MVP Features You Should Cut (And 5 You Shouldn't)

I've helped 30+ founders ship MVPs in the last three years. The same five features always get built that no one needs. The same five always get skipped that always matter by week three. This post is the cut list.

If you're staring at a Notion doc with 47 features and a six-week timeline, you're already losing. Pre-product-market-fit, every feature is a liability: code to maintain, a bug surface, a thing you'll have to migrate later. Cut hard. Then cut again.

5 features to cut before you write a line of code

These are the five I see in nearly every MVP scoping doc. They eat 30-40% of the build budget and contribute roughly zero to your first 100 users. Cut them now.

1. Cut the custom admin dashboard. Use the database GUI (Supabase Studio, Retool, Postgres in TablePlus). You will look at this dashboard 12 times before you have 100 users. Don't burn 40 engineering hours on it.
2. Cut multi-language support. Internationalization touches every component, every email, every error message. Ship in English. If a French customer wants in, that's a great problem and a paid feature later.
3. Cut SSO and SAML. Magic links or email plus password covers 100% of pre-PMF B2B sales. Enterprise SSO is a deal-closer for $50K+ ACV deals, not for your first ten customers paying $49/month.
4. Cut the public API and developer docs. You don't have a developer audience yet. Building OAuth2, rate limiting, webhook signing, and a docs site is two weeks of work for zero revenue. Add it when a paying customer asks.
5. Cut customizable email templates. Founders love this one. Users do not. Ship one transactional email per type, hard-coded, with your brand. Customization is a v3 feature.

Bonus cut: dark mode. I know. But dark mode is a 4-8 hour rabbit hole when you account for testing every component, charts, syntax highlighters, and Tailwind's class soup. Ship light only. Add dark mode in week 8 if anyone asks.

Why founders keep building these (and shouldn't)

Every cut-list feature has the same psychological pull: it feels like real work, it's tractable, and the spec is clear. Building an admin dashboard is satisfying. Talking to your tenth user who hates your onboarding is not. Founders default to the satisfying work and starve the painful work.

I covered the validation side of this in [How to Validate a Startup Idea Before Writing a Line of Code](/blog/how-to-validate-startup-idea-before-building). Read that first if you haven't started building yet. If you're already mid-build, this cut list is your second chance.

5 features you should never cut

These five are non-negotiable. Skip any of them and you'll spend week three doing emergency surgery instead of talking to customers. They cost less than $50/month total at MVP scale.

1. Authentication done properly. Use Clerk ($25/month free tier covers 10K MAU) or Auth.js with Postgres. Roll-your-own auth is the most expensive 'free' decision in software. Hashing passwords correctly, session rotation, password reset emails, rate limiting login endpoints — all solved problems. Use the solution.
2. Core CRUD with audit logging. The user can create the thing, read the thing, update the thing, delete the thing. Every state change writes a row to an audit_log table with user_id, action, timestamp, and a JSON diff. You will thank past-you the first time a customer asks 'where did my data go?'
3. Payments via Stripe Checkout. Stripe Checkout is a hosted page. Plug in a price ID, redirect, listen for the webhook. Two hours of work. Stripe charges 2.9% plus 30 cents. Don't build Braintree comparisons or 'we'll add billing later.' Charge from day one.
4. Error reporting via Sentry. Free tier: 5K events/month. Wire it into your Next.js app in 10 minutes. The first time a user hits a 500 and you see the stack trace before they email support, you'll understand. Ship without this and you're flying blind.
5. Product analytics via PostHog or Plausible. Track three events: signup, activation moment (the thing that proves the user got value), and the paywall hit. PostHog free tier covers 1M events/month. Plausible is $9/month if you only need page views. You cannot improve what you do not measure.

The cut list vs keep list at a glance

Here's the quick visual. If you're scoping with a co-founder or a contractor, screenshot this and align on it before the kickoff call.

How to defend the cut list against your own brain

Founders cave on the cut list under three pressures: investor demos, competitor envy, and 'just one more feature' from a single loud beta user. Here's how I handle each in client engagements.

• Investor demo pressure: investors don't fund features, they fund evidence of demand. Ten paying customers on a janky MVP beats a polished demo with zero revenue every time. Show the Stripe dashboard, not the admin panel.
• Competitor envy: your competitor has 30 engineers and three years of build time. You have you, two months of runway, and a question to answer. Different game, different rules.
• Loud beta user: one user is not your market. If three paying users ask for the same thing, build it. If one free user asks, log it in a 'maybe' doc and move on.

What about design polish?

Polish is a keep, but a constrained one. Use a component library (shadcn/ui, Mantine) and pick a single accent color. That's your design system. Don't hire a designer for v1. Don't run a brand exercise. Don't build a Storybook. The MVP needs to be legible and functional, not beautiful. Beautiful comes after PMF.

If you want the full opinionated stack, see my post on [the best MVP tech stack for 2026](/blog/mvp-tech-stack-2026) — it pairs well with this cut list.

The 80/20 of MVP scope

If you cut everything on the cut list and ship everything on the keep list, your MVP is roughly 1,200-1,800 lines of application code, six database tables, and three external services (Stripe, Clerk, Sentry). It costs under $50/month to operate at 0-1,000 users. It ships in 7-14 days with a senior architect. Anyone telling you a different shape is selling you complexity.

I run this exact scope as my [MVP Build Sprint](/services/mvp-sprint) starting at $3,500. Two-week ship cycle, daily demos, no scope creep. The cut list is built into the engagement.