The Discovery Sprint Contract: How We Charge for Thinking Before Building

A prospect emails you a two-paragraph brief and asks for a fixed quote. You spend six hours scoping, send a number, and lose to the agency that quoted half. Six months later, that agency is in change-request hell and the client is calling you again. The fix is not better estimation — it's refusing to quote without a paid discovery sprint first.

We've been running paid discovery sprints as a standalone product for several years across web, mobile, and AI builds. This is the version of the playbook we'd hand a founder running a 10 – 40 person shop today.

Why free scoping is a tax on your best clients

Free scoping has three failure modes and they all compound.

First, you under-scope because you're working with marketing-grade information. The client's "simple admin panel" is a role-based permissions system with audit logs, and you won't find out until sprint four.

Second, you select for the wrong clients. Prospects who demand free detailed quotes before signing anything are, in our experience, the same prospects who later demand free detailed changes. The ones willing to pay for thinking are the ones who respect the engineering work.

Third, you subsidise losers with winners. If your close rate on cold quotes is 20%, every won project is paying for four lost ones in sales engineering time. That cost shows up somewhere — usually as a thinner margin on the delivery phase, which is where you actually make money.

A paid discovery sprint flips all three. The client self-qualifies by paying, you work with real information, and the sprint itself is profitable rather than a cost centre.

What goes into the sprint

We run discovery as a fixed-scope, fixed-fee engagement of one or two weeks depending on project complexity. The deliverables are concrete, not a slide deck full of adjectives.

The actual artefacts we ship

At the end of a discovery sprint, the client receives a tangible package they own regardless of whether they continue with us:

• A product requirements document with user stories grouped by epic
• A proposed technical architecture with service boundaries, data model sketches, and third-party dependencies named
• A clickable low-fidelity prototype of the three to five highest-risk screens
• A build estimate with ranges per epic, not a single number
• A risk register with mitigation options
• A proposed delivery plan with suggested team shape and cadence

The clickable prototype is non-negotiable. Non-technical stakeholders cannot evaluate a requirements document, but they can click a Figma prototype and say "wait, that's not what I meant." Finding that out in week two of discovery is cheap. Finding it out in week ten of a build is a change request war.

What we deliberately leave out

Pixel-perfect design, final copy, production-grade schemas, and anything that pretends the plan survives contact with the first sprint. Discovery is about reducing uncertainty to a level where a build estimate is honest, not about pre-building the product.

Pricing the sprint

Here's the part most agencies get wrong: they price discovery as a loss leader. We price it as a real engagement with real margin, because if it isn't profitable standalone, you'll resent doing it and your team will rush it.

In our experience, discovery sprints land in these bands for a typical two-week engagement with a senior engineer, a product lead, and a part-time designer:

• Small web app or feature-set addition: lower five figures USD
• Mobile app with backend, or AI-infused product: mid to upper five figures
• Platform rebuild or multi-tenant SaaS: low six figures

The exact number matters less than the ratio. We aim for discovery to be roughly 5 – 10% of the expected build cost. Below 5% and you're giving it away; above 10% and clients balk because they haven't yet seen you deliver.

The credit clause

To get hesitant clients across the line, we offer a partial credit: if they sign a build contract within 30 days of discovery completion, 50% of the discovery fee is credited against the first build invoice. Not 100% — that signals discovery was never really valuable. Fifty percent signals "we're confident you'll want to continue, and we're sharing some upside."

The contract clauses that matter

The discovery sprint contract is short, usually three to five pages, but a few clauses do most of the work.

1. SCOPE: Fixed deliverables listed in Appendix A.
   No commitment to build phase.

2. IP: All discovery artefacts transfer to Client on
   final payment. Agency retains right to reuse
   generalised patterns and non-confidential methods.

3. ESTIMATES: Build estimates provided are ranges,
   valid for 60 days, and contingent on scope as
   documented in the PRD.

4. NO-BUILD OUTCOME: Either party may decline to
   proceed to build phase for any reason. Discovery
   fee is non-refundable.

5. CREDIT: 50% of discovery fee credited against
   first build invoice if build contract signed
   within 30 days.

The IP clause is where founders get sentimental and lose money. The client paid for the artefacts, so they own them. If they take your PRD to a cheaper shop, that's their right — and it rarely happens, because the PRD is tightly bound to your architectural assumptions and your team's working style. In the handful of cases where it did happen, the client came back within six months because the cheaper shop couldn't execute what we'd scoped.

The no-build clause protects both sides. We've walked away from discoveries that revealed the client wasn't actually ready to build, or where the honest estimate was triple what they'd budgeted. Walking away cleanly, with the client holding useful artefacts, has generated more referrals than most of our completed projects.

Selling discovery to a prospect who wants a quote

The pitch is not "we need to understand your business." Every agency says that. The pitch is risk reduction with a dollar figure attached.

Try this framing on your next call:

"A fixed quote on a two-paragraph brief is either padded by 40% to cover our unknowns, or it's wrong and we'll both suffer through change requests. A two-week paid discovery gives you a real estimate, a prototype you can show your board, and an exit if you don't like what you see. If you proceed with us, half of what you paid comes off the first invoice."

Clients who refuse this are almost always clients who will refuse to pay for scope changes later. Losing them at the sales stage is a feature.

Handling the "we already know what we want" objection

Some prospects genuinely do have detailed specs, usually because they've built similar systems before or have a strong internal product team. For these, we offer a compressed one-week technical discovery focused on architecture and estimation only, at roughly 60% of the standard price. We skip the prototype and PRD, since they already exist.

This variant has saved deals with CTOs who felt insulted by the full discovery pitch. Giving sophisticated buyers a sophisticated option matters.

What changes inside your team

Running paid discovery as a product changes who you hire and how you staff. Discovery is senior work — it needs an engineer who can push back on bad ideas, a product lead who can facilitate stakeholder sessions without losing the thread, and a designer comfortable with low fidelity. Juniors can shadow but shouldn't lead.

We keep a small pool of people who rotate through discovery between build sprints. It's intense work and burns people out if they do it continuously, but used in rotation it keeps senior staff sharp and exposed to new domains.

Billing-wise, discovery utilisation looks different from build utilisation. A discovery team of three people over two weeks is roughly 240 billable hours against a fixed fee, so the effective rate per hour is what matters, not the headline number. Track it that way internally or you'll under-price the next one.

Where we'd start

If you're running an agency that still quotes cold, pick your next three qualified inbound leads and offer only a discovery sprint — no fixed build quote, no exceptions. Price it at 7% of your gut-feel build estimate. Write the five-clause contract above. Expect to lose at least one of the three, and expect the two that convert to be better clients than your average.

Do that for a quarter, measure close rate and project margin against your baseline, and decide from data whether paid discovery becomes your default. For most shops we've talked to, it does. For more on how we structure engagements end-to-end, see our services overview.