly could fill and then priced it below market for a single discipline, they’ve just told you a great deal about how they think, what they understand about the work, and how they intend to close the gap between what they need and what they’re willing to pay, which is with your evenings. Read the scope, read the number, and if the two don’t match, the number is the honest one.