Requirements Package Before Development Starts: What Discovery Should Produce

What discovery artifacts are — brief, SRS, roles, flows, NFR — why each matters and what it is for. Phased estimation, not fixed price for the whole product.

Requirements Package Before Development Starts: What Discovery Should Produce

Requirements Package Before Development Starts: What Discovery Should Produce

By the Smartym Pro team | July 2026

Series: Discovery & documentation (3 parts):

  1. Requirements package (this article) — what must be ready before estimate and code
  2. Discovery process and D1–D11 artifactshow you get there
  3. AI for documentation draftsprompts, workflow, consistency

Introduction

This article explains the document types that typically appear at the end of discovery — before development starts: what each contains, why it matters, and what it is for in conversations with your team. It is not a formal gate to "bring twelve PDFs." It is a shared vocabulary that reduces uncertainty between client and engineering.

The package is built in the discovery process from a brief and workshops. Here we focus on what you end up with and why those artifacts matter for phase-one scope and estimation.


Where the package sits in the lifecycle

Idea / brief
     ↓
Discovery (workshops)     ← part 2
     ↓
Requirements package (this article)
     ↓
Estimate and budget (D10)
     ↓
Sign-off → contract
     ↓
Development
Stage Package What you discuss
Before discovery None Ballpark only
After discovery Documents below Phase-one estimate
After sign-off Signed scope Sprint 0, delivery

Documents in the package: what, why, what for

Document D# Why it matters What it is for
Project brief D2 Without "why," teams price visible features only Goals, users, MVP, metrics
SRS (by module) D1, D4 One picture of functional scope Phase scope, backlog, contract
Roles matrix D4 Hidden admin/permission complexity Security, cabinet estimates
User flows D4 SRS prose does not replace the user path UX gaps, testability
Wireframes D11 Removes UI ambiguity UI-heavy products without full design
Sitemap / IA D1 Screen count and navigation UI scope boundaries
Data model (lite) D7 Backend is often harder than it looks API, migrations, admin
Integrations D6 Main schedule and risk driver Realistic phase estimate
NFRs D5 "Works" ≠ fast / secure Architecture, compliance
MVP scope (MoSCoW) D3 Without priorities, everything gets priced Phase one, not full roadmap
Out of scope D1 Stops scope creep Phase boundary
Risks & assumptions D8 Makes unknowns explicit When estimates change
Roadmap / phasing D9 Products evolve in iterations Planning phases 2, 3…
Technical constraints D7 Feasibility Stack, legacy, hosting

SRS: structure and modules

The SRS is grouped by modules so requirements are easier to read and align — that is document structure, not a call for fixed price on every module across the full product roadmap.

Recommended SRS sections:

  1. Introduction
  2. Product overview
  3. User roles
  4. Functional requirements (by module)
  5. Non-functional requirements
  6. Integrations
  7. Business rules
  8. Data requirements
  9. Security
  10. Acceptance criteria
  11. Constraints and assumptions
  12. Out of scope

Example modules: Authentication → User Profile → Dashboard → Admin → Payments → Notifications → Reports → Integrations.

Per module: description, roles, requirements, business rules, edge cases, acceptance criteria, dependencies.


Roles matrix

Role View Create Edit Delete Approve
Guest Public pages
User Own data Requests Own data
Manager Team data Yes Team data Limited Yes
Admin All data Yes Yes Yes Yes

Without this matrix, admin panel and permission work is almost always under-estimated.


User flows

Describe key journeys step by step, for example registration:

  1. User opens registration page
  2. Enters email and password
  3. System validates input
  4. Confirmation email sent
  5. User confirms email
  6. User completes profile
  7. Access to dashboard

Minimum flows: registration/login, core business process, payment (if any), approval, admin, notifications, error/rejection paths.


Integrations

For each integration document:

  • name and purpose
  • direction (in / out / both)
  • data exchanged
  • authentication method
  • API docs / sandbox link
  • frequency, webhooks, retry
  • client-side owner
  • [UNKNOWN] where details are missing — do not invent

NFRs (summary)

Category Examples
Performance Page load, API latency, concurrent users
Availability Uptime target, backup/DR
Security HTTPS, RBAC, audit log, OWASP baseline
Compliance GDPR, consent, retention — flag "legal review needed"
Scale Users, RPS, file volume, data growth

Prioritize Must / Should / Could as in part 2.


Minimum vs full package

Minimum for a meaningful estimate:

  1. Project brief
  2. SRS (at least top MVP modules)
  3. Roles and permissions
  4. User flows
  5. Wireframes or Figma (low-fi)
  6. Integration list
  7. MVP scope (MoSCoW)
  8. Out of scope
  9. NFR priorities
  10. Estimate request format

For a tighter estimate add: detailed AC, data model, UI states, edge cases, legacy migration, deployment, SLA, documentation requirements.


Example folder layout

/project-docs
  01-project-brief.md
  02-srs.md
  03-user-roles-permissions.md
  04-user-flows.md
  05-sitemap.md
  06-data-model.md
  07-integrations.md
  08-non-functional-requirements.md
  09-acceptance-criteria.md
  10-mvp-scope.md
  11-out-of-scope.md
  12-estimation-request.md
/design
  figma-link.txt
  wireframes.pdf
/assets
  logo, brandbook, screenshots
/legal
  privacy-policy.md, terms.md

AI can speed up drafts and help keep artifacts consistent across files — see part 3.


Phased estimation, not fixed price for the whole product

These artifacts help everyone understand what is being discussed. The commercial number usually applies to phase one (MVP), not a multi-year roadmap priced as one lump sum.

Principle Practice
MoSCoW Must Boundary of the current phase for estimation
Iterations After phase-one release — refined scope and a new estimate for phase two
Do not expect fixed for everything Fixed price fits a signed phase when uncertainty is low
Otherwise Range, T&M with a cap, or hybrid

See engagement models pros and cons. For discovery process and ballpark → fixed phase levels, see part 2.

Reasonable ask after the package: assumptions, open questions, phase-one estimate, options for later phases, risks, out of scope, recommended engagement model — not one number "for the entire SRS."


Next in the series

  • Part 2 — discovery as a process: RACI, D1–D11, MVP/tenders, ballpark vs fixed phase
  • Part 3 — AI: drafts, prompts, consistency and impact review

Need a discovery workshop or help turning a brief into this package? See our approach or tell us about your project.


General guidance on software delivery practice — not legal or procurement advice.