The reason that actually holds up
The usual pitch for leaving Amazon is the referral fee. It is the weakest argument on the list, because once you have paid for traffic, a checkout, fulfilment and returns yourself, the fee you saved is mostly spent again — and you have taken on the work Amazon was doing.
The real reason is that on Amazon you never own the customer. You do not get their email address. You cannot tell them a new product exists. You cannot win them back after a bad month, run a subscription, or find out that a third of your buyers are the same fifty people. Every one of those is normal on your own store and structurally impossible on the marketplace.
So the honest framing is not "escape Amazon". Most brands should run both — Amazon carrying the volume and the discovery, the site carrying the brand, the margin and the relationship. What the site changes is that your business stops being a tenancy.
Shopify unless there is a reason not to
For most brands Shopify is the right answer and we will say so before you have paid us anything. It launches faster, costs less to run, and its checkout converts better than almost anything built from scratch — which matters more than any other single page on the site.
Custom is worth it when there is a requirement Shopify genuinely cannot meet: unusual product configuration, a pricing model that is not per-unit, an integration with a system that has no app. That is rarer than the agencies pitching custom builds suggest, and the tell is whether they asked what you need before recommending it.
Where the build effort actually pays back is the product page and the path to it — the gallery, the variant picker, the shipping and returns answer, and how fast it all appears on a mid-range phone. A beautiful homepage nobody scrolls is the most common place a budget goes to die.
Migrating without losing your rankings
If you already have a site, the redirect map is the part that decides whether the move costs you anything. Every existing URL that has traffic or links pointing at it needs a 301 to its nearest real equivalent on the new site — not a blanket redirect to the homepage, which Google treats as a soft 404 and drops.
This is the step that gets skipped, and it is why brands relaunch a better-looking store and watch organic traffic fall off a cliff the same week. We build the map from your existing analytics and Search Console data before the new site goes live, and check it after the switch rather than assuming it worked.
The same discipline applies to the things nobody thinks of: product schema, canonical tags, the sitemap, and the image URLs — which carry their own search traffic. We wrote up the wider case for a direct site here.