A page that ranks well at launch and slips about a week later is usually settling, not failing. Google tends to give a fresh page a temporary lift, showing it higher than its long-term signals would earn, so it can gather real interaction data fast. Once enough of that data arrives, the page settles to the position its actual relevance and authority support. The early high was a trial, the later position is the verdict, and the gap between them is the system learning, not punishing.
This is observed behavior rather than a documented promise, but the pattern is consistent enough to plan around. A brand-new URL has thin evidence behind it, so the engine cannot yet tell how well it serves the query. Floating it up briefly is an efficient way to collect clicks, dwell, and refinement behavior from live searchers. After a few days to a week or so, the page has been seen enough times for those signals to outweigh the initial boost, and it drops into its earned slot.
The one-week timing falls out of that mechanism directly. The trial lasts roughly as long as it takes to accumulate meaningful data, which depends on how much traffic the query draws, so a high-volume term settles faster and a quiet one can take longer. A page that holds its launch position is one whose real signals justify it. A page that slides is one whose trial flattered it, which is information, not a sentence.
This is why the drop should not be filed as a penalty or a glitch. A penalty follows a violation and a glitch is a malfunction, but settling is the normal arc of a new page finding its level. Reacting to it as damage leads to thrashing the page right when it needs time to prove out, which only resets the clock on the data the engine is trying to collect.
When your next page launches strong, note its launch position but treat the level it holds after a week or two as its true baseline, and judge whether the page needs work from that settled number rather than from the trial high.