Data Story
We rebuilt our website with AI, moved faster, and held our search visibility.
Website · SEO
July 2026
We rebuilt datastory.nz with AI writing every line of the code, to answer two questions at once. How fast can we publish a real site when AI does the building? And can an AI-built site come through a full platform migration, every URL rewritten, without losing the search visibility the old site had earned? So we did exactly that, and watched what happened.
Publish faster, and keep every ranking through a full migration
A traditional website build moves at the speed of its development queue. Design waits on build, build waits on review, and weeks pass before anything ships. We wanted to compress that. Use AI to write the code, iterate in hours instead of days, and get a production site live in a fraction of the usual time.
Speed only counts if the output holds, and a migration is where that gets tested. Rewrite every URL at once and the rankings the old pages earned can vanish if the redirects are wrong. If AI-written code could carry the whole site across that, it could carry anything.
Build with AI, ship on a modern stack
We used AI to write the site on a modern component-based stack, moving from idea to working page in hours. The same speed that built the site let us be thorough where it mattered most: the migration.
Let AI write the build
Design and copy went straight into AI-assisted development on a modern framework, compressing the slow middle of a normal web project.
Map every old URL to its new home
Before launch we generated a redirect for every old address on the site, not a sample, so nothing was left to resolve itself.
Make each redirect permanent and check it lands
Every redirect is a permanent one, and we confirmed each target resolved to a real page rather than a dead end or a chain.
Watch the code prove itself
At launch we instrumented Google Search Console and GA4, so we could see whether the AI-built site held its rankings in real time instead of assuming it had.
Shipped fast, and the rankings came through the rewrite
100%
of the site's code, written by AI
+18%
Organic clicks, like-for-like on the same search queries
13.8 → 12.5
Average Google position through a full URL rewrite
2.8% → 3.8%
Click-through rate from Google search results
Judging a migration means comparing the same searches before and after, not the site totals. Datastory.nz appeared for 177 queries in both the 51 days before launch and the 51 days after. On those queries, average Google position improved from 13.8 to 12.5, and clicks rose 18%.
Those clicks came off 14% fewer impressions, so the improvement is in how often people chose the result: click-through rate went from 2.8% to 3.8%. Same searches, better position, more people clicking.
The site-wide numbers moved in a different direction, and they measure a different thing. Google now surfaces the new site for a much broader set of searches, 265 of which the old site never appeared for at all. Most enter low, around position 20, which drags the site-wide average down even as the established pages rank better. That is a widening, not a loss, and it is why the like-for-like read is the one that answers the question the rebuild was testing.
The real test of AI-written code is not whether it looks right. It is whether it survives contact with production. Ours moved a whole site to a new platform and came out ranking better on the searches that already mattered.
Ready to scale your marketing?
Let's grow your business and reach your digital marketing goals. Let's do this together.
Book a Free Strategy Call
