You’re staring at a proposal draft with the price field blank, and the two pieces of advice you’ve heard most are fighting in your head: “charge what you’re worth” and “start cheap to get experience.” Both are wrong for where you’re sitting right now. There’s a third path — a defensible starting price for an AI service you’ve never delivered, built from inputs you actually have.

This is the number you can say out loud on a discovery call without flinching, without apologizing, and without secretly hoping the client doesn’t push back.

Why the two dominant pieces of pricing advice both fail you

“Charge what you’re worth” assumes you have a track record that proves the worth. You don’t. Not yet. Quoting $15,000 for an automation you’ve never shipped is a great way to lose the deal and confirm to yourself that AI services don’t pay.

“Start cheap to get experience” assumes the only thing you’re missing is reps. But pricing at $500 for a project that takes you 40 hours doesn’t build a business — it builds a trap. You finish the project, you have a “case study,” and now your reference price is $500. The next client expects the same. You’ve anchored yourself into a business that can’t sustain you.

Both pieces of advice ignore the actual situation: you have real domain expertise, you have a service that will take real time to deliver, and your client is taking a real risk by hiring an unproven operator. The price has to respect all three.

The four inputs that should drive your starting price

Forget hourly rates. Forget “what the market charges.” A defensible starting price for an unproven AI service comes from four inputs you already have access to:

  • Domain expertise. The years you’ve spent in your industry are not zero. A marketing operator building an AI lead-qualification system is not the same as a generalist building the same thing. Your expertise is what makes the AI useful — price it in.
  • Scope and delivery effort. Estimate the hours honestly, then add 40%. First-time builds always take longer than you think. If it takes 30 hours of focused work, plan for 42.
  • Client risk. Your client is hiring someone without a portfolio for this specific service. That risk is real, and they should pay less than they would for an established operator delivering the same outcome. Not half. But noticeably less.
  • Margin to sustain the business. What does the price need to be for this to be a real business and not a hobby? If you can’t see a path to a 60%+ margin at this price, you’re not pricing — you’re subsidizing your client.

Notice what’s not on this list: hourly rates, competitor pricing, and “what feels right.” Those are downstream signals. They’re not the inputs.

Underpricing versus undervaluing — and which one will actually break you

These two failure modes sound similar. They’re not.

Underpricing for momentum means pricing below your sustainable rate on purpose, for a defined window, to win your first one or two clients and build proof. It’s a deliberate trade — you accept lower margins now in exchange for case studies, testimonials, and the confidence that comes from delivering. It has a clear exit: once you’ve delivered two or three engagements, your price moves up.

Undervaluing into an unsustainable business is what happens when you price low because you’re scared. There’s no exit. No plan to raise the rate. You just keep quoting cheap because every new client feels like the first one. Six months in, you’re booked solid at rates that don’t pay your bills, and you can’t raise prices without losing every existing client.

The bigger risk is almost always undervaluing — because it compounds. Underpricing is a one-time discount; undervaluing is a permanent ceiling.

How to tell which one you’re doing:

  • If you can name the specific number of clients at which you’ll raise your price, you’re underpricing for momentum.
  • If you can’t, you’re undervaluing.

Write the trigger down before you send the quote.

The starting-price framework — produces a range, not a number

You want a range, not a single number, because a range gives you room to negotiate without flinching and forces you to think about scope rather than just price.

Here’s the framework:

  1. Estimate the all-in delivery hours. Be honest. Include discovery, build, revisions, and handoff. Multiply by 1.4 for first-time-builder buffer.
  2. Set a sustainable hourly floor. This is the rate at which the business actually works for you — typically $100-200/hr for solo AI service operators, depending on domain. This is not what you bill. It’s the rate that makes the math work.
  3. Calculate the floor price. Hours × hourly floor = the absolute minimum you’d take this engagement for. Below this, you’re undervaluing.
  4. Apply the expertise multiplier. If your domain expertise is directly responsible for the outcome (not just the AI execution), multiply the floor by 1.5 to 2.5x. Deep expertise + proprietary judgment = the high end. Generic execution = the low end.
  5. Apply the unproven-operator discount. Take 15-25% off the top of the expertise-adjusted number. This is the honest acknowledgment that the client is taking a risk on you. After two or three delivered engagements, this discount disappears.

The result is your starting price range — floor on one end, expertise-adjusted-minus-discount on the other.

Worked example: pricing an AI-powered lead qualification system

Let’s run a real one. You’re a marketing operator with eight years of B2B SaaS experience. You’re quoting an AI-powered inbound lead qualification system — it ingests form submissions, enriches the lead data, scores against an ideal-customer profile, and routes the hot leads to a Slack channel with a one-line summary. You’ve built pieces of this before, but never the full system, and never for an external client.

Running the framework:

  • All-in hours estimate: Discovery (4) + system build (22) + testing (6) + revisions (5) + handoff and documentation (3) = 40 hours. With the 1.4x first-build buffer: 56 hours.
  • Hourly floor: You decide $125/hr is the rate at which this is a real business for you. Below that, you’re working for free in a costume.
  • Floor price: 56 × $125 = $7,000.
  • Expertise multiplier: Your B2B SaaS marketing experience is directly responsible for the ICP scoring logic — the AI is the execution engine, but the judgment is yours. That’s a 2x multiplier. $7,000 × 2 = $14,000.
  • Unproven-operator discount: This is your first external delivery of this exact system. Take 20% off. $14,000 × 0.8 = $11,200.

Your starting price range is $7,000 to $11,200. You quote $11,000 and you’re prepared to land at $9,500 if the scope tightens. Below $7,000, you walk — that’s the floor for a reason.

Notice what just happened. You didn’t pull a number from the air. You didn’t ask Reddit. You didn’t undercut yourself by quoting $2,500 because it felt safer. You also didn’t quote $25,000 because some LinkedIn post said premium pricing was a mindset. You produced a defensible number you can explain — to the client, and to yourself.

The pricing checklist — apply this to your own first quote

Use this before you send any proposal. If you can’t answer every line, you’re not ready to quote.

  • I have estimated all-in delivery hours honestly, including discovery, build, revisions, and handoff. Example: 40 hours base + 40% buffer = 56 hours.
  • I have multiplied the hours by 1.4 to account for first-time-builder reality. Example: confirmed 56 hours, not the optimistic 40.
  • I have set a sustainable hourly floor that makes this a real business, not a hobby. Example: $125/hr — the rate below which I’m subsidizing the client.
  • I have calculated my floor price (hours × hourly floor). Below this, I walk. Example: 56 × $125 = $7,000 floor.
  • I have named the specific domain expertise the client is paying for — not the AI. Example: 8 years of B2B SaaS ICP scoring judgment.
  • I have applied an expertise multiplier (1.5x to 2.5x) based on how directly that expertise drives the outcome. Example: 2x — my judgment defines the scoring logic.
  • I have applied an unproven-operator discount of 15-25%. Example: 20% off, because this is my first external delivery of this exact system.
  • I have a starting-price range, not a single number. Example: $7,000 floor to $11,200 top of range.
  • I have named the specific trigger that ends my discount — number of delivered engagements before the price moves up. Example: discount disappears after engagement #3.
  • I can explain the price out loud, in plain language, without using the words “worth,” “premium,” or “value.”

If you can check every box, send the proposal.

How to say the price out loud

The price is going to come up on a call. Here’s how to handle it without apologizing or overselling.

Don’t do this:

“So, um, I was thinking somewhere around $11,000 — but I’m flexible, you know, if that’s too much we can definitely work something out…”

Do this:

“For this engagement, the price is $11,000. That covers discovery, the full system build, two rounds of revisions, and handoff with documentation. The scope assumes you can give me access to your form data and CRM within the first week — if that shifts, we’ll talk about timeline. Any questions on what’s included?”

That’s it. State the number. State what’s in it. State the one assumption that could move it. Ask if they have questions.

You don’t justify a price by talking more. You justify it by what’s in the proposal, and by being the kind of operator who can say the number without your voice tightening.

The line that matters: the goal isn’t to be cheap or to be expensive — it’s to be defensible. Cheap loses you a business. Expensive loses you a client. Defensible wins both.

→ Once you have a price, the next problem is the conversation that closes the deal.

If pricing your service still feels like guesswork after applying the checklist, that’s because pricing isn’t really a number problem — it’s a positioning problem. Members of the NextBuild cohort leave with a pricing model built around their specific service, not pricing advice they have to translate. We build the model alongside them, against real client conversations, until the price is something they can quote without flinching.

You don’t need permission to charge what the work is worth. You need a number you can defend.

You just built a defensible number from four inputs and pressure-tested it against a checklist — which is exactly the kind of work that’s easier when you’re not doing it alone, staring at a blank price field at midnight. That’s what NextBuild is built for: experienced operators architecting their AI-powered, productized service alongside our team in focused build sprints, not watching a course about it. We sit with you to shape the offer and the pricing model around your actual expertise, then help you sell the first one to a real client. If you’re ready to turn this framework into an engagement you can quote and close, come build it with the NextBuild cohort.