Skip to Main Content

// Service · Websites & Web Systems

Websites &
Web Systems

Your website is your best salesperson. Works 24/7. Never takes a vacation.

Most companies think of their website as a "digital business card." We think differently. Your site is the first — and sometimes last — touchpoint with a customer. It decides whether they stay or leave. Buy or look for a competitor.

In 2026, a site has to be fast, beautiful, clear, and adapt to every screen. But also something less talked about: it has to be built for an environment where AI bots crawl it and decide whether to surface it in their answers. We build from scratch. No templates. No shortcuts.

Show me how we work

// 01 · What we build

What we build

From a brochure site to a full platform. From a landing page to a marketplace. Each one built for its specific goal.

  • 01

    Corporate websites

    Your company's main site. Custom design, copy that converts, excellent performance. Not a $15 template.

  • 02

    Landing pages that convert

    Single pages focused on one action. Leads, sign-ups, purchases. Built-in A/B testing.

  • 03

    Online stores

    Shopify, WooCommerce, or fully custom. From product to checkout — a smooth, frictionless flow.

  • 04

    Web systems & platforms

    Systems users log into every day. SaaS, B2B, internal portals.

  • 05

    Content & media sites

    News sites, magazines, blogs. Built for SEO and speed from the ground up.

  • 06

    Marketplaces

    Two-sided platforms with payments, user management, ratings and reviews.

  • 07

    SaaS product sites

    Your product's marketing site. What it does, how to sign up, how to pay. The full flow, end to end.

// Which site do you need

Brochure site, landing page, store or web system — which one do you actually need?

"Website development" covers five very different products, each with its own timeline and budget. Before anyone talks design, it helps to know which one you are building.

  • Corporate / brochure site

    When it is the right call
    A company that needs a professional presence, service pages and leads — without complex business logic.
    Typical timeline
    4–8 weeks
    How we build it
    Next.js + headless CMS, green Core Web Vitals, SEO and GEO from day one.
  • Landing page / campaign

    When it is the right call
    One message, one audience, one action. Paid campaigns, launches, A/B tests.
    Typical timeline
    1–3 weeks
    How we build it
    Static and extremely fast, full measurement, variants for testing.
  • Online store

    When it is the right call
    Selling products directly. Catalog, cart, checkout, shipping and inventory.
    Typical timeline
    2–4 months
    How we build it
    Shopify or WooCommerce when that is enough; a custom headless store when the experience or integrations demand it.
  • Web system / web app

    When it is the right call
    Users log in and get work done: customer portal, admin system, internal tool, SaaS.
    Typical timeline
    3–6 months
    How we build it
    Fully custom — frontend, backend, permissions, integrations. At this point it is software development in every sense.
  • Platform / marketplace

    When it is the right call
    Two or more sides (suppliers and customers), payments, ratings, complete business flows.
    Typical timeline
    4–8 months to MVP
    How we build it
    Architecture planned for growth, an operational backoffice, and a clear v2 roadmap.

Not sure where your product lands? That is exactly the first question in the discovery meeting — and it drives the budget more than any design decision.

For complex web systems and SaaS — see our software & systems development service

// 02 · The approach

We don't build on templates.

Not because we're snobs — because templates don't serve you. They serve the company that sold them. A custom-built site runs faster, is easier to maintain, and actually reflects who you are.

  • 01 · 2026 standards

    Built for the right era

    Core Web Vitals, accessibility (EAA + IS 5568), Schema.org, structure crawlable by AI agents — not just Google.

  • 02 · Design before code

    Design first. Then code.

    Changing a screen in Figma is cheap. Changing it after it's coded is expensive. Every site is fully designed before a single line of code.

  • 03 · SEO + GEO from day one

    Found on Google and AI

    Built for rankings from the foundation up — semantics, structure, schema, speed. AI bots find you, Google ranks you.

// Web systems and portals

A system for customers, employees or partners?

A web system is not a website with a login. It is a place where people get work done: submit, approve, compare, report. The first distinction is who the user is and what the business action is — because that decides the sign-in, the permissions and what has to be simple.

System typeWho the user isThe business actionWhat it specifically requires
Customer portalExisting customers — consumers or businesses — outside the organisationSee a status, submit a request, pay, download a document, open a ticketSimple sign-in, self-service, self-explanatory screens; zero training
Employee portal / internal systemEmployees in different roles, sometimes field staff tooRun a process: draft → approval → execution → closure, with management reportsRole-based permissions, multi-step processes, organisational sign-in (SSO)
Partner / B2B portalSuppliers, distributors, agents — other organisations with their own usersSubmit a bid, update stock, order, report — against your dataSeparation between organisations, user management on the partner side, a data interface (API) alongside the screens
Admin / back-office systemThe team that operates the serviceSee everything, handle exceptions, approve, exportAn audit log, search and reports, admin permissions separated from end users
  • Customer portal

    Who the user is: Existing customers — consumers or businesses — outside the organisation

    The business action: See a status, submit a request, pay, download a document, open a ticket

    What it specifically requires: Simple sign-in, self-service, self-explanatory screens; zero training

  • Employee portal / internal system

    Who the user is: Employees in different roles, sometimes field staff too

    The business action: Run a process: draft → approval → execution → closure, with management reports

    What it specifically requires: Role-based permissions, multi-step processes, organisational sign-in (SSO)

  • Partner / B2B portal

    Who the user is: Suppliers, distributors, agents — other organisations with their own users

    The business action: Submit a bid, update stock, order, report — against your data

    What it specifically requires: Separation between organisations, user management on the partner side, a data interface (API) alongside the screens

  • Admin / back-office system

    Who the user is: The team that operates the service

    The business action: See everything, handle exceptions, approve, export

    What it specifically requires: An audit log, search and reports, admin permissions separated from end users

Most systems we build combine two or three of these types on one engine — a customer portal plus an admin system for the team, or an environment for suppliers alongside one for project owners.

What a portal includes — in practice

  • 01

    Sign-in

    Email and password, an SMS code, a Google or Microsoft account, or organisational SSO — depending on the audience. Password reset and lockout after failed attempts are part of the baseline, not an extra.

  • 02

    Permissions

    Who sees what, who changes what, who approves. By role, and sometimes by organisation or branch. Actions that need two people are defined up front.

  • 03

    Documents

    Upload, secure storage, versions, PDF generation, signing when required. Who may see each document follows from permissions — not from a link.

  • 04

    Processes and states

    Each entity has a lifecycle: draft, pending, approved, closed. Who moves it between states, and what happens in exceptions — cancellation, rollback, change after approval.

  • 05

    Notifications

    Email, SMS or an in-app alert when something is waiting for action. To whom, when, and how often — so they do not get ignored.

  • 06

    Reports

    The questions management needs to answer from the system without asking a developer. An Excel export is almost always the first report requested.

How does the system connect to the business?

A web system almost never stands alone. Every integration has to answer three questions: what passes through, how often and in which direction, and what happens when the other side is unavailable.

SystemWhat passes throughFrequency and directionWhat happens on failure
CRMCustomers, contacts, ticketsTwo-way, real-time or every few minutes; the source of truth for the customer is defined up frontThe system keeps working on the data it has; updates are queued and sent when the connection returns
ERP / accountingOrders, invoices, inventory, price listsUsually orders go out, inventory and prices come in; sync cycles are definedAn order does not vanish — it is marked 'pending sync' and someone sees it
Legacy system / filesHistorical data, spreadsheets, documentsA one-off migration at the start, sometimes periodic imports until the old system is retiredAn exceptions report: which rows were not imported and why, so they can be fixed and rerun
Messaging (email / SMS / WhatsApp)Alerts, confirmations, codesOutbound only, immediateAutomatic retry; the message is logged in the system even if the external provider fails
Organisational sign-in (SSO)The user's identity and roleOn every sign-inNo sign-in without authentication — but a clear message and a way to reach support
  • CRM

    What passes through: Customers, contacts, tickets

    Frequency and direction: Two-way, real-time or every few minutes; the source of truth for the customer is defined up front

    What happens on failure: The system keeps working on the data it has; updates are queued and sent when the connection returns

  • ERP / accounting

    What passes through: Orders, invoices, inventory, price lists

    Frequency and direction: Usually orders go out, inventory and prices come in; sync cycles are defined

    What happens on failure: An order does not vanish — it is marked 'pending sync' and someone sees it

  • Legacy system / files

    What passes through: Historical data, spreadsheets, documents

    Frequency and direction: A one-off migration at the start, sometimes periodic imports until the old system is retired

    What happens on failure: An exceptions report: which rows were not imported and why, so they can be fixed and rerun

  • Messaging (email / SMS / WhatsApp)

    What passes through: Alerts, confirmations, codes

    Frequency and direction: Outbound only, immediate

    What happens on failure: Automatic retry; the message is logged in the system even if the external provider fails

  • Organisational sign-in (SSO)

    What passes through: The user's identity and role

    Frequency and direction: On every sign-in

    What happens on failure: No sign-in without authentication — but a clear message and a way to reach support

Connecting to a system with a documented API is a completely different job from connecting to a legacy system with no interface. It is one of the first questions in discovery, because it drives scope more than most design decisions.

Users and permissions — an example

A table like this is usually the first document we write together. The example here is invented — a service portal for a company with business customers — not a client's data.

RoleSeesCreates / editsApproves
Customer (a user at the client organisation)Only their own organisation's tickets, orders and documentsA new ticket, a new order, contact details—
Manager at the clientSame as customer, for all users in their organisationAdding and removing users in their organisationOrders above a defined amount
Service repTickets and orders of the customers assigned to themTicket status, replies to the customer, internal notes—
Team leadAll tickets and orders, workload per repAssigning customers to reps, reply templatesCredits and cancellations
ManagementAggregate reports; no access to individual tickets by default——
System adminEverything, including the audit logRoles, permissions, integration settingsPermission changes for other users
  • Customer (a user at the client organisation)

    Sees: Only their own organisation's tickets, orders and documents

    Creates / edits: A new ticket, a new order, contact details

    Approves: —

  • Manager at the client

    Sees: Same as customer, for all users in their organisation

    Creates / edits: Adding and removing users in their organisation

    Approves: Orders above a defined amount

  • Service rep

    Sees: Tickets and orders of the customers assigned to them

    Creates / edits: Ticket status, replies to the customer, internal notes

    Approves: —

  • Team lead

    Sees: All tickets and orders, workload per rep

    Creates / edits: Assigning customers to reps, reply templates

    Approves: Credits and cancellations

  • Management

    Sees: Aggregate reports; no access to individual tickets by default

    Creates / edits: —

    Approves: —

  • System admin

    Sees: Everything, including the audit log

    Creates / edits: Roles, permissions, integration settings

    Approves: Permission changes for other users

Two rows in this table save most of the arguments later: who approves, and what management does not see by default.

Testing, migration and handover

  1. 1

    Acceptance criteria per process

    For every process in the system we write down in advance what counts as 'working': who does what, with which data, and what the result is. Testing — ours and yours — is done against that list, not against a feeling.

  2. 2

    Data migration

    What moves, what stays, how mismatched fields are mapped and how it is verified. A trial run on a copy of the data before the real cutover, and an exceptions report for anything that did not import. When needed — a parallel-run period.

  3. 3

    Rollout and adoption

    A pilot with one user group before opening to everyone; training for the people who will work with the system; and measurement of what is actually done in it during the first weeks — because a process nobody uses is a process to fix, not to explain.

  4. 4

    Handover and maintenance

    The repository in your name, cloud accounts and domain in your ownership, operating and configuration documentation, and a list of what was built against the specification. Maintenance after that — monitoring, updates, incident handling — under a separate agreement, within the scope and service hours it sets.

What this looks like in a real system

Michrazit is a two-sided portal: an environment for project owners and a separate one for suppliers, permissions that keep each side from seeing the other, a five-stage tender flow, and a platform team that sees both. The case study has a section describing exactly how the process was built into the system — roles, data and integrations.

// Sites and systems we built

Websites and web platforms we built — and that work

Three web products you can open today: a two-sided tender platform, an employee-onboarding system and a real-time gaming layer. What we built in each — in the full case study.

All case studies — products we built end-to-end

// 03 · Technology

Our stack

A full stack — from modern frontend, through headless CMS and commerce, to CDN, search and analytics. We pick the right tool for each site.

  • Frontend
    • Next.js
    • React
    • Vue
    • Astro
    • Remix
    • TypeScript
  • CMS
    • Sanity
    • Contentful
    • Strapi
    • Headless WP
  • E-commerce
    • Shopify
    • WooCommerce
    • Medusa
  • Hosting & Edge
    • Vercel
    • Netlify
    • Cloudflare
    • AWS
  • Search
    • Algolia
    • Meilisearch
    • Elasticsearch
  • Analytics
    • Google Analytics
    • Plausible
    • PostHog
  • Performance
    • Edge Functions
    • CDN
    • Image Optimization
    • ISR
  • SEO & Schema
    • Schema.org
    • OG
    • Sitemaps
    • Hreflang

// 04 · How we work

How we work

  1. 01STEP 1

    Understand the goal

    Brochure, sales site, platform? Who's the audience? What's the single most important action? Everything else builds from there.

  2. 02STEP 2

    Structure & UX

    Sitemap, information architecture, user flow. Three clicks max to the goal.

  3. 03STEP 3

    Custom design

    Our designers start every site from a blank Figma page. No templates, no "what everyone else is doing."

  4. 04STEP 4

    Frontend & backend dev

    Two-week sprints. Something to see at the end of each one. No surprises, no "we'll show you in two months."

  5. 05STEP 5

    SEO & speed optimization

    Pre-launch: Core Web Vitals, accessibility, on-page SEO. All green. No compromises.

  6. 06STEP 6

    Launch & migration

    Move from staging to production. Domain, SSL, analytics — all handled.

  7. 07STEP 7

    Ongoing support

    A site isn't "ship it and forget it." It needs updates, new features, optimization. We stay with you.

// 05 · Who it's for

Who this is for

  • 01

    B2B companies

    Need a site that signals trust and expertise from the first second.

  • 02

    Startups

    Need a launch site before a funding round or going to market.

  • 03

    Established businesses

    With an old, slow site that needs a serious upgrade.

  • 04

    Stores

    Want to launch online commerce — or upgrade what they have.

  • 05

    Founders

    With a platform idea — and a need for someone to build it from scratch.

// 06 · Why us

Why us

  • 01

    Design + dev under one roof

    A full UI/UX team in-house. You won't need an outside designer who doesn't speak code.

  • 02

    Obsessive about speed

    Our PageSpeed scores are green. Because every second of slowness is a customer walking away.

  • 03

    SEO from day one

    Built for rankings from the foundation. Not bolted on after the fact.

  • 04

    Service that doesn't end at launch

    Monitoring, alerts and fixes — within the scope and service hours set in the maintenance agreement.

// 07 · FAQ

Frequently asked questions

  • 01.How much does a website cost?

    It depends on scope. A brochure site — ~$7K-$17K. A site with custom features — $17K-$45K. A custom web platform or store — $30K-$120K. We give a precise quote after one discovery meeting.

  • 02.How long does it take to build a website?

    A brochure site — 4 to 8 weeks. A more complex site — 2 to 4 months. A web platform — 3 to 6 months.

  • 03.WordPress or custom development — which is better?

    It depends. WordPress is great for content sites and blogs. For platforms, web apps, or advanced experiences — custom development with Next.js gives a faster, more secure, more precise result. We'll tell you straight what fits.

  • 04.Does this include design?

    Yes. For us, design and development are one process. A full in-house UI/UX team.

    UI/UX design — how our design process works

  • 05.Does this include SEO?

    Technical SEO — yes, built into every site we ship. Content optimization and authority building — ongoing work we can manage with you.

    SEO + GEO — organic search and visibility in AI engines

  • 06.What about GEO — optimization for AI?

    Yes. We build sites that appear in answers from ChatGPT, Claude, and Perplexity. In some niches it's already 30% of traffic — and only growing.

  • 07.Does this include hosting?

    We recommend modern hosting (Vercel, Cloudflare, AWS) and help with setup. The control and ownership stays with you.

  • 08.Do you build bilingual sites (Hebrew/English)?

    Yes. Multilingual is something we do a lot. RTL/LTR, translations, dedicated SEO per language.

  • 09.What about maintenance?

    Monthly maintenance packages available — security updates, backups, small features, fixes. Or on-demand.

  • 10.Who owns the code and the site?

    You do. No "buts." Code, domain, hosting accounts — all yours.

  • 11.I have an old site — can you upgrade it?

    Yes. Sometimes a smart upgrade beats building from scratch. We'll audit, recommend, and execute.

    Legacy upgrades — modernizing old sites and systems

  • 12.Do you build content sites and blogs?

    Yes. With a headless CMS for easy editing, and a structure Google and AI both love.

  • 13.What is the difference between a customer portal and an internal system?

    The user and the business action. A customer portal serves people outside the organisation — they have had no training, they compare you with competitors, and their experience is part of the product; so it needs simple sign-in, self-explanatory screens and self-service. An internal system serves employees who can be trained, with multi-step processes, role-based permissions and management reports. Often both run on the same engine — with a separate interface for each audience.

  • 14.How is data moved from an existing system into the new web system?

    With a migration plan written before development, not after it: what moves, what stays in the old system, how mismatched fields are mapped, and how you verify everything arrived — usually with a trial run on a copy of the data, and sometimes a parallel-run period in which both systems operate side by side. The quality of the existing data is what determines the size of this work, which is why it is checked early.

Last reviewed:

// 08 · Let's talk

Need a website that actually works?

A 30-minute intro call. No strings attached. We'll figure out together if we're a fit.

Straight. To the point. On time.