A significant rewrite re-triggers how search engines evaluate the page, and the outcome cuts both ways: it can lift the page if the new version fits the query better, or dip it temporarily (sometimes for weeks) while the new content settles and gets re-assessed. The risk is not uniform, and the catch is that the better a page already ranks, the more a big rewrite has to lose.
The mechanism is worth understanding before you decide. When you change a page modestly, search engines mostly carry forward what they already concluded about it. When you change it substantially, you have effectively presented a new document for the same URL, and it gets weighed again against the competition. That re-evaluation is where the swing comes from. A page that was underperforming has room to gain because the old assessment was holding it back. A page that was already winning has further to fall, because the rewrite puts a working result back into play and the settling period can land lower before it recovers, if it recovers.
This is why “rewrite freely, it can only help” is the wrong frame. It can help, and often does when the page was genuinely weak, but it carries a real cost-benefit that scales inversely with current performance. Rewriting a page stuck on page two is mostly upside. Rewriting a page sitting comfortably at position three is a gamble, because you are risking a stable, earning position in exchange for a chance at a marginally better one, and the settling dip is real money while it lasts.
None of this argues against rewrites. It argues for matching the size of the change to how much the page has to lose, and for going in with eyes open about the temporary instability that follows a major rewrite rather than expecting an immediate, monotonic improvement.
Before you commit to a big rewrite, check where the page currently ranks and what it earns. If it is underperforming, rewrite with confidence. If it is already performing well, weigh the likely gain against the settling-period risk, and consider whether a smaller, targeted edit gets most of the benefit at a fraction of the exposure.