The redirects are in place, the site has been switched over, everything looks good. Two weeks later the curve in Search Console points downwards, and the question nobody can answer is on the table: is this normal? This article is about exactly that part — not preparing a relaunch, but the weeks that follow it.
One thing we treat as already done: every old address needs a new destination, and that mapping is made before the switch — how we set it up is described on our website relaunch page. Everything here starts at the point where that is handled and the numbers still do not look the way they should.
Without a baseline, a relaunch cannot be judged
The most common reason a relaunch gets debated for months instead of assessed is mundane: there is no before figure. People remember that "there was more going on", but there is no number to test that against.
Two things make this unrecoverable later. First, Google Search Console keeps its performance data for only about 16 months. Second, that data is tied to addresses and search queries: if you no longer know which pages had clicks and for which queries, you are searching a report that does contain answers — but only to questions you can no longer ask.
We work through this list before every switch-over. It takes about an hour:
- The Search Console performance data as an export, covering the last twelve months — once broken down by page, once by query. Two files that sit on your own computer, not only in the account.
- The twenty pages with the most impressions and the twenty with the most clicks. They are rarely the same twenty, and that difference is exactly what becomes informative later.
- For your five most important pages, the queries they are found for, with impressions, clicks and average position.
- The number of indexed pages from the page indexing report, with the date beside it. One number, written down, nothing more.
- Visitor numbers for the last three months, split by source: search, direct, referrals, social.
- The number of enquiries in the last three months and how they arrived — form, phone, email, Google Business Profile. If that number is recorded nowhere, it is an estimate. Write it down anyway and label it as an estimate.
- The measured loading time of your home page and two subpages, on mobile. Without a before measurement, "faster now" is a feeling.
- A screenshot of the results page for your three most important search terms. The only evidence of what the result actually looked like before — title, description, neighbours.
When you compare later, compare periods of equal length and, where possible, the same season. Setting four weeks after launch against the strongest week of last year produces a drop that does not exist. A trade business in January and the same business in May are two different data sets.
What a normal dip looks like — and what an unusual one looks like
A dip straight after the switch-over is the rule, and the reason is mechanical: search engines know your site in its old form. They have to fetch the new structure, follow the redirects, evaluate the new pages and update their records. That does not happen in a day, and not for all pages at once, but spread over weeks — important pages usually earlier, rarely visited ones considerably later.
An unremarkable pattern looks like this in practice: the decline starts in the days after the switch, affects many pages at once, but is shallow rather than vertical. The number of indexed pages dips briefly and rises again. Individual pages come back at different times, and impressions recover noticeably before positions settle.
What stands out is something else. These five patterns are not proof of a fault, but each one is a reason to look rather than wait:
- Impressions fall to almost zero across the whole site instead of declining. A technical exclusion looks like this; a re-evaluation does not
- The number of indexed pages keeps falling for several weeks instead of steadying
- A clearly defined group of pages does not come back at all, while the rest behaves normally
- The decline does not start on the switch-over day but distinctly later — in which case the cause is probably elsewhere
- After six to eight weeks there is no upward movement at all
The widespread rule of thumb that "after four weeks it is fine again" is exactly that: a rule of thumb. It depends on how often your site is crawled, how many addresses changed and how large the site is. As a rough orientation it works; as a commitment it does not.
Diagnosis in the order that saves time
When visibility does not return, the conversation usually turns to content and copy immediately. That is the most expensive way in, because it costs weeks and skips the more common causes. The reverse order makes more sense — each step is what makes the following ones meaningful:
- Is the site reachable at all? A block for search engines left over from the build phase, a forgotten password protection, or an instruction not to include the pages — the quietest cause, and checked in five minutes.
- Are the pages indexed? The page indexing report gives you not just the number but, for every excluded page, the reason. Reading those reasons is the most productive part of the whole diagnosis.
- Do the old addresses land on the right new ones? A sample from your before list, in the browser. This should have been settled before launch and is still where things go wrong most often.
- Are the important pages still the same in substance? Relaunches shorten text, move sections into images, or spread them across several pages. A page that used to answer a question in full and now only touches on it is a different page.
- Has internal linking become thinner? Tidier menus and leaner footers look better and cut off routes at the same time.
On that last point, a check you can run without any tool: count the clicks from the home page to one of your important pages. What used to be one click away and now takes three gets fetched less often — and if it can only be found through the internal search, not at all. This is the cause most often missed, because looking at the site does not reveal it.
| Symptom | Possible cause | How to check |
|---|---|---|
| Almost all pages lose their impressions at the same time | The site is blocked for search engines, usually a leftover from the build phase | Inspect any address in Search Console: is a block or exclusion reported? |
| The number of indexed pages keeps falling week after week | Pages were dropped, or several old addresses point at the same destination | Open the page indexing report and read through the reasons given for excluded pages |
| Individual, previously strong pages are gone, the rest is stable | Those exact pages were merged or heavily shortened | Open the old address and compare the new destination page with the old content |
| Impressions hold up, but clicks collapse | Titles and descriptions were rewritten during the relaunch | Put the click-through rate per page before and after launch side by side in Search Console |
| Search results still show the old titles | The record for those pages has simply not been updated yet | Inspect the address individually and look at the date of the last crawl |
| Visitor numbers are normal, but no enquiries arrive | The form, or the sending of it, is not working | Submit the form yourself and check the inbox, not the confirmation page |
| Important pages are barely fetched any more | They are harder to reach internally than they used to be | Click from the home page and count how many clicks it takes to get there |
| The drop came weeks after launch, not on launch day | The cause probably is not the relaunch | Overlay the launch date and the date of the drop on the same curve |
The first weeks: one change, then wait
The strongest impulse after a drop is to fix everything at once. Revert the text, rework the titles, rebuild the menu, add a few redirects — all in one afternoon. The problem is not the effort but that you will never learn what worked. And when it happens again next time, you start from zero again.
In this phase we hold to four rules:
- One change, then wait at least two weeks. Before that there is no data capable of proving anything
- Log the date and the change in a single file. Two columns are enough; what matters is that nobody has to guess later
- No new addresses and no structural rebuild during this phase. Both create precisely the movement you are trying to observe
- The exception to all of the above: a technical exclusion gets fixed immediately, not observed
We rebuilt our own website in August 2026 and held to the same discipline — including the uncomfortable part, where you look at a curve for two weeks and do nothing.
What does not come back
One expectation is better said out loud beforehand, because otherwise it disappoints afterwards: not every position returns exactly, and that is not necessarily anyone’s mistake. A ranking is not property. It is the outcome of a comparison, and during the weeks of your relaunch other people changed their pages too.
There is also an effect the relaunch itself creates. If three thin pages were merged into one good one, the mix of queries changes: a few very specific terms drop away, while the remaining page becomes clearer for the more important ones. In a list of positions that looks like a loss. In the enquiries it does not.
So the more robust question after a relaunch is not "are we back at the same position", but: are as many suitable enquiries coming in through search as before, or more? That figure is in your before list — this is the point where the hour of work at the start pays for itself.
Common questions about visibility after a relaunch
A decline in the first weeks is the rule, because search engines have to fetch and re-evaluate the changed structure. It is unremarkable when it is shallow, affects many pages at once, and the number of indexed pages steadies again. A fall to near zero looks different and is usually technical.
That depends on how often your site is crawled, how many addresses changed and how large the website is. The usual order of magnitude is a few weeks, and the recovery does not happen for all pages at the same time. A specific date cannot be promised, and you should be sceptical of anyone who does.
Overlay the launch date and the start of the decline on the same curve. If the decline begins in the days after the switch-over, the connection is likely. If it starts weeks later, look for other causes first instead of continuing to work on the relaunch.
At least four things: which pages get clicks and impressions, for which queries, how many pages are indexed, and how many enquiries recently arrived by which route. Export the Search Console performance data as a file rather than relying on the account — the data there only goes back about 16 months.
Some of them, by fixing the cause — a missing redirect, a technical exclusion, a page that no longer exists. Restoring one particular old position, on the other hand, is not something anyone can deliver, because it always comes out of a comparison with others. The trend in enquiries from search says more.
Free consultation
We look at your website up front and show you the biggest levers for more enquiries.
