FDE vs Consultant

Row six is the sharpest one. A consultancy is structurally rewarded for the engagement continuing; an FDE organisation is rewarded for it ending well. If your FDE job is measured on the first thing rather than the second, the label has slipped.

a16z’s own thesis for why software companies are hiring FDEs is that they are deliberately trading gross margin for a moat — accepting the lower-margin economics of services because deployment depth is defensible.⁸ That is a clear-eyed description of a real trade-off, not a denial that services are involved. The critique lands in three places:

Palantir famously handed forward-deployed staff a copy of Keith Johnstone’s Impro — a book about improvisation and status dynamics — which tells you how seriously the strongest FDE cultures take the non-technical half.¹⁴ Treat those skills as engineering skills, because they determine whether your engineering ships.

One structural note in the FDE column: the big consultancies are now standing up FDE practices of their own, and Microsoft committed $2.5B to a 6,000-expert organisation doing embedded deployment work.⁷ The two worlds are converging from both ends — which means the label on your business card will matter less over time than the four questions in section 02.

Say a client asks for an invoice-triage workflow, scoped at 12 weeks. A consultancy staffs it with two people at 40 billable hours a week. A product company sends one FDE, and the client pays the same licence however long the work takes. Halfway through, the client asks for six more weeks. All the numbers are illustrative.

Steps 2 and 4 are the same request with opposite signs. That is the whole incentive argument in two rows. Step 6 shows where an FDE org gets its leverage: not from billing more hours, but from each engagement making the next one shorter. Step 7 is the mirror image. On hourly billing, the reuse that makes an FDE org more efficient makes a consultancy smaller, unless it changes how it prices. That is why fixed-fee and outcome-priced consulting exist, and why the question to ask is how the work is paid for, not what it is called. If your org would celebrate step 2, it has drifted into the top loop of the diagram.

Here Case is the choice, Time is what it costs in weeks, Space is what you or the company are left holding, and Why is the catch.

None of these is wrong on its own. A bespoke build is right when the customer is genuinely unusual. The trap is when the bespoke choice becomes the default and nobody writes down what the product should absorb. A quick test at the end of any engagement: list what you built, and mark each piece as product, reusable pattern or one-off. A list that is all one-offs is a consulting engagement, whatever the contract called it.

• “Isn’t this just consulting?”

  You may be asked this in an FDE loop, or ask it yourself. A strong answer names the three differences, then says which drift you would watch for and how you would notice it. Candidates coming from consulting carry the extra burden of showing code that ran in production. Bring one system you built, what it replaced, and who runs it now.

• Dev versus Delta

  Palantir keeps its forward-deployed org separate from the product org that builds the platform. The FDE vs FDSE vs Deployment Strategist page covers the split. It is the clearest example of the bottom loop as an org chart.

• Your own performance review

  The metrics in your review are the business model, written down. Adoption, product contributions and engagements that ended cleanly point to the bottom loop. Utilisation, hours and extensions point to the top one, whatever the title says. Ask to see the review template before you accept an offer.

• Margin conversations

  The a16z thesis cited above treats FDE headcount as margin traded for a moat. When FDE time starts showing up as a priced line item, ask how it is measured internally. Priced is fine. Measured on hours is the drift.