Rebuilding an existing website is a different job from building a new one. You already have content that works, addresses that other sites link to, and pages that rank. Losing any of those costs more than the rebuild saves.
Short answer
A rebuild has to protect what already works before it improves anything. Record which pages get visits and links, keep their addresses or redirect them properly, and carry the content that performs. Generation helps with the pages that were always thin, and the pages that already earn should be moved rather than regenerated.
Export the list of pages that get visits, the pages that get search impressions, and the pages that other sites link to. Those three lists define what you are not allowed to break.
Most small sites have a handful of pages doing nearly all the work, and they are rarely the ones the owner expects. Finding out before the rebuild is far cheaper than discovering afterward.
Our page on auditing a site covers how to assemble that picture, and our page on the performance report covers where the search half of it comes from.
Every address that changes is a risk. Links from other sites point at the old one, a search engine or an answer engine has to relearn the new one, and anything you miss becomes a dead page.
Keeping the same addresses removes the whole problem. Where a new structure genuinely improves the site, map every old address to its closest new one before anything goes live.
Then redirect properly, one hop, permanent. Our page on redirects covers the difference between the two kinds and why chains cost you.
Carry anything that performs. A page with links and visits has earned its position, and replacing it with a fresh version discards the evidence that it works in exchange for nothing.
Regenerate what was always thin. The service page nobody finished, the empty section, the pages that exist as placeholders. Those are where generation genuinely adds something.
And rewrite what is wrong. Old prices, old coverage, credentials that lapsed. A rebuild is the moment those get fixed, because nobody will go looking for them later.
This order protects what works first and improves second. Reversing it is how rebuilds lose visibility, and the loss usually shows up weeks later when it is expensive to diagnose.
Check that every old address resolves to something sensible. A sample is not enough on a rebuild, because the missed ones cluster in the parts of the site nobody thought about.
Submit the new sitemap and watch coverage. Our page on the coverage report covers what to look for, and the first two weeks are when problems surface.
Expect a dip. Some movement after a rebuild is normal even when everything was done correctly, and the mistake is reacting to week one by changing things again.
Changing addresses for tidiness. A cleaner address is worth very little and costs every link pointing at the old one. Keep the address unless the change fixes a real problem.
Dropping pages quietly. A page nobody remembers may still be the one another site links to, which is why the export comes before the decisions rather than after them.
And regenerating the pages that were working. Those pages are evidence of what your audience responds to, and a rebuild is the worst possible moment to discard that.
A rebuild does not have to be a single event. Improving the worst pages one at a time carries almost no risk and produces most of the benefit, which suits a site that is already performing.
The case for doing it all at once is structural. Where the whole shape of the site is wrong, fixing it page by page never converges and the mess persists.
Our page on rebuilding a site covers the wider decision, including how to tell which of the two situations you are actually in.
It can, if addresses change without redirects or if performing pages are replaced. Done carefully, with every old address mapped and the pages that work carried over, the effect is usually a short dip and then recovery.
Wherever possible, yes. Every changed address costs the links pointing at it and has to be relearned. Change one only when it fixes a real problem, and redirect permanently in a single hop when you do.
The ones that were always thin or never finished. Pages with visits and inbound links have earned their position, and replacing them discards evidence that they work in exchange for nothing.
Usually a few weeks, and it varies with site size. Some movement in the first days is normal. The common mistake is reacting to week one by changing the site again before anything has stabilized.
15-day free trial. Card required. Cancel before day 15 and you pay nothing.