Should You Charge for a Discovery Phase?
Yes, for any project you cannot price confidently after a single conversation. Keep the first call free, because that call is qualification and it is your job to establish whether there is a project at all. Everything after it - the requirements, the data model, the integration list, the risks, the written scope - is work, and work that a competent client can see the value of. Charge a fixed fee, deliver in days, and credit the fee against the build if they go ahead.
The reason to do this is not that discovery is hard, though it is. It is that free scoping puts you in an impossible position: you have to do enough thinking to quote accurately, while being paid nothing for it, while competing against people who will guess a number quickly and adjust later. Charging removes all three problems at once. You get paid for the thinking, you take the time to do it properly, and you stop competing on speed of guessing.
The Handoff Test
Here is the standard we hold discovery to. Could the client take what you delivered, hand it to a completely different builder, and get an accurate quote from it without talking to you? If yes, you produced something with independent value and the fee was obviously fair. If no, you produced a sales document with a price tag on it, and the client will feel that even if they cannot name it.
That test is uncomfortable on purpose. It rules out the version of paid discovery that is really just a proposal with extra pages, and it forces the deliverable to contain the specific things another builder would need: what is being built, what it connects to, what the data looks like, what is explicitly out of scope, and where the risk sits. Write it so it survives you leaving, and you never have to defend the invoice.
What a Paid Discovery Actually Delivers
A document and a number. The document says what is being built and what is not. The number is a fixed-price quote for the build, or a tight range with the conditions that would move it named explicitly. Anything less and you are selling a meeting. These are the sections that earn the fee.
- The outcome in one paragraph, in the client's own language, agreed with them before anything else is written.
- A feature list split into in-scope, out-of-scope, and later. The out-of-scope list is the most valuable part of the entire document and the part clients read twice.
- The integration list: every external system involved, who owns access to it, and whether that access has actually been confirmed rather than assumed.
- The data model at a level a non-technical person can follow. What records exist, what they contain, and who is allowed to see them.
- The risk list: what could make this take twice as long, with a plain statement of what happens to the price if it does.
- A build quote and a delivery window, with the assumptions the numbers rest on written next to them.
The out-of-scope list is where most of the future arguments get prevented. Scope creep does not usually arrive as a big new demand, it arrives as a small thing that everyone assumed was included. Writing down what is not included, during a phase the client is paying for, is the cheapest insurance available, and it works alongside the practices in [how to handle scope creep on client projects](/blog/how-to-handle-scope-creep-on-client-projects).
How to Price a Discovery Phase
Price it as a proportion of the build you expect to quote - around a tenth is a workable rule - with a floor low enough that a client will say yes without a committee and high enough that you will take it seriously. Fixed fee, never hourly. Paid before you start, not on delivery. And credited against the build if they proceed, which is what makes it an easy yes rather than a second decision.
A structure that works across project sizes
| Element | The rule | Why it is that way |
|---|---|---|
| Fee basis | Roughly a tenth of the expected build price, with a floor | Scales with the amount of thinking required, and the floor stops small projects being unprofitable to scope |
| Fee type | Fixed, quoted up front | Hourly reintroduces the exact argument about time that paid discovery exists to end |
| Payment timing | Before the work starts | A client who will not pay for discovery will not pay for a build, and you find out for the price of an email |
| Credit against build | Yes, in full, stated in writing | Turns the fee into a deposit in their mind and removes the objection before it is raised |
| Duration | Days, not weeks | Momentum is the asset. A discovery that drags loses the enthusiasm that started it |
| Ownership of the document | Theirs on payment | It is the only version that passes the Handoff Test, and it is what justifies the fee |
Crediting the fee against the build is doing more work than it looks. It reframes the decision from should I pay for scoping into should I put down a deposit, which is a decision clients already know how to make. It also means that the only person who is out of pocket is the one who does not proceed, which is exactly the person you wanted to charge in the first place.
If you are also still working out the build number itself, price the two together rather than in sequence. The discovery fee should follow from the build quote, not float free of it, and the reasoning behind the build quote is covered in [how to price a Claude Code agency project](/blog/how-to-price-a-claude-code-agency-project).
Free Call or Paid Discovery: Where the Line Sits
The line is between finding out whether there is a project and working out what the project is. The first is sales and it is on you. The second is consulting and it is on them. Most builders draw this line far too late, then wonder why they spend two unpaid weeks on deals that never close.
What belongs on each side of the line
| Activity | Free | Paid |
|---|---|---|
| First conversation about the problem | Yes | |
| Establishing budget range and timeline | Yes | |
| Explaining how you work and what you have built | Yes | |
| A rough order-of-magnitude figure with caveats | Yes | |
| Detailed requirements gathering with their team | Yes | |
| Reviewing their existing codebase or systems | Yes | |
| Designing the data model or the integration approach | Yes | |
| A written scope with a fixed price attached | Yes | |
| A technical proof of concept | Yes |
The free call still needs to be a good one, because it is where you decide whether to offer discovery at all. The questions that make it useful are the same ones in [client discovery questions before you quote](/blog/client-discovery-questions-before-you-quote) - you are just asking them to qualify rather than to scope, and stopping the moment the answers start requiring real work to obtain.
How to Sell It Without Losing the Deal
Present it as the first phase of the project rather than as a gate before the project. The wording that works is close to this: the honest answer is that I cannot give you a firm price yet without understanding the integrations properly. What I would suggest is a scoping phase, fixed fee, delivered in a week, and you end up with a written scope and a fixed quote. If we go ahead with the build, that fee comes off the total.
- Lead with the reason, not the fee. The reason is that a firm price requires real work, and clients accept that immediately when it is said plainly.
- Say the deliverable out loud. A written scope and a fixed quote is a thing. A discovery phase is not, until you describe it as one.
- Name the credit in the same breath as the price. Left to a second sentence, it sounds like a concession you were forced into.
- Give it a short, specific duration. A week sounds like a project. A few weeks sounds like a delay.
- Do not apologise anywhere in the paragraph. The moment you sound uncertain about charging, they become uncertain about paying.
If the discovery offer sits inside a written document, keep it to a single page with the deliverable, the fee, the timeline and the credit visible without scrolling. The structure that does this well is the same one in [the one page AI proposal that closes](/blog/one-page-ai-proposal-that-closes), just aimed at a smaller decision.
Why Free Scoping Costs You the Project
Free scoping does not just cost you the hours. It costs you the deal, in a way that is almost invisible while it is happening. Because the work is unpaid, you rush it. Because you rushed it, the quote carries padding you cannot justify. Because the number looks padded, the client negotiates. And because you have already sunk a week into it, you accept less than you should to avoid wasting the week.
- You do the thinking anyway, but compressed into the gaps between paid work, which is where mistakes come from.
- You quote defensively. Unknowns become contingency, contingency becomes a bigger number, and the bigger number invites the price conversation.
- You have no leverage in that conversation, because the sunk cost is entirely yours and both of you know it.
- Nothing is on paper, so the scope stays negotiable throughout the build and every assumption resurfaces later as a request.
- Serious buyers and tyre-kickers get identical effort from you, which means the tyre-kickers are subsidised by your real clients.
That last one is the argument that usually lands. Somebody is paying for all that free scoping, and it is the clients who did say yes. Charging for discovery moves the cost onto the person creating it, which is the ordinary way every other professional service works.
When Not to Charge for Discovery
This is not a universal rule and treating it as one will cost you work. Skip the paid phase when the project is small enough to quote from a conversation, when the client is already yours, or when you genuinely want the relationship for reasons beyond this one project. Judgment beats policy here.
- Small fixed-scope jobs. If you can quote it accurately in one call, charging to scope it looks like friction for its own sake.
- Existing clients. You already know their systems and their people, which is most of what discovery buys. Fold the scoping into the work.
- Work you want for the case study. If the reference is worth more to you than the fee, decide that deliberately and say nothing about it.
- Very early conversations. Discovery is offered after the free call establishes that there is a real budget and a real problem, never before.
- Anything you would decline at full price. A paid discovery is not a way to make a bad project tolerable.
There is also a version of this that scales down neatly. Not every project needs a week of scoping, and a short paid audit with a two-page write-up is often the right size for a smaller build. The principle survives the shrinking: the client pays for the thinking, and the thinking arrives as a document.
What Happens If They Do Not Buy the Build
You got paid, they got something useful, and you found out cheaply that they were not going to proceed. That is a good outcome, not a failed sale, and the sooner you believe that the easier the whole model becomes to run. The alternative outcome - two unpaid weeks and the same answer - is strictly worse in every direction.
- Deliver the document properly regardless. Half-delivering because you sense they are drifting is the one move that turns a neutral outcome into a bad reputation.
- Ask for the reason and record it. A pattern in why clients stop after discovery tells you something real about your positioning or your pricing.
- Follow up on a calendar, not on a feeling. Budgets return, and a discovery document ages well because the problem is usually still there.
- Treat it as a paid case study. You now have a full picture of a real project and a client who paid you, which is a stronger reference than most free work produces.
Some of these come back months later with the same document and a signed budget. That only happens if the follow-up is systematic rather than sentimental, which is the argument in [how to follow up with a client without being annoying](/blog/how-to-follow-up-with-a-client-without-being-annoying). A discovery client is the easiest person on your list to follow up with, because there is a specific artefact to reference.
Running the Discovery Phase Itself
Keep it tight and finish it. The failure mode is not underpricing, it is letting a week become a month while you wait for access to a system nobody can find the credentials for. Set the shape before you start and hold it.
- Take payment and send a short agreement naming the deliverable, the duration and what you need from them.
- Ask for access on day one. Systems, credentials, the existing codebase, the person who actually knows how the current process works.
- Run one structured session with the people who use the thing, not only the person who is buying it.
- Write the out-of-scope list before the in-scope list. It is faster, and it exposes disagreements while they are still cheap.
- Send a draft mid-way and ask for corrections. Nothing in the final document should be a surprise on delivery day.
- Deliver the document and the quote together, walk them through it live, and ask for the decision on that call.
Delivering it live matters more than it sounds. A document sent by email gets skimmed, filed, and discussed without you in the room. Walked through on a call, it becomes a conversation about the build, and the build quote is right there in front of both of you at the moment the client is most convinced that you understand their problem. That is the same reason kickoff works better live, as covered in [the client kickoff call for an AI build](/blog/client-kickoff-call-ai-build).
Short, practical drops on skills, MCP, agents, prompts, and more. No spam, unsubscribe anytime.
Frequently asked questions
What is a paid discovery phase?
A short fixed-fee engagement, usually days rather than weeks, in which you work out exactly what is being built and produce a written scope with a firm price attached. It sits between the free qualifying call and the build itself. The client pays for the thinking, owns the resulting document, and normally has the fee credited against the build if they proceed.
How much should I charge for discovery?
A workable rule is around a tenth of the build price you expect to quote, with a floor low enough to be an easy yes and high enough that the work is worth doing properly. Always a fixed fee rather than hourly, because hourly reintroduces the argument about time that paid discovery exists to end. Take payment before you start.
Should the discovery fee be credited against the build?
Yes, in full, and say so at the moment you name the price. Crediting it turns the decision from should I pay for scoping into should I put down a deposit, which clients already know how to make. It also means the only person out of pocket is the one who does not proceed, which is precisely who you wanted to charge.
What if the client refuses to pay for discovery?
Take it as information rather than as an objection to overcome. A client who will not fund a small fixed-fee phase is unlikely to fund a build, and you learned that for the cost of one conversation instead of two unpaid weeks. If the project is small enough to quote from a single call, quote it and skip discovery entirely.
How long should a discovery phase take?
Days, not weeks. Around a week is the shape most builds want, and smaller projects can be scoped in two or three days with a shorter write-up. The risk is not underpricing it, it is letting the phase drift while you wait for system access, so ask for credentials and contacts on day one and hold the end date you promised.
Is paid discovery the same as a proposal?
No, and the difference is the Handoff Test. A proposal is a sales document that argues for hiring you. A discovery deliverable is a specification that another builder could quote from without ever speaking to you. If what you are producing fails that test, you are charging for a proposal, and clients can feel the difference even when they cannot articulate it.
Last reviewed by Duncan Rogoff on August 21, 2026
