Compound field notes / AI assessment
AI consulting pricing: start with the return
Our assessment is $2,500. Builds target a 5–10x first-year return. Here’s how we check the numbers before agreeing on a scope and fee.
5 min read
In this guide
Compound's AI Assessment is $2,500. We price builds around a target of 5–10x return within the first year, then agree on a fixed scope and fee before work starts.
That target has to survive the math. It means the expected first-year benefit should be five to ten times the total first-year investment, including the build and the cost of running it. It's a basis for deciding what to build and what to spend, not a guaranteed result.
I spent ten years in operations at Saucey, ending as VP of Operations. When I look at a process, I want to know where the time goes, what mistakes cost, and what gets left waiting. Those details tell us whether a build is worth doing.
The investment lens
What the 5–10x target means
For each unit of total first-year investment, the model targets five to ten units of economic benefit.
Total first-year investment
Target economic benefit
What you're paying for
The assessment takes 14 days. It gives you a ranked plan, the numbers behind the recommendations, tool choices, and a fixed-price scope for the first automation. You can take that plan to another builder. The deliverables guide shows what it contains.
Two terms matter before you book:
- Refund: if I don't find a durable, reliable solution to your bottleneck, you get the full $2,500 back.
- Build credit: start a Compound build within 30 days of assessment delivery and the full assessment fee comes off its price. The credit expires after that. Count the assessment once in your budget.
For the build, we agree on the work and price after the assessment. The first automation is live within 30 days of the build starting. We keep building until it works as scoped, with no overage.
The build includes:
- 30-Day Watch after launch.
- Documentation your team can refer to.
- A live team handoff to walk through the system.
- Human approval for automation that touches customers.
The scope also spells out the boundaries. Adding another workflow later is a separate piece of work.
Software subscriptions and usage charges belong in the budget from the start. Optional ongoing support is $1,500–$2,500 a month, covering improvements to the existing system. A new system needs its own scope.
What 5–10x means in the proposal
Here's the calculation we use:
Expected first-year benefit ÷ total first-year investment = benefit multiple.
A 5x multiple means five dollars of expected benefit for every dollar invested. At 10x, it's ten dollars. Those are the targets we work toward when scoping and pricing a build.
That's a benefit multiple. Net return on investment (ROI) subtracts the investment first, so check which calculation you're comparing.
Use the first 12 months from the start of the project for both sides of the calculation. If it takes time to build, train the team, and reach normal usage, the benefit needs to reflect those months. A full year's savings on a system that runs for only nine months would overstate the return.
Check that the investment includes:
- Build and setup: the agreed fee and your team's setup time.
- Running costs: software subscriptions, usage charges, and maintenance.
- Ongoing review: deduct those hours from the time recovered, or add their cost to the investment. Count them once.
- Assessment credit: apply it once if the build qualifies.
Start with a week you can measure
Suppose a task takes eight hours a week today. After automation, someone still spends two hours reviewing the output and fixing exceptions. That's six hours recovered.
Using illustrative assumptions of 50 working weeks and a $40 hourly cost, here's how the annual capacity value changes with the time recovered. These are examples, not client results or a forecast.
| Hours recovered each week | Hours across 50 weeks | Capacity value at $40/hour |
|---|---|---|
| 3 | 150 | $6,000 |
| 6 | 300 | $12,000 |
| 8 | 400 | $16,000 |
The middle row gives you $12,000 of capacity value in a full operating year. You'd adjust that for the launch date before using it in the first-year calculation. Then compare it with all the costs of the proposed system.
Be honest about what those hours are worth:
- More capacity: if payroll stays the same, faster work doesn't automatically improve the bank balance. Name the work your team will do with the recovered time.
- More customer work: estimate the additional gross profit after delivery costs. Don't count those same hours again as labor savings.
- Lower expenses: identify the expense that goes away and when it ends.
I'd also want to see the lower estimate. What happens if you recover three hours instead of six? If the case falls apart there, narrow the scope, look for a cheaper approach, or wait until the task volume justifies the work.
A useful build still has to earn its place
The Newgarden case study covers a check-photo-to-invoice workflow that recovered roughly six hours a week. It's a specific outcome from a specific process. It doesn't establish a return multiple for a different business.
Your existing software may already solve the problem with some setup. A connection between two tools might be enough. The buy-or-build guide walks through those choices.
For a discovery call, bring one task that's eating your week, the tools it touches, and a rough count of how often it happens. That's enough to start checking whether the AI Assessment is a sensible next step.