How Do I Safely Update a Website Someone Else Built?

The job came in as "it's ready and approved, just one step." The client wanted a new page added to their website. We had not built that site, and we had not been the last people to publish it. That turned out to matter more than anything else about the task.

When you take over a website from someone else, you inherit their decisions, their shortcuts and their unfinished notes. Most of the time that is harmless. Sometimes one of those leftovers can take the whole site out of search or wipe most of it, and nothing on screen will warn you.

Trap 1: The Ready Copy Was Hiding the Site From Google

The folder that looked like the finished site was really a draft build. It was set up to keep search engines out, on purpose, so the unfinished version would not show up in search. It did that in two places:

• The file that tells search engines where they may go said, on its first line, that the site was in draft mode and blocked every search engine.
• The homepage and every blog post carried a hidden tag telling search engines not to list the page, with a note saying to remove it at launch.

Those settings had been removed when the real site launched. They were never removed from the draft copy, because nobody expected that copy to be published again.

What we do now: before publishing any folder to a live site, we search it for those hidden "do not list" tags and read the search engine file in full. Then we pull the same files from the live site and compare them. Here the live site was clean and the folder was not.

Trap 2: "Upload Two Files" Would Have Deleted Everything Else

Some websites are published from a saved history of their code, so a small change only touches what changed. Others are published by uploading a folder by hand. This site was the second kind.

On a hand-uploaded site, whatever you upload becomes the entire site. So "just add the new page" done the obvious way, by uploading the new page and the file that links to it, would have left the client with a website of two files.

Trap 3: Doing It Safely Takes One Extra Step

The hosting service this site used lets you publish by sending a list of every file on the site, each with a unique fingerprint. Any file whose fingerprint it already holds is kept exactly as it is and is not sent again. It then asks only for the files it does not have.

We also checked the homepage and the search engine file by their fingerprints before publishing, so a mistake in the list would stop the job instead of quietly dropping a page.

Trap 4: We Did Not Have the Real Homepage

Our copy of the homepage did not match the live one. After undoing every difference we knew about, a small part still did not match. We could not rebuild the client's live homepage from the files we had.

It is tempting to fix that by copying the homepage from the browser and uploading it back. We did not, because the hosting service changes pages slightly as it serves them. Uploading the served copy would bake those changes into the site's own files.

So we made the change smaller instead of bigger. The new page went live on its own, and we skipped adding it to the site menu, because the menu lived in the file we could not rebuild. The client's emails linked straight to the new page, so the menu link was polish, not the requirement.

One More Lesson: Check Before You Report

The same night, several jobs on the list were already done, finished earlier in another session. Each one took a single check to confirm, and skipping that check would have meant wasted work and a wrong update. It also went the other way: twice, something reported as finished was not. Both mistakes cost trust. We check the live site or system, not the note about it.

Questions to Ask Before Anyone Touches Your Site

• Did you build this site, or were you the last to publish it? If not, what are you checking first?
• Is this site published from saved code, or uploaded by hand?
• Have you compared the files you are about to publish with what is live right now?
• Have you checked for any "do not list in search" settings left over from a draft?
• If you cannot rebuild a page exactly, what will you leave alone?

Common Questions

Can updating my website really remove it from Google?

Yes. A draft or test copy of a site often carries settings that tell search engines to stay away, so it does not show up in search before it is ready. If that copy is published over the live site, those settings go live with it. The pages still look perfect to you and your customers, but search engines are being told to drop them.

Why would uploading one new page delete the rest of my site?

Some sites are published by uploading a folder by hand, with no saved history of the code behind them. On that kind of setup, whatever you upload becomes the whole site. Upload a folder with only the new page in it and the site becomes only that page. Check how the site is published before you touch it.

What if nobody has the original files for my website?

Then make the change smaller, not bigger. Add what is new on its own and leave the pages you cannot rebuild exactly as they are. Copying the live pages from the browser and uploading them back is not safe, because the hosting service changes pages slightly as it serves them. Delivering most of the job safely beats delivering all of it on a guess.