Discovery phase deliverables that justify their cost

A client wants a custom application. They've drawn three screens in Figma, described the workflow in a Notion doc, and want a quote by Friday. Most contractors will guess. The good ones propose a discovery phase first — 2-4 weeks of paid work that produces a real scope document and quote.

Clients hate paying for discovery. "It's just talking." Until the discovered scope is 3x what they originally described — at which point they're grateful they didn't sign the wrong contract.

The discovery phase justifies its cost only if it produces concrete deliverables. Otherwise it does feel like "just talking."

The deliverables that work

1. Written scope document

10-30 pages covering:

• User stories with acceptance criteria for every screen.
• Data model with entity diagrams.
• Integration list with each third-party service.
• Authentication and authorization model.
• Out-of-scope items explicitly listed.

This document is the foundation for the fixed-price quote. Without it, fixed price is gambling.

2. Technical architecture proposal

Diagrams showing:

• Component architecture (services, databases, queues).
• Sequence diagrams for critical flows (payment, signup, key user journey).
• Deployment topology.
• Hosting and infrastructure choices with rationale.

Tools: Excalidraw or Miro for collaboration, Mermaid for code-version-controlled diagrams.

3. Interactive prototype

Figma or Penpot covering the 5-10 most important screens. Not pixel-perfect, but clickable enough that stakeholders can walk through the flow.

This catches UX assumptions early — usually the client realizes 2-3 of their imagined features don't make sense, or new features emerge during walkthrough.

4. Risk register

List of identified risks with mitigation:

This shows the client you've thought beyond happy paths. Reassuring to non-technical buyers, useful to technical ones.

5. Estimation with assumptions

Not just a single number, but range with reasoning:

• Optimistic: $X if all integrations work as documented.
• Realistic: $Y assuming typical 20% rework rate.
• Pessimistic: $Z if data migration takes the upper bound estimate.
• Quoted: $Y, with $5K reserved for change requests.

6. Project plan with milestones

Gantt or simplified timeline with:

• 3-7 milestones (not 30 micro-tasks).
• Each milestone has deliverable and acceptance criteria.
• Payment tied to milestone acceptance.
• Realistic dates with explicit buffer.

7. Recommended team and engagement structure

Who's on the project, what role, % allocation. Client decides if they want junior or senior leads. Pricing follows.

What discovery doesn't produce

• Production-quality code — that's the build phase.
• Pixel-perfect design — that's the design phase.
• Detailed API documentation — that's the build phase.
• Promises about exact timeline outside high-level milestones.

Pricing discovery

Fixed price for discovery itself, typically $5-30K depending on project size. Discovery should be 5-15% of the total project budget.

Common structures:

• $5-10K for small projects (~$100K total): 2 weeks, lighter deliverables.
• $15-25K for medium projects ($200-500K total): 3 weeks, full deliverables.
• $30K+ for enterprise: 4-6 weeks, deep dive with stakeholder interviews.

What if the discovery shows the project is wrong?

Good. That's exactly what discovery is for. Common discoveries:

• Client should buy off-the-shelf instead of build.
• Project as described is 3x the budget — rescope.
• The real problem is different from the stated one.
• Third-party API doesn't support a critical feature.

Discovery saved them from a six-figure mistake. They'll be back when they have a better project.

Verdict

Discovery justifies its cost by producing seven concrete deliverables: scope document, architecture, prototype, risk register, ranged estimate, milestone plan, team structure. Without these, discovery looks like "just talking." With them, the client has documents they can show internally and a basis for confident commitment.