A PM I've worked with for years forwarded me an article last week with a one-line note: "Have you thought about repositioning?" The piece was the usual 2026 sermon — the market's a barbell now, machine translation ate the cheap end, so climb toward advisory work, transcreation, direct clients, retainers. Sensible enough. Also almost entirely irrelevant to what I do.
I've read a dozen versions of this advice this year. All of it assumes a market I don't live in.
The barbell has no weight on my end
Here's the picture the industry keeps painting. General translation runs $0.08 to $0.15 a word. Specialist work — legal, medical, technical — climbs to $0.15 to $0.30, and true subject-matter experts push past that. MTPE has settled into a third tier underneath everything, $0.05 to $0.15, usually 60 to 70 percent of a full rate. The middle's supposedly being hollowed out. The move, we're told, is to run for the top before the machine catches you.
That story makes sense if you're one of ten thousand English-to-Spanish generalists watching DeepL undercut you. There's a bottom to fall out of because there's a crowd standing on it.
Turkmen doesn't have a crowd. On most projects I'm the only vendor the agency has, and they know it. There's no race to the bottom because there's no one to race. The commodity tier the articles warn about never formed for my pair — the machine output isn't cheap-but-usable, it's fluent-and-wrong, confident in exactly the places it's mistaken. You can't build a $0.05 tier on a draft that needs rebuilding. So the pressure everyone describes, the one crushing generalists in big languages, mostly doesn't reach me.
That sounds like good news. It isn't, exactly. The same scarcity that protects my rate also strips out every other lever the advice assumes I have.
Direct clients and retainers are a fantasy at my volume
The headline number people love: direct clients pay 30 to 60 percent more than agency subcontracting for the same pair. Senior freelancers supposedly pull 60 to 75 percent of income from repeat direct clients and retainers — three to ten thousand a month for steady high-volume work.
Name me the company with enough steady Turkmen to justify a retainer. I'll wait.
The Turkmen work of the world is lumpy and sporadic. A government tender package here, an oil-and-gas compliance batch there, a software UI string dump, a stretch of nothing, then three legal files landing the same afternoon. No single end client generates the recurring volume that makes a retainer rational for either side. And the direct-client premium assumes the client can evaluate the work and manage the process. A firm needing Turkmen once a quarter can do neither — that's precisely why the agency exists. The agency isn't a middleman skimming my margin. For my pair it's the only structure that makes the work happen at all.
So when the advice says "go direct to escape the agency haircut," it's telling me to abandon the one channel that reliably routes work to me. No thanks.
What actually moves the number
Strip out tier-climbing and direct-client hunting, and what's left? Two things, and neither is glamorous.
Scarcity is real leverage, but only if you're the version of scarce that's also easy to work with. I'm not the cheapest Turkmen option — I might be the only serious one — but that means nothing if I'm slow to reply or I hand back a file that breaks their QA. My rate holds because I'm the node the PM never has to think about. Segmented XLIFFs come back clean. The Latin-to-Cyrillic script question gets asked before delivery, not after. When their automated checker screams about false positives in Turkmen morphology, I explain it instead of ignoring it. That reliability is worth more to a good PM than shaving two cents, and they know it. Being irreplaceable and being pleasant are two different skills, and you need both.
The second thing is refusing the discount, and meaning it. Roughly half of surveyed translators won't cut rates for MTPE, and they're right — post-editing a bad draft often eats as much time as translating clean. For Turkmen it's worse. Cleaning up fluent-and-wrong output means reading every line with more suspicion than I'd read a blank page, because the errors are camouflaged as competence. I don't offer 60 percent for that. I offer my rate, and if the file is genuinely light, I'll say so myself — but the machine doesn't get to set my price.
The positioning advice isn't wrong. It's just addressed to someone standing in a stadium. I'm standing in a room with the door shut and my name on it. Different math.
The repositioning I'm doing isn't upward, toward strategy decks and cultural-assessment invoices. It's sideways — deeper into the handful of agencies that already know what I'm worth, making myself the one they'd fight to keep. That's not a growth hack. It's just the only market I've got, worked properly.