The true cost of a forward deployed motion

Every vendor selling to large accounts eventually has the meeting: enterprise deals are stalling in deployment, the board has read about forward deployed engineers, someone proposes hiring two. What rarely happens in that meeting is arithmetic. Here it is, with the sourced numbers separated from our assumptions.

The visible cost

Line item	Number	Status
Median FDE base salary	$173,816	Sourced — Bloomberry, 1,000 postings; 70% add equity
Benefits on top of wages	~31% of total comp → ~$252K	Sourced — BLS ECEC (all-civilian average)
Travel, tooling, management overhead	+10–20%	Our estimate for a travel-heavy role
Fully loaded	$280K–$400K/yr	Model; top end = senior markets plus equity
Concurrent engagements per FDE	~2	Our working assumption — complex accounts run below two, which makes this math worse, not better
Field cost per engagement, per year	$125K–$200K	Model output

That last row is the number that matters: it is the effective ACV floor for white-glove deployment. Below it, every engagement is a loss before you’ve paid for hosting. This is the quiet reason the motion has been a luxury good, reserved for the seven-figure contracts that can absorb it.

The cost your CFO sees later

Field engineering has to land somewhere on the P&L, and both destinations hurt. Book it as cost of revenue and it drags gross margin toward services territory: SaaS professional services run break-even to roughly 5% gross margin per CloudZero’s benchmarks, against a median of 80% on software revenue (Aleph/Benchmarkit, 342 companies) — and once services pass ~15% of total revenue, the market starts reading you as a services business. The reference points: Accenture at ~32% gross margin; Palantir’s Foundry at ~80%.

Book it as sales and marketing instead — Palantir’s actual trick — and you’ve made a bet. Palantir’s S-1 describes running Acquire-phase pilots at its own expense at negative contribution margin, recovered when Scale-phase accounts reach ~55% contribution. That bet pays only if field cost per account falls over time. Andrusko’s test for anyone running the motion: does forward deployed effort decline on mature accounts? If not, the CAC never amortizes and revenue scales linearly with headcount — a body shop with SaaS pricing.

The cost nobody books at all

The loaded rate assumes those hours go to FDE work. Audit a real week and a large, unmeasured share goes elsewhere: chasing an access grant, assembling the status email, hunting for the decision someone made on a call three weeks ago. Nobody has published a defensible time split for the role — we’re instrumenting design-partner engagements to produce one. On top sits the opportunity cost: what that engineer didn’t build, and the deals you declined because the team was full.

What changes the arithmetic

Every number above hangs on one variable: engagements per person. That’s the variable Gravel moves — agents carry the coordination layer, your team keeps the judgment, and one person supervises engagements instead of carrying them.

	Traditional	With agents on coordination
Fully loaded cost per person	$280K–$400K	$280K–$400K, plus software
Concurrent engagements	~2	~6 supervised (our design target)
Field cost per engagement-year	$125K–$200K	~$47K–$67K before software
ACV floor where white-glove pencils	≈ $125K+	Reaches well below $100K

On the software line: for the gap to close, Gravel’s subscription would have to cost more than ~$80K per engagement per year. It is not close to that.

Two honest caveats. The six-engagement ratio is a design target, not an audited industry average — design partners get the measured distribution first. And agents compress coordination hours, not judgment hours; a motion staffed with people who can’t make deployment calls fails at any ratio. The direction is hard to argue with — MIT’s data has vendor-partnered deployments succeeding at three times the rate of internal builds (67%, versus roughly a third of that), so the motion earns its cost when you can afford to run it. The magnitude is what we’re instrumenting design partners to prove.