The deployment is done. It feels like the end of the project, but whether a redesign worked usually becomes clear in the two weeks after launch. Deciding in advance what to watch — and what to ignore — keeps you from debating a rollback every time a number twitches.
Axis 1 — indexing
Three things in your search console: are 404s rising, are the new addresses being indexed, and did the redirects take effect as intended.
Rising 404s are the clearest signal. If addresses moved and something was missing from the redirect map, it shows up here — and the response is simple: add the missing address to the map. Re-submitting the sitemap right after launch speeds up discovery of the new URLs.
Rankings moving around for a few days is ordinary; the search engine is re-evaluating changed pages. But if two weeks pass with no sign of recovery and the decline is concentrated in one group of screens, that is not re-evaluation — check that group’s addresses, titles and internal links first.
Axis 2 — speed
New designs tend to gain weight. Another font, larger imagery, a little animation. Each is small; together they delay the first screen.
Measure the same screens under fixed conditions. Pick three top entry pages, run three measurements on the same device profile at the same time of day and take the median. Never judge on a single reading — network conditions alone move the number substantially.
There is a precondition here: comparison requires that you measured before the redesign. Measure for the first time after launch and you have no way to tell whether the number is good. If nothing else, record a baseline now for the next redesign.
Axis 3 — conversion
The first thing to check is not a metric but whether the form actually arrives. Enquiry forms dying quietly after a redesign is extremely common — mail settings not carried across to the new environment, a notification address left at an old value, or spam protection blocking legitimate submissions.
Submit one yourself and check the inbox and the spam folder. If there is an auto-reply, confirm that arrives too. Do this on launch day, and once more a few days later.
Then look at the funnel. Check each step on the conversion path for increased drop-off. If the call to action moved, the change shows on that screen; if a form gained fields, it shows on the form.
Normal, and not normal
Post-launch observation and reading the numbers are covered further in the SEO archive, and continuous monitoring of the state after deployment is what an InfraGuard pass is for.
Next part
If what you observe is bad, you have to go back. The final part is the actual rollback procedure — code and database, and the order to reverse them in.