Quality methodology

Quality is not promised, it is verified.

Written standards, automated checks on every delivery, review by an architect and a release checklist before every go-live: the care we take with your platforms is a method, not an intention.

Our conviction

Care shows in what nobody looks at.

A website can look successful on launch day and become costly within months: a forgotten permission, a disabled cache, a test configuration left in production, a back-office nobody knows how to use. These invisible details make the difference between a platform that lasts and one that will have to be rebuilt.

That is why we have written down our requirements, rule by rule, and tooled their verification. Every rule has an identifier, is cited in our reviews and reports, and applies the same way to all our projects, whatever the team.

Our standards

Seven standards, over 170 written rules.

Our standards are not habits: they are versioned documents, maintained by our architects and applied on every project.

Architecture

Data model, entities, fields, pages, configuration: how a Drupal platform is built to stay consistent and evolve.

Editor experience

Readable admin menus, tables with every useful column, filters, bulk actions, a dashboard: a back-office designed for the people who publish.

Release gates

Thirty-six checks to pass before every delivery, from code quality to go-live. The audit report serves as the release checklist.

Security

The verifiable application of our security policy: access, workstations, projects, third-party tools, AI agents, incidents.

SEO

Markup, metadata, URLs, indexing, Core Web Vitals: the technical requirements that make a website visible, on Drupal as on Next.js.

Development and Git

Drupal coding standards, repository hygiene, branches, commits, reviews, releases: shared practices so that anyone can pick up someone else's work.

Every delivery

Three levels of control on every change.

No line of code reaches production without passing three complementary filters. Tools settle the mechanical, people keep the judgement.

1. Automated checks

On every proposed change: Drupal coding standards, static analysis, strict typing on the Next.js side, dependency audit. What fails cannot be merged.

2. AI-assisted review

An agent reviews every change against our standards: architecture, logic, risks, citing the rule concerned. It does not re-detect what tools cover; it provides a first analysis.

3. Review by the lead architect

The architect responsible for your project is automatically assigned as reviewer and is the only one who can approve. The final decision is always human.

Before every go-live

The release checklist, point by point.

Our release gates cover everything that must be true before a version ships. Here is the essence.

Quality

No errors or warnings in the logs, coding standards met, architecture and back-office standards reviewed.

Security

Permissions checked role by role, back-office tested with every role, no pending security advisory, security headers configured, development modules absent.

Configuration

Environment differences isolated, editorial configuration protected, translations checked, error reports hidden in production.

Performance and robustness

Compression and caching enabled, scheduled tasks in place, healthy status report, logs sent to the system.

Going live

Sitemap, redirects from old URLs, indexing open, email sending verified, anti-bot protection and cookie management working.

Project

Compliance with the specifications validated with you, manual operations communicated, no links to test environments.

From development to operations

A controlled delivery flow, then monitored.

Sprints and continuous acceptance testing

Developments are delivered in iterations and tested by your teams as they go, not only at the end. Go-live is approved by your acceptance testing.

A documented chain of environments

Development, staging, production: every project has its environments and branches described, and an automated check ensures no change reaches production without going through staging.

A three-month warranty

After go-live, defects are fixed under warranty for three months, before the platform moves to maintenance.

Continuous monitoring

Security vulnerabilities, overdue updates, security headers, performance and accessibility, production errors: the platforms we maintain are scanned continuously.

AI and quality

The same rules for people and for agents.

We use AI in our production, within a framework: it speeds things up, it does not decide.

Standards readable by agents

Our standards are exposed to the AI tools working on your projects: they produce code that follows the same rules as our developers.

An AI usage charter

Every change declares how AI was used, and a developer attests to having reviewed it. Without that attestation, nothing can be merged.

The agent proposes, people decide

Code review, monitoring, improvement proposals: AI analyses and suggests, an architect or a project manager decides.

Neovia Academy

Living rules, carried by a trained team.

Our standards evolve with Drupal, Next.js and our projects, and the team keeps training through Neovia Academy, our internal platform: code quality, security, best practices. The same requirements, from one project to the next, whoever is involved.

Discover Neovia Academy

FAQ

Quality: your questions.

What is a release gate?

A release gate is a condition that must be met before a version of a website can be delivered: for instance, no pending security advisory, or permissions checked for every role. Our thirty-six release gates form the release checklist of every delivery, and an audit report attests to it.

Do your standards apply to a website you did not build?

Yes. When taking over, we audit the platform against our standards, which gives a precise review and a prioritised list of fixes. The rules then apply to every change.

How do you check code quality?

At three levels: automated checks on every change (coding standards, static analysis, dependencies), an AI-assisted review that cites our rules, and the review of the project's lead architect, the only one who can approve.

Do you use AI to develop our projects?

Yes, within a written framework: our standards are exposed to AI tools, every change declares how AI was used, and a developer attests to having reviewed it. AI speeds things up, it does not decide.

What happens after go-live?

A three-month warranty covers defects, then the platform moves to maintenance: updates, continuous monitoring of security, performance and errors, and enhancements.

Can we read your standards?

They are internal, but we are happy to walk you through them, and the audit reports on your platform are meant for you.

Client experience

A partner engaged at every step

Discover how we support our clients, from strategy to production — a genuine human collaboration.

Discover the Neovia client experience→

What digital experience for your customers?

Let's talk about your needs. We'll show you how we can support you throughout your project.

Get in touch