The candidate had a beautiful resume. Ex-big-tech, ex-consultancy, a slide deck about "transformation" so polished it squeaked. The company hired him in three weeks, and by week nine they had a forty-page assessment of their problems, zero working software, and an operations team that had stopped returning his messages.
I have now watched some version of this movie at least a dozen times. The mistakes hiring forward deployed engineers are remarkably consistent, because the role looks like jobs it is not. It is not consulting, it is not standard software engineering, and it is not staff augmentation with a fancier title. Hire for the wrong job and you get the wrong person, every time.
Short answer: companies botch FDE hires by hiring consultants in FDE clothing, skipping the client-readiness check, interviewing like big tech, optimizing for cheap, writing no outcome into the offer, ignoring the communication demo, and hiring before a real first problem exists. Here are the seven failure modes, in roughly the order they kill engagements.
Mistakes Hiring Forward Deployed Engineers Make: The Consultant Clone Comes First
The most common of all the mistakes hiring forward deployed engineers, and the most expensive. The candidate interviews brilliantly because interviewing is their core skill: they have spent years being assessed on how they talk about work, not on the work itself. Ask what they shipped last quarter and you get a story about "driving alignment." Ask to see the repo and the room goes quiet.
Spotting them is easy if you look. Consultants-in-disguise describe outcomes at the program level ("led a digital transformation initiative") and go vague at the commit level. Real FDEs get specific fast: "the invoice parser broke on scanned PDFs, so I added a fallback OCR pass and accuracy went from 71 to 94 percent." One of those sentences belongs in your building. The other belongs in a boardroom you are paying rent on.
Give every finalist a small paid build exercise: a working prototype against a sanitized slice of your actual problem, due in a week. Consultants in FDE clothing self-select out at this step, which alone is worth the fee.
Mistake 2: Skipping the Client-Readiness Check
Sometimes the hire is fine and the company is the problem. Before you hire an FDE, confirm three things: someone can grant this person data access within five business days, a decision-maker will actually answer questions, and a process exists that is painful enough to matter.
If the answers are no, you are hiring a very expensive person to sit in your lobby. I watched one firm burn two months of a six-month engagement waiting for read access to their own ERP, because nobody had the authority to approve it and the person who did was on a sabbatical. The FDE was brilliant. The engagement was a statue.
Mistake 3: Interviewing Like Big Tech
The job is maybe 60 percent building and 40 percent diplomacy with people who did not ask for software. Then companies run a loop of algorithm puzzles that selects for neither. You do not need someone who can invert a binary tree on a whiteboard. You need someone who can watch a dispatcher work for an hour, spot the real bottleneck, prototype a fix by Thursday, and explain it to the dispatcher without using the word "asynchronous."
Better signals: a walkthrough of something they built for a non-technical user, a story about a time the client's stated requirement was wrong and what they did about it, and a live session where they interrogate a messy process description. If your interview could be given unchanged to a backend candidate at a search company, it is the wrong interview.
Mistake 4: Optimizing for Cheap
FDE pricing looks alarming next to a contractor's day rate, so procurement negotiates it down, and then acts surprised when the senior people vanish and the work lands on someone's first rotation. The rate was never the expensive part. A failed engagement costs the fee, plus six months, plus the credibility of the next person who tries to fix anything in that building.
Benchmark against reality before you haggle: forward deployed engineer salary data tells you what the market actually pays for people who can do this, and the delta between "market" and "the number procurement liked" is usually smaller than one month of a stalled project.
Mistake 5: Writing No Outcome Into the Offer
"Help us with AI" is not a scope. "Modernize our systems" is not a scope. A scope sounds like: reduce order-entry time from eleven minutes to four, or cut month-end close from nine days to five. Without a number, you cannot evaluate the work, the FDE cannot prioritize, and the engagement drifts toward whatever is most interesting instead of whatever is most valuable.
Write one to three measurable outcomes into the agreement before anyone starts. Then, once the person is running, the question of whether it is working becomes answerable; there is a whole discipline of measuring FDE performance, and it starts with having something to measure against.
Mistake 6: Ignoring the Communication Demo
Every technical screen should include one test that has nothing to do with code: have the candidate explain a technical decision to a non-engineer on your team, live. Not a prepared talk. Hand them a scenario, give them ten minutes, and watch whether your ops lead's eyes glaze over or light up.
This is not soft-skills theater. An FDE who cannot make a warehouse manager care about a queue retry policy will spend every engagement waiting for "requirements" that never arrive, because the requirements live in the heads of people they failed to reach. The demo predicts the job better than any algorithm question ever written.
Mistake 7: Hiring Before You Have a First Problem
An FDE without a queue drifts into meetings. It is not a character flaw; it is physics. The role is defined by shipping against real pain, and a hire made "because we'll need capacity soon" arrives to a vacuum, starts attending steering committees, and within a quarter has become the thing you were trying to avoid.
Have the first problem identified, sized, and sponsored before the start date. Have a rough sense of problems two and three. If you are still figuring out how to structure the offer, you are also still figuring out what the person will do, and that is too early to sign anything.
The pattern under all seven: companies hire for the job they know how to interview for, not the job that exists. Hire for shipping plus diplomacy, get your own house ready first, and the role pays for itself. Hire a deck, get a deck.