Help Centre

Search engine visibility and robots.txt

Hide your marketplace from Google while you build it, show it again at launch, and change the robots.txt file search engines read.

Search engines find your pages with programs called bots or crawlers. Most of the time you want them to: that’s how people find your marketplace. But while you are still building, or for a private community, you may want to keep them away. Yclas gives you two controls: a switch that asks search engines not to list your site, and the robots.txt file that tells them which parts to visit.

Hide your site from search engines

  1. In the admin panel, go to Settings › General.
  2. In Access & privacy, switch on Hide from search engines.
  3. Click Save changes.

Every page now carries a “noindex, nofollow” instruction. Google, Bing and other well-behaved search engines won’t add your pages to their results, and will drop pages they already have the next time they visit. Your site stays open to people: anyone with the address can still use it.

To appear in search results again, switch Hide from search engines off and click Save changes. Then submit your sitemap in Google Search Console to speed things up. See Get your site into Google.

Forgetting this switch is the most common reason a launched marketplace never shows up in Google. Add it to your launch checklist.

Which option to use

You want to… Use
Keep a site out of Google, but open to visitors Hide from search engines
Close the site to everyone but your team while you build Maintenance mode
Open the site only to signed-in members Private site

Maintenance mode and a private site keep search engines out on their own, since they can’t see the pages either.

Your robots.txt file

robots.txt is a small public file at yoursite.com/robots.txt that tells crawlers which addresses they may visit. Yclas writes a sensible one for you:

User-agent: *
Crawl-delay: 10
Disallow: /search.html
Disallow: /user$
Disallow: /user?
Disallow: /*?*sort=
Sitemap: https://yoursite.com/sitemap.xml

It asks crawlers to slow down a little, skips search-result pages and sorted lists (every combination of filters would otherwise be a new page to crawl, which wastes their time and slows your site), skips the member directory, and points them to your sitemap. Your listings, categories, pages and seller profiles stay open. (The search address follows your site’s language, so it may not be /search.html on your site.)

Change it

You can replace the file with your own, for example to block a particular bot that is hammering your site.

  1. Go to Settings › General › Advanced.
  2. Paste the full content you want into Robots.txt.
  3. Click Save changes, then open the link under the box to check the result.

Your text replaces the default completely. Start from the default above and change only what you need, and keep the Sitemap: line, or search engines lose track of your sitemap. To go back to the default, empty the box and save.

A single Disallow: / under User-agent: * blocks your whole site from every search engine.

Only well-behaved bots follow robots.txt. Scrapers and spam bots ignore it, so it isn’t a security tool. To stop spam, see Fight spam.

ads.txt

Right under Robots.txt there’s an Ads.txt box. If you show ads from Google AdSense or another ad network, they ask you to publish an ads.txt file listing who may sell ads on your site. Paste the line they give you and click Save changes. See Banners and ad slots.

Updated 7 October 2026 · Suggest an edit