What Does the MVP Development Process Look Like, Week by Week?
Five stages, an 8-week calendar with design running ahead of code, what each stage produces, where the design-to-development handoff usually goes wrong — and what changes when one team does both.

The MVP development process runs in 5 stages over 6–12 weeks: a 30-minute scoping call (day 0), a written scope and fixed price (within 5 working days), design one to two weeks ahead of code (from week 1), a deployed build demoed every week (from week 2), and ship, run and hand over (weeks 6–12). That is how Kinetico, a product design and engineering studio, runs it; because the same people design and build, there is no handoff stage.
Most guides to the MVP process are step-lists — 4 steps at Toptal, 9 at ScienceSoft — that tell you what to think about and not when it happens or who does it. This one is a calendar. It is the process behind our MVP development service (branch 4 of our guides), and the weeks and tiers here are the same as on that page and in the cost guide, so nothing on this site disagrees with itself.
Key Takeaways
- •5 stages: scoping call (day 0) → written scope + fixed price (≤5 working days) → design 1–2 weeks ahead of code (week 1→) → build with weekly deployed demos (week 2→) → ship, run, hand over (weeks 6–12).
- •Timeline by tier: Lean 6 weeks · Standard 8 · Full 10–12. The design track and the build track run in parallel, design one to two weeks ahead — the calendar below is the Standard tier.
- •The design-to-development handoff is where two-vendor MVPs lose weeks: unspecified states, drifting design systems, QA against mocks, nobody owning the gap. With one team the stage does not exist.
- •Your job as the founder is one hour a week — open the demo, use it, decide — plus 5–10 real users to try the weekly build. Scope changes are written, not Monday surprises.
- •The MVP does not get rewritten at traction if it ships on a standard stack, with tests in the QA line and monitoring from week 1, and if roles and integrations are added as written changes.
What are the stages of the MVP development process?
Five, and each has a date, an output you can point at, and a characteristic way of going wrong:
| # | Stage | When | Output | How it goes wrong |
|---|---|---|---|---|
| 01 | Scoping call | Day 0 · 30 minutes | The problem, the constraint (budget, date), and whatever exists — deck, Figma, spreadsheet, no-code prototype | Skipping it and sending a feature list: the vendor prices your list instead of your problem |
| 02 | Written scope + fixed price | Within 5 working days | Modules, screens, states, integrations, weeks, one number. Nothing starts until this exists | A 'fixed price' with no written scope behind it — it is a rate in disguise |
| 03 | Design, 1–2 weeks ahead of code | Week 1 → | Flows, screens, every state (empty, error, loading, edge), and a coded design system engineering builds against | Design finished in full before any code; or design and build by different companies |
| 04 | Build with weekly demos | Week 2 → | A signed-in, deployed build every week; core workflow end to end first; your GitHub organisation from the first commit | Progress reported in percentages instead of a URL you can log into |
| 05 | Ship, run, hand over | Weeks 6–12 | Production deploy in your accounts, monitoring, documentation; repo, design files and infrastructure in your name | A handover that depends on a retainer; QA 'moved to after launch' |
Compare that with the published step-lists. ScienceSoft's nine steps put “design MVP UX and UI” as step 5, then back-end, then front-end, then testing — a strictly sequential process with design finished before code starts, and design and coding described as separate phases. Toptal's four steps are all design (PRD, information architecture, wireframes and prototypes, design QA) and end where engineering begins. Neither is wrong; both describe a process with a handoff in the middle of it, which is the thing this article is about.
How long does the MVP development process take, week by week?
6–12 weeks depending on tier. This is the Standard tier — 2 user roles, 12–20 screens, core workflow plus admin, payments or 2 integrations — as two parallel tracks. Read it across: design is always one to two weeks ahead of the build it feeds.
| Week | Design track | Build track |
|---|---|---|
| Week 1 | Scope locked; data model; core flows | Repo, CI, auth, database |
| Week 2 | Role-one screens, all states | Core object CRUD; first deployed demo |
| Week 3 | Role-two / admin screens | Core workflow end to end |
| Week 4 | Payments and settings screens | Admin, roles and permissions |
| Week 5 | Empty, error and edge states | Payments or integrations wired |
| Week 6 | Design QA against the build | Notifications, settings, polish |
| Week 7 | Handover files, design system docs | QA, performance, monitoring |
| Week 8 | — | Production deploy, docs, handover |
Two things to notice. The first deployed demo is in week 2, not week 8 — from then on progress is a URL you log into, not a percentage. And design QA (week 6) is against the running build, by the people who designed it, before the QA and performance week. The other tiers compress or extend the same shape:
| Tier | Weeks | Scope | Week-last looks like |
|---|---|---|---|
| Lean | 6 | 1 role · 6–10 screens · sign-in + one core workflow · 0–1 integration | Users can sign in and complete the core loop |
| Standard | 8 | 2 roles · 12–20 screens · core workflow + admin · payments or 2 integrations | Users pay; you can operate it |
| Full | 10 – 12 | 3 roles · 20–30 screens · billing, notifications, reporting · 3+ integrations | It runs as a business, not a demo |
For contrast, ScienceSoft publishes an MVP timeline of “from 2 weeks to 8 months” depending on approach — true, and not something you can plan a raise around. Fixed weeks are only possible because the scope is written first; the week-by-week for the Lean and Full tiers is in the MVP cost guide.
Where does the design-to-development handoff fit — and why is it the risky stage?
In the common two-vendor process the handoff sits between “design done” and “engineering starts”: a Figma file, some annotations, and a call. It is where MVPs lose weeks, because nothing on the design side was built to survive being read by strangers later. Here is the same six moments in a two-vendor process and in a one-team process:
| Moment | Design vendor → dev vendor | One team designs and builds |
|---|---|---|
| Specifying the screens | Designer annotates Figma; developer reads it later; questions arrive after the designer has moved on | The person who drew the screen writes the component; unspecified states are noticed while designing |
| States and edge cases | Happy path is designed; empty, error, loading and permission states are guessed at in code | Every state is designed in week 5 at the latest, because the same people will have to build it |
| Design system | Tokens live in Figma; a second copy is hand-built in code; the two drift | Tokens and components live in the repo; Figma is a view of the same system |
| Design QA | Against static mocks, if at all; 'looks close enough' ships | Against the running build, week 6, by the designer who is also on the build |
| Change during the build | A change touches two vendors, two backlogs and two invoices | One conversation; same-size swaps absorbed, additions written up with weeks and cost |
| Who owns the gap | Nobody — each vendor did their part | The team; there is no gap to own |
If you are stuck with two vendors, the mitigations are known: keep the designer engaged through the build, put the design system in code rather than only in Figma, and schedule design QA against the running product (Toptal's fourth step). They cost weeks and a second invoice. The reason we scope prototype and MVP as one engagement — the case made in PoC vs prototype vs MVP — is that the prototype simply becomes weeks one and two of the build, and the handoff never happens.
What actually happens in each stage?
Stage 1 — the scoping call (day 0)
Thirty minutes. You bring the problem you are solving, the constraint (budget and date), and whatever exists — a deck, a Figma file, a spreadsheet, a no-code prototype. The most useful thing you can say is the sentence that starts “I don't know if…”: it tells us which risk the MVP has to retire, which decides the tier. We will ask what we can remove; a good answer to that question is worth more than a feature list.
Stage 2 — the written scope and fixed price (within 5 working days)
Modules, screens, states, integrations, weeks and one number, in a document you can send to another vendor for a like-for-like quote. Nothing starts until it exists. It is also the contract for change: swaps of the same size are absorbed inside the price; additions are written up with their weeks and cost before any work starts.
Stage 3 — design, one to two weeks ahead of code (week 1 →)
Week 1 locks the data model and the core flows. From week 2 the designers are one screen-set ahead of the build: role-one screens with every state, then role-two and admin, then payments and settings, then the empty, error and edge states — in that order, because that is the order the build needs them. The design system (tokens, components) lives in the repository, so the product and the design cannot drift apart. The people doing this are the same people writing the code, which is why there is nothing to hand off.
Stage 4 — build with weekly demos (week 2 →)
Week 1 sets up the repository in your GitHub organisation, CI, auth and the database. Week 2 ships the first deployed, signed-in build. The core workflow goes end to end before anything else is broadened; then admin, roles and permissions; then payments or integrations. Every week you get a URL and an hour: use the build, decide the open questions, set the priority. TypeScript end to end — React with Next.js, a typed backend, PostgreSQL — deployed in your accounts.
Stage 5 — ship, run, hand over (weeks 6–12)
Design QA against the build, then a QA and performance week with monitoring wired, then production deploy, documentation and the handover: repository, deployment accounts, design files and docs in your name. There is no licence, retainer or lock-in required to keep running what was built. Then the MVP goes and does its job — finding out whether people come back and pay — which is a different process, and yours.
What does the founder do during the MVP development process?
Less than you fear, and not nothing. The failure mode on the founder side is either disappearing (and finding out at launch) or re-scoping every Monday. The job, by week:
| When | Your job |
|---|---|
| Before week 1 | The scoping call; read the written scope; say what to cut |
| Every week | One hour: open the demo URL, use it, decide the open questions, set next week's priority |
| Weeks 2–5 | Line up 5–10 real users to try the weekly build; bring back what they got stuck on |
| Week 5–6 | Approve payments copy, pricing, emails; get the production accounts created in your name |
| Handover | Confirm you can log into every account; read the docs; run the product for a week before asking for changes |
How do you keep the MVP from being rewritten at traction?
The MVP is v1, not a draft — the code you keep. Four process choices make that true without slowing week 6:
- A boring stack you can hire for. TypeScript, React with Next.js, PostgreSQL. It costs no more to build on, and it is the largest hiring pool if the work moves in-house — see Next.js and React development.
- Tests where they pay: in the QA line, on the core workflow and the money paths. Full coverage on an MVP is a way to spend week 6 twice; no tests is a way to spend year two on a rewrite.
- Monitoring from the first deploy. Logs and error tracking in week 2 cost an afternoon; adding them after the first outage costs a customer.
- Roles and integrations as written changes, not mid-build surprises. The second role is a full tier of work; adding it by stealth in week 5 is how architectures get bent. Add it as the next engagement, on the same codebase.
What is not on the list: micro-services, multi-region, a platform team. Those are what “from MVP to enterprise” means later, and the process above is designed so that when the day comes it is an extension, not a restart.
Frequently asked questions
What is the MVP development process?
The MVP development process is the sequence from idea to a deployed first version real users can sign into. In Kinetico's engagements it has five stages: a 30-minute scoping call (day 0); a written scope and fixed price (within 5 working days); design running one to two weeks ahead of code (from week 1); build with a deployed, signed-in demo every week (from week 2); and ship, run and hand over (weeks 6–12). Published step-lists elsewhere run from 4 steps (Toptal) to 9 (ScienceSoft); the difference is whether design and engineering are one team or two.
How long does the MVP development process take?
6–12 weeks in Kinetico's tiers: 6 weeks for a Lean MVP (1 role, one core workflow), 8 for Standard (2 roles, admin, payments or 2 integrations), 10–12 for Full (3 roles, billing, notifications, reporting) — plus up to 5 working days for scoping before week 1. ScienceSoft publishes 2 weeks to 8 months depending on approach. Timelines stretch when a second vendor has to re-interpret the design or when scope changes arrive without a written change.
What are the steps to build an MVP?
Decide the one risk the MVP must retire; write the scope down (roles, screens, states, integrations, weeks, price); design the flows and every state one to two weeks ahead of code; build the core workflow end to end first and demo a deployed build weekly; wire payments or integrations; QA against the design and the scope; deploy in your own accounts; hand over the repository, design files, docs and monitoring. Kinetico runs those as five stages over 6–12 weeks.
What is the design-to-development handoff, and how do you avoid problems with it?
The handoff is the point where finished designs (usually Figma) are passed to a separate engineering team to build. Problems come from re-interpretation: unspecified states, spacing and behaviour guessed at, screens that drift from the design, and QA against static mocks. You avoid it either by keeping the designer in the build with a coded design system and design QA against the running product — or, as Kinetico does, by having the same people design and build so there is no handoff stage at all.
What happens after the MVP launches?
Handover, then learning. At handover you own the repository, deployment accounts, design files and documentation, with monitoring in place; no retainer is required to keep it running. Then the MVP does its job: real usage answers whether people come back and pay. Keep it from becoming a rewrite by shipping on a standard stack, keeping tests in the QA line, and adding integrations and roles as written changes rather than mid-build surprises.
How much does the MVP development process cost?
$15,000–$50,000 fixed at Kinetico: Lean $15k–$22k over 6 weeks, Standard $22k–$35k over 8 weeks, Full $35k–$50k over 10–12 weeks, with scoping inside the price. The money splits roughly scoping 5–10%, design 20–25%, engineering 50–60%, QA and handover 10–15%. Market ranges and the full breakdown are in our MVP cost guide.
Who wrote this, and where the process comes from
Written by Pukar Khanal and the Kinetico team, a product design and engineering studio in Pokhara, Nepal, where the same senior people design the product and write the code that ships it. The stages, calendar and tiers are the engagement terms published on the service page, reproduced here without change. Competitor step-lists and timelines are quoted from ScienceSoft and Toptal as read in August 2026. This page replaces our earlier “From MVP to Enterprise Scale” playbook; the engineering advice that survived is in the last section.
Published 2026-08-17 · Last reviewed 2026-08-17 · Author: Pukar Khanal, Kinetico
Continue your research
How Much Does It Cost to Build an MVP in 2026? →
Scope tiers, product types, phase split, and the Lean and Full week-by-week.
Do I need a PoC or prototype first?PoC vs Prototype vs MVP →
Which risk each retires, and why the prototype should be weeks one and two of the build.
Who should run this process?Staff Augmentation vs Outsourcing vs Product Studio →
Who directs, who owns, and where each model stops working.
Ready for stage 1?MVP Development by One Team →
One scoping call → written scope and a fixed price within 5 working days.
Stage 1 is a 30-minute call. Bring the problem, the constraint and the “I don't know if…” sentence; you get the written scope and a fixed number within 5 working days.
Related articles
SaaS MVP Development: What Goes In, What Stays Out, and How Many Weeks It Takes
What a SaaS MVP is and how it differs from a regular MVP; the nine feature layers a first subscription product needs and which to buy rather than build; what to leave for v2; how the scope maps to 6/8/10–12-week tiers at $15k–$50k fixed; published timelines and costs; the stack; what to measure after launch; how to choose a SaaS MVP development company.
PoC vs Prototype vs MVP: What's the Difference, and Which Do You Need?
Proof of concept vs prototype vs MVP, with real numbers: a PoC retires technical risk in 1–3 weeks, a prototype retires usability and investor risk in 1–3 weeks, an MVP retires market risk in 6–12 weeks ($15k–$50k). Pick by the risk you need to kill.
How Much Does It Cost to Build an MVP in 2026? Real Numbers by Scope
MVP development cost in 2026: $15k–$50k over 6–12 weeks with a design-and-engineering studio — by scope tier, product type, phase and region, with a week-by-week timeline and the estimation model behind the numbers.