Build → Rank → Convert → Operate With Claude Code
Four problems, in the order websites meet them, plus one that arrives when a site has to move, one that belongs to a single platform, and one that belongs to the people who do all of this for clients. Most projects fail at one of them, and each needs genuinely different work — which is the reason there are seven products rather than one.
The order is not arbitrary
You cannot convert traffic you do not have, and you cannot rank a site that does not exist. That sounds obvious and it is routinely ignored: the most common wasted quarter in web work is optimising conversion on a page nobody reaches, or optimising titles on URLs that are about to be canonicalised away.
So the order matters, and knowing which stage you are actually in matters more than any individual tactic.
Which one are you in?
The honest test is what happens when you look at your own numbers.
If you cannot answer "is it live and does it pass its own checks" — you are at Build. Everything downstream is premature.
If it is live and Search Console shows few impressions — you are at Rank. Conversion work on 40 sessions a month cannot be measured, and any change you make will be indistinguishable from noise.
If impressions and sessions are real but nothing happens — you are at Convert, and the first job is finding out where people stop rather than guessing at fixes.
If it is live, found and converting, and the question is keeping it that way — you are at Operate. This is the stage that lasts longest, and the one where the failures are quiet: a certificate that stops renewing, a form that stops delivering, a plugin with a published exploit, a backup nobody has restored. The work is a rhythm of checks, not a project.
If the site has to move — a new platform, a new host, a new domain, a new URL structure — you are at Migrate, and the risk is not the build but what gets lost on the way: URLs, redirects, titles, images, forms, analytics. The work is an inventory, a map and a before/after comparison, with a go/no-go at the end and a rollback that was prepared before the DNS change. It is a project with a start and an end, unlike Operate; it is also the point where most sites quietly lose a year of search traffic.
If you build websites for clients — and the site is fine but the engagement is not: the scope that drifted, the proposal written the night before, the revisions that never end, the handoff that was a folder — you are at Deliver. That is not a stage of the website; it is the operating system of the business that builds them, and it is solved once and reused on every client.
If the store runs on Shopify and the catalogue is the work — products, collections, SEO, metafields, inventory and redirects at a scale the admin cannot keep up with — that is not a stage but a platform. The Shopify Automation & Admin API Toolkit drives the Admin API from Claude Code, with a preview before every write.
If you own the first three for more than one site — you are running the lifecycle, not a stage, and the useful thing is a repeatable system rather than three separate purchases.