On a site with existing pages and existing authority, the first movement usually shows inside four to eight weeks, and it comes from fixing what is already there rather than from anything new. On a genuinely new domain competing for a term with real money behind it, six to twelve months is the honest range, and anyone shortening that in a pitch is describing a different term than the one you asked about.
The useful version of this question is not how long until it works. It is what should have moved by when, so you can tell a slow start from a dead one.
Weeks one to four: impressions, not clicks
The first thing that moves is impressions on terms you were not previously visible for at all. That is Google testing you, and it happens before any click arrives. If impressions are flat after a month of published work, something structural is wrong: the pages are not indexed, the crawler is blocked, or the pages are duplicating something you already had.
Weeks four to twelve: the existing pages move first
Almost always, the pages that already existed move before the new ones do. That is the shape of the thing. On Phrase Studios, our own label, the work started with rewriting what was there rather than publishing more, and over six weeks search sessions went from 177 a month to 2 711 while Google impressions went from 3 652 to 89 340. The homepage alone gained 872% more clicks. It was a small base, and that is stated plainly on the case study, along with the Search Console exports underneath.
If impressions have not moved in a month, waiting six more will not help. Something structural is wrong.
Months three to six: the new work starts to compound
Published pages take time to accumulate the signals that move them up: internal links from later pages, other people citing them, and enough impressions for Google to have an opinion about whether people are satisfied. This is the stretch where an engagement feels slowest and is usually working hardest, and it is also where most people give up.
What makes it faster, and what genuinely does not
Faster: an existing domain with history, a site that is already technically sound, a topic where you have real first hand experience to write from, and internal linking done deliberately rather than by a related posts widget. Not faster: publishing more of the same thing, keyword density, or paying for links, which buys a risk rather than a ranking.
One thing that has genuinely changed the timeline is AI search. Citations inside AI answers can move much faster than rankings do, because the assistants are assembling an answer rather than ordering a list. On the same brand, AI feature impressions went from 290 to 23 503 over the same six weeks, which is a different mechanism worth understanding on its own.
When to conclude it is not working
Three months with no movement in impressions, on a site where pages are indexed and crawlable, is a real answer. So is six months of rising impressions with no clicks, which usually means you are ranking for the wrong intent: visible on questions asked by people who will never buy. Both are diagnosable long before the twelve month mark.
What we do with it
We fix and rewrite what exists before publishing anything new, because it moves first and it tells you whether the site can move at all. The channel is on the SEO and GEO page.