Owners keep the labor savings
Most AI roadmaps assume the money customers spend on labor will move to software, and that payroll becomes revenue for whoever automates it.
A contractor who cuts three roles and saves $150,000 won’t pay you $120,000 for that. They might pay $15,000 and keep the other $135,000 as profit. They made the change and carried the risk, so they treat the savings as theirs.
The assumption
Labor spend moves to software.
Price against the payroll you replace and keep most of it.
What owners do
They keep the savings.
Expect to capture about 10% of the savings, 20% at most. The rest goes to profit.
The labor savings you create become the budget for your next product.
One P&L, step by step
Take a residential remodeler doing $10M a year at a 5% profit margin, or $500,000 of profit. The owner would like to run at 10%. All figures here are illustrative.
Your first product is an AI front office that answers calls, books appointments and follows up on leads. It removes $1M of labor from the P&L.
If you ask to split the difference, with the owner running at 10% and paying you $500,000, they’ll say no. At $100,000 they’ll say yes, and they now run at 14%.
Your second product is an AI field supervisor. It checks each job’s photos and daily logs against the plan and flags mistakes before the drywall goes up. At a 5% margin the owner won’t spend $200,000 on it, because that’s 40% of their profit. At 14%, fewer callbacks and fewer weekends on job sites are worth $200,000 to them.
| Step | Owner’s profit | Margin | Paid to you |
|---|---|---|---|
| Today | $500K | 5% | $0 |
| The AI front office removes $1M of labor | $1.5M | 15% | $0 |
| The split-the-difference ask (refused) | $1.0M | 10% | $500K |
| Product one, priced at 10% of the savings | $1.4M | 14% | $100K |
| Product two, once the savings show up | $1.2M | 12% | $300K ($100K + $200K) |
Illustrative
Illustrative
Illustrative
Why the owner buys the second product
At 5%, the owner can’t spare money for anything that doesn’t cut costs. At 14%, they can pay for quality and for their own time.
The owner made the change, so they see the savings as their own. Asking for half feels like a tax on their work.
Labor replacement is the first thing most AI vendors build, and larger customers can build it themselves. Expect its price to fall first.
By the second pitch, the owner has seen the first product move the margin in their own books.
Price the second product on its own results, such as fewer callbacks or faster payment. Don’t present it as a share of the savings.
Capture 10% now and 20–40% later
Asked up front, the owner refused to give you half the savings. Sold in sequence, you can earn the same amount: about 10% for the first product and 20–40% for the second. At 40%, you earn the $500,000 they turned down, and the owner still ends at the 10% margin they wanted.
We save you[savings],Labor removed from their P&L, in dollars.charge[about 10%],The first product.then sell you[a second product]One they couldn’t afford at their old margin.for[20–40%].Priced on its own results.
Where to find the second product
Look for a problem the owner has always had but never spent profit to fix. It usually sits in one of four places.
| Where, and what it is | Why a 5% business won’t buy it | Why a 14% business will |
|---|---|---|
| Processing and financing fees come straight out of a thin margin. | Getting paid sooner and funding more jobs is worth a point or two. | |
| The owner has no time to negotiate with suppliers. | The savings show up on every order, so the cost is easy to justify. | |
| Rework is treated as a cost of doing business. | Catching mistakes protects the margin the first product created. | |
| They’d rather put up with it than spend profit on it. | With money to spare, their own time becomes worth paying for. |
The owner’s time is the hardest to build for. Every job is local and custom, so checking whether a junior team did the work right is hard to automate. Owners would also pay the most to stop doing it themselves.
Why the subscription won’t grow enough
Most vertical software is a system of record with workflow on top. We expect general-purpose AI to learn each person’s workflow within the next year, leaving the system of record as a database with some workflow attached. Customers doing $10–20M a year are already building their own tools. They’ll keep paying a monthly subscription, but that fee won’t grow to $50,000 a year.
The deflation trap
More value, faster, for less money.
The subscription becomes a volume business, and payments carry the margin.
The two-product plan
Save them money, then sell them the next product.
Contract value grows from $5,000 to $25,000 to $50,000 by adding products for the same customer.
Hold the data the second product needs
A second product needs information that general-purpose models can’t see and the customer can’t easily rebuild. You get it by running the systems the business depends on, such as its phones, payments and job records.
Plan it in six steps
- Measure the labor your product removes, in dollars, for a typical customer.
- Price the first product at about 10% of those savings. Don’t ask to split the difference.
- Show the margin change in the customer’s own P&L before you pitch anything else.
- Pick the second product from the four places above: a problem the owner wouldn’t pay to fix at their old margin.
- Price it on its own results, at 20–40% of the original savings, so the owner still ends above their target margin.
- Schedule the second pitch in the annual plan for when the savings appear. A 5% business won’t buy both products at once.
Check the plan
- You can state the savings in dollars for one real customer.
- The first product costs the customer about a tenth of what it saves them.
- After both products, the customer keeps at least half the savings.
- The second product is one they wouldn’t buy at their old margin and would buy at their new one.
- You run a system (phones, payments, job records or the end-customer relationship) that the second product depends on.