Retail, ecommerce and professional services
Online shops, catalogues and service sites rebuilt as static HTML that search engines and AI assistants can read, with product, price and shipping rules in your own database, feeds for Google Merchant Center and Meta, and senior maintenance after launch.

What usually goes wrong
- We redesigned the shop last year and sales did not move. Nobody can tell me why.
- Customers ask ChatGPT what to buy and it names three competitors, never us.
- The catalogue lives in three places: the shop, the feed, and a spreadsheet the warehouse keeps.
- Some orders cost us more in shipping than we charged, and we only find out at the end of the month.
- Every price or stock change goes through the agency and takes two weeks.
- Merchant Center keeps rejecting items and nobody here can read the errors.
- The photos look right in the studio and take forever to load on a phone.
What we build
- A storefront on infrastructure you own: catalogue, series, variants, cart, checkout, customer accounts and orders, with your repository and your own Supabase and Vercel projects.
- Category, series and product pages pregenerated as real HTML, each with its own metadata and JSON-LD, protected by build guards so a page can never quietly go empty.
- Product feeds regenerated on every build for Google Merchant Center, Meta and Pinterest, plus sitemap, robots policy for search and answer crawlers, and Search Console setup.
- Shipping and pricing rules in the database, so the estimate a customer sees and the amount charged on the order come from the same calculation.
- Card payments through your own bank virtual terminal or gateway, with the amount re-read server-side, the signature verified and repeat notifications made idempotent.
- An admin panel for your team: orders, products, prices, stock, series, shipping and role-based invitations, instead of a plugin ecosystem you then maintain.
- Migration off WooCommerce, Shopify or a legacy CMS, verified item by item, followed by a CARE retainer with ServiceDesk and a written monthly report.
What changes in this sector
Two things changed at the same time. People now ask an assistant before they open a shop, and most shops are invisible to it. GPTBot, ClaudeBot, PerplexityBot and OAI-SearchBot fetch raw HTML and do not run JavaScript. When we tested the store of a premium tableware and home linen brand in Madrid in August 2026, all 65 of its URLs returned zero characters of body text. The catalogue only existed after JavaScript had run, so to an assistant the shop had no products at all.
The second change is less visible. In retail the margin lives in the plumbing: carrier tariffs by weight and zone, VAT on a preparation fee, strike-through prices with no genuine previous price behind them, stock that disagrees with the warehouse. None of that is a design problem. It is money leaving the business one order at a time, and it only surfaces when somebody reads the rules line by line.
In B2B and professional services the same problem takes a different shape. There the catalogue is the work itself: references, articles, specifications. KD.tech, a Spanish semiconductor company, had 327 bilingual pages worth reading and a site scoring 69 out of 100 for SEO on mobile. Same fix, different content.
What we build
We rebuild the storefront on infrastructure you own: Vite, React and TypeScript on the front, Supabase for Postgres, storage, authentication and edge functions, Vercel for hosting, and your repository on GitHub. Every category, series and product page is pregenerated at build time with its own title, description, canonical, Open Graph tags and JSON-LD, and a build guard fails the build if a page drops below 400 characters of text, so readability cannot quietly regress.
Product data goes where machines can read it: variants, prices, availability and images in one database, feeds regenerated on every build for Google Merchant Center, Meta and Pinterest, a sitemap built from the same route list as the site, and a daily job that reruns the prerender so static pages follow stock and price changes.
Conversion work here is engineering, not persuasion. Shipping is calculated in the database by weight and zone, so the estimate in the browser and the amount charged on the order come from the same function. Card payments run through your own bank terminal with the amount re-read on the server, the signature verified and repeat notifications made idempotent. Images are regenerated as WebP at three widths. Your team changes prices and stock from an admin panel instead of calling us.
What it costs
A shop with catalogue, checkout, payments and an admin panel is a fixed price from 12,000 EUR, and usually lands between 12,000 and 25,000 EUR depending on catalogue size and integrations, payable in six to ten monthly instalments. A catalogue or reference site without checkout starts from 7,600 EUR. Both begin with a Discovery Sprint at a fixed price from 4,900 EUR, half of it credited against the build if it starts within 60 days. After launch, a CARE retainer from 850 EUR per month covers maintenance, evolution and the ServiceDesk, and search and AI visibility work runs inside the same contract.
How we prove it
We publish what we can attribute, and nothing else. On the tableware rebuild the catalogue went from zero to 105,212 characters of readable HTML across 65 URLs, 24 products and 93 variants were migrated through the public WooCommerce Store API and checked one by one, and 137 images were regenerated as WebP at three widths. Five audit passes ran before launch and more than 70 findings were fixed, among them a VAT error on the preparation fee worth about 80 euros per 100 orders. The shop has not traded long enough to report sales, so we quote no sales figures.
Proof
Questions, answered
How much does a custom online shop cost?
Can you connect the shop to our ERP, our warehouse and our payment provider?
Why do ChatGPT and Perplexity never mention our products?
Do we have to leave Shopify or WooCommerce?
Who looks after the shop after launch?
Start the engine
Tell us what you are building in a few short questions. A senior engineer answers in writing within 48 business hours, with a first take on scope, timeline and price.
