Drupal 7 → 11 in 2026: the migration playbook — AI Robos
HOME / BLOG / ARTICLE
DRUPALJUN 18, 2026 · 12 MIN READ

Drupal 7 → 11 in 2026: the migration playbook.

What actually breaks, what's safe to defer, and the audit-first path we use on estates with millions of nodes — refined over twenty years of Drupal migrations.

If you're still on Drupal 7 in 2026, you're running years past end-of-life on extended support that gets more expensive every renewal. The good news: the migration path to Drupal 11 is the most mature it has ever been, and going 7 → 11 directly — skipping intermediate hops — is now the standard route. Here's the playbook we run.

1. Audit before you estimate

Every failed migration we've rescued started with an estimate made before an audit. Two weeks of inventory changes everything:

Content: which types and fields are actually used? Estates routinely carry 40% dead content types.
Modules: for each contrib module — D11 equivalent, core replacement, or genuinely custom need?
Custom code: most D7 custom modules exist to do things Drupal 11 does in core. Delete before you port.

2. Migrate content with scripts, not hope

Core's Migrate API handles the standard cases; the value is in what you wrap around it. We write per-content-type migration scripts with automated verification at each step: node counts reconciled, broken references flagged, media deduplicated, URL aliases preserved and redirected. On a 2.3M-node state government estate, that verification layer is what let us say "zero lost content" in the contract — and mean it.

A migration you can't verify is a migration you'll be re-running in production, at night, with the client on the call.

3. Rebuild the theme; don't port it

D7 themes don't translate — PHP templates to Twig is a rewrite whatever a migration tool promises. Treat it as the opportunity it is: a design-system pass with Single Directory Components, WCAG 2.2 AA built in rather than bolted on, and Core Web Vitals as an acceptance criterion. Editors get a component library; the next redesign stops being a re-platform.

4. What's safe to defer

Not everything belongs in phase one. Headless front-ends, search overhauls, personalization, AI-assisted editorial — all cleaner to add on a stable Drupal 11 base than to bundle into the migration. The discipline is a two-column list: "needed to launch" and "easier after launch." Defend it weekly, because every stakeholder tour adds a wish.

5. Go live without drama

Content freeze windows measured in hours (not weeks), a delta-migration for anything created during the freeze, redirects tested against the top 10k URLs by traffic, and a rollback plan nobody expects to use. Launch on a Tuesday morning. Boring launches are the deliverable.

Ashish
Founder & CEO, AI Robos
Talk to us
NEXT ARTICLE
React Native vs Flutter in 2026: how we actually choose →
All articles