You run a 120-person distribution company. Dispatch lives in a spreadsheet that three people maintain by hand, and you've finally decided to bring in an embedded engineer to fix it. So you do the responsible thing and collect three proposals for the same job.

The first quotes $2,200 a day. The second quotes $18,000 a month for "embedded capacity." The third wants almost nothing fixed up front, just 30% of the savings it can measure, whatever "measured" turns out to mean.

Same job, three numbers, and no arithmetic that compares them. It's like being quoted rent, a lease, and a timeshare for the same apartment.

FDE pricing models come in exactly these three flavors: day rate, retainer, and outcome-based. The real difference isn't the number on the quote, it's who holds the risk when the work goes sideways. Pick the model by how well you can define the work, not by which total looks smallest. If this is your first engagement, take a day rate on a short pilot and stop overthinking it.

The Quote That Made No Sense

The reason those three proposals felt incomparable is that they were pricing three different things. The day-rate firm is selling hours of a skilled person's life. The retainer firm is selling reserved capacity, like a lane on a highway they promise to keep open for you. The outcome firm is selling a result, or at least a share of one.

When owners get confused comparing FDE pricing models, it's usually because they line the totals up side by side and try to pick the cheapest. That's comparing apples to orchards. A $2,200 day rate for a four-week pilot is $44,000. An $18,000 monthly retainer over six months is $108,000. A 30% share of $400,000 in measured savings is $120,000. The totals barely matter until you ask the only question that does: if this goes wrong, who pays for the mistake?

Keep that question in your pocket. It earns its keep on all three models.

Day Rate: The Honest Default

The day rate is the workhorse of embedded work because it prices the one thing everyone can verify: the engineer showed up and worked. Typical market ranges for a senior embedded engineer run roughly $1,500 to $3,000 a day, with the top of the band for people with a track record in your industry and the bottom for strong generalists. Yes, that stings the first time you write it on a purchase order. It's still usually cheaper than the fully loaded cost of hiring wrong.

Here's the trade you're making. You bear all the scope risk. If the problem turns out to be three problems in a trench coat, the engagement just runs longer and you keep paying. The engineer bears almost none; they get paid whether your dispatch spreadsheet was a molehill or a mountain range.

That sounds bad until you realize what you get in exchange: total flexibility. You can redirect a day-rate engineer on Tuesday morning when you discover the real fire is somewhere else. Try doing that with an outcome contract. The day rate is the right default for first engagements, for fuzzy problems, and for anything where "we'll know the scope when we see it" is an honest sentence. For a fuller picture of what the money buys across an engagement, our complete FDE cost breakdown walks the ranges from pilot to year-long program.

Retainer: The Capacity Bet

A retainer buys a block of embedded time each month, say 10 to 15 days, at a modest discount to the day rate. The pitch is simple: you have a steady stream of weird problems, and you want the same engineer who already knows your systems to keep solving them without re-quoting every time.

Retainers earn their keep when two things are true. First, you genuinely have a pipeline of work, not one project wearing a fake mustache. Second, continuity matters, meaning the engineer's accumulated knowledge of your operations is itself valuable. A retainer that survives month three usually does so because the engineer has become the person who knows why the integration breaks every time the ERP updates.

The trap is "use it or lose it." Some months you won't have enough work, and you'll either pay for idle days or invent tasks to justify the block. I've seen retainer engineers assigned to reorganize a wiki in week seven because the ops lead felt guilty. Negotiate a rollover clause, or a break clause after a pilot period, and watch utilization like a hawk. If it drops below about 70% for two months running, the retainer is a gym membership.

Outcome-Based: The Aligned Minefield

Outcome-based pricing is the seductive one: pay on measured results. Hours saved, errors prevented, revenue that was stuck and now moves. Alignment! Skin in the game! What's not to love?

Three things, mostly. First, the baseline. If the deal is "30% of hours saved," somebody has to agree on how many hours the process took before, and that number is always fuzzier than anyone admits. Second, the measurement. Who counts the savings, with what tooling, and what stops the client from quietly changing the process so the savings shrink on paper? Third, the gaming problem. An engineer paid on one metric will optimize that metric, occasionally into absurdity, like automating a task so aggressively that three new manual tasks appear just out of scope.

None of this makes outcome pricing bad. It makes it a contract for situations where the measurement is boring and undisputed. One client of mine, a mid-size freight brokerage, paid purely on invoices processed per week through the new system versus the old one. Clean metric, one counter, both sides could see it. That's when outcome pricing is perfect: one measurable pain, one trusted meter. For everything fuzzier, a hybrid works better, meaning a reduced day rate plus a success bonus tied to a metric you both trust. If you want the math for what those savings are actually worth, the automation ROI walkthrough has a spreadsheet-ready version.

Who Bears the Risk in Each of the FDE Pricing Models

Lay the three models side by side and the pattern shows itself. There are really three risks in any embedded engagement, and each model deals them differently.

Roughly speaking, a day-rate engagement puts maybe 85% of the risk on you, a retainer splits it closer to 60/40, and a true outcome deal flips it to something like 25/75. There's no free lunch in that spectrum. You're not choosing a price; you're choosing which side of the table the anxiety sits on.

Picking the Model for Your Situation

Decision guide, from a lot of watched engagements:

  1. First engagement with a new partner? Day rate, two to four weeks, one defined problem. You're buying information about them as much as software. If the pilot goes well, every later conversation gets easier.
  2. Proven partner plus a steady stream of problems? Retainer. The discount is real, the continuity is the point, and you've already calibrated trust with a pilot or two.
  3. One measurable pain with an undisputed meter? Outcome or hybrid. Invoice counts, hours per week on a named task, error rates on a defined process. If you can write the metric in one sentence and both sides nod, this can be the cheapest model of all.
  4. Can't define the problem yet? Then no pricing model will save you. Buy three days of discovery at a day rate and come back to this list.

And if you're still torn between an embedded engineer, an agency, and buying something off the shelf, that's a different fork in the road entirely. The build vs. buy vs. FDE comparison is the map for that one.

The owners who get burned on pricing aren't the ones who picked the "wrong" model. They're the ones who never asked which risk they'd just agreed to carry. Ask that question in the first meeting, out loud, and watch how the person across the table answers. The quote will make a lot more sense after that.