Every web project workflow lists roughly the same eleven steps. They are not equally risky. Three of them absorb almost all the pressure when a deadline moves, and what happens to those three determines whether the launch was a success or just an event.
Nobody hand-writes SEO for fourteen thousand products. The workable approach is to sort pages into three tiers — templated, written, and excluded — and spend real effort only where it changes anything. Here is how I decide the tiers and what goes into each.
A build is finished when someone who was not there can change it without breaking it. This covers what to standardise before the first page, what to leave behind at handover, and a simple six-month test for whether the structure held.
A menu page is not a design problem, it is a maintenance problem. Three ways to structure one in WordPress, what each costs the client after launch, and the questions that decide which is right for a given restaurant.
A product CSV is not a transport format, it is the deliverable — and spreadsheets damage it in ways that stay invisible until import. Here is the prep sequence I run on supplier data, the validation column that replaces manual checking, and the point where a spreadsheet stops being the right tool.
A catalog migration is decided before a single row is imported. This is the field-mapping method I use on large WooCommerce catalogs, the four decisions that have to be made per field, and the validation passes that catch problems while they are still cheap.