WooCommerce Wishlist Bot Traffic: Why Bots Add to Wishlists

Written by

in

<!– target keyword: woocommerce wishlist bot traffic –>

<!– meta description: Crawlers follow ?add_to_wishlist= links and can create wishlist rows for visitors who don’t exist. Here’s how to confirm it and stop it. –>

WooCommerce Wishlist Bot Traffic: Why Bots Add to Wishlists

You check your wishlist plugin’s table and the numbers do not match your customers. There are rows for people who never bought anything, never created an account and, as far as you can tell, never visited the site. Your server log tells a similar story: the same ?add_to_wishlist=123 URL, over and over, from user agents that are not browsers.

This is not a bug in your wishlist plugin. It is what happens when a feature built for a click is exposed as a URL that anything can request.

Why a wishlist button is a link a crawler can follow

Some wishlist plugins render the button as an ordinary link rather than a form submission:

?add_to_wishlist=123&_wpnonce=...

A browser sends that request when a shopper clicks. So does a search engine, a social preview fetcher, an AI training crawler, a price-monitoring bot or a script someone wrote in an afternoon. None of them needs to understand the page. They only need to follow links, and your shop and category pages contain one on every product, on every page of pagination.

That is the multiplier. A catalogue with 2,000 products and 100 paginated category pages is not one wishlist URL a crawler can find. It is thousands of them, and the same crawler may come back next week.

On some wishlist plugins, following that URL is not a read-only event. The plugin treats the request as a shopper adding an item and writes a row. A crawler that never renders the page still creates the row, because the row is written by PHP on the server before any JavaScript runs.

How to confirm it is bots, not shoppers

Before changing anything, get evidence from your own site. Three checks are enough for most stores.

1. Look at the user agents hitting the wishlist URL. In your access log, filter for add_to_wishlist and count the user agents. Browsers send long, varied strings. Crawlers often announce themselves (AhrefsBot, GPTBot, Bytespider, curl, python-requests), and spoofers send a browser string that is identical across thousands of requests from rotating addresses.

2. Compare wishlist rows with customers and orders. If rows are being created for sessions with no account, no order and no other page views, they are not shoppers browsing. Look at the timestamps too: a burst of rows a few seconds apart, at 3 a.m., with no product page views in between, is a pattern no human produces.

3. Run in monitor mode first. A request-level shield can record the block and challenge verdicts it would have applied without blocking anything. Give it 48 hours of real traffic, then read the log. If the same crawler names keep appearing on wishlist URLs, you have your answer from your own data rather than from an article.

One caution: not every unexplained row is a bot. A logged-out shopper can legitimately add an item, and some plugins create a row for a real visitor whose cookie later expires. The question is not “are there rows without accounts” but “are these requests coming from software that is not a browser”.

Why robots.txt and a firewall rule are only half a fix

The obvious first move is a Disallow rule in robots.txt. It is free and it is worth doing, but it is a request, not an enforcement. A crawler that ignores it — and a user-agent spoofer always ignores it — still gets through. It also does nothing about requests that are already queued or already indexed.

The second obvious move is a firewall or CDN rule that blocks the wishlist query string. That can work well at the edge, and if you already run a WAF it may be the cheapest option. Two costs come with it. Address-based rules are blunt: shared addresses, mobile networks and privacy VPNs mean you can block real shoppers along with the bot. And a rule written against one query argument misses the moment a plugin changes its parameter name or a crawler starts using a slightly different URL.

Neither approach looks at the request the way your wishlist plugin does. That is the gap a request-level shield fills.

What actually stops it

The reliable approach is to decide, before your wishlist plugin runs, whether a request looks like a shopper. Four layers, cheapest first:

  • Nonce check. A request with a missing or malformed nonce is rejected before the wishlist plugin sees it. That removes naive requests which omit the token, but not anything that copies a full URL, nonce included.
  • Known-crawler block. Match the user agent against a list of crawler, scraper and automation signatures. This is a blocklist, so it is a first filter, not a guarantee.
  • A JavaScript and signed-cookie challenge. A real browser can run a tiny script and return with a signed cookie. A spoofed user agent that cannot execute JavaScript cannot pass, and neither can a crawler that does not run scripts. This is the layer that catches bots that lie about who they are but cannot run a browser.
  • Rate limiting. Cap requests per visitor fingerprint and site-wide. When the cap is hit, return HTTP 429 with Retry-After rather than doing the work.

Order matters. If the challenge runs after the wishlist plugin has already written a row, you have paid the database cost and only cleaned up afterwards. The check has to sit in front of the wishlist handler.

Do not forget the search leak

Bot traffic and search indexing are the same URL problem. If ?add_to_wishlist= URLs are crawlable, they can be indexed, and every indexed variant is another invitation. Three rules help, but they do different jobs:

  • Disallow: /*add_to_wishlist=* (and the hyphenated variant) in robots.txt reduces crawling. It does not remove a URL that is already indexed, and it can stop a crawler from seeing the noindex response, so it is not a de-indexing tool on its own.
  • X-Robots-Tag: noindex, nofollow on responses carrying the argument is the signal that keeps the URL out of the index and removes an entry that is already there, provided the crawler is still allowed to fetch it.
  • A noindex robots meta tag and a canonical link with every wishlist argument stripped consolidate variants that are still crawled: the meta removes the action URL from the index, and the canonical points at the clean product URL.

If you need to de-index an already-indexed URL, apply the noindex first and add the Disallow afterwards, once the URL has dropped out. A request that carries no wishlist argument should be untouched: no header, no meta change, no canonical rewrite. That is the line between cleaning up an action URL and accidentally noindexing your shop.

What this does not fix

Be honest with yourself about the edges before you buy or install anything.

A crawler that runs a full headless browser can execute the challenge script and pass. A signed-cookie test raises the bar; it does not make the site invisible. A full-page cache or CDN that serves the wishlist URL without invoking PHP is safe from a database-write point of view — no PHP, no wishlist row — but the request will not appear in any log the plugin keeps. And a wishlist plugin that processes its action extremely early, before other plugins load, can still win the race on some configurations.

These are reasons to test in monitor mode, not reasons to do nothing.

A sane rollout order

  1. Install the shield and leave it in monitor mode. Nothing is blocked; every block and challenge verdict is recorded, while allowed requests are not.
  2. After 48 hours, read the log. Filter by reason. Confirm the blocked requests are bots, and check your server logs or analytics to confirm your real shoppers were allowed.
  3. Switch to enforce. Watch the log for the first day, particularly for any real customer who hits the challenge twice.
  4. Add the SEO rules if they are not already on, then fetch a wishlist URL and a normal product URL and compare the headers and canonical tag.
  5. Set log retention to the shortest period you actually review. Seven days is a reasonable default; there is no reason to keep hashed visitor data longer.

If the log shows nothing after a week of real traffic, you may not have this problem. That is a useful result: leave the shield in monitor mode or remove it, and spend the effort elsewhere.

The bottom line

Wishlist bot traffic is not mysterious once you look at the request. A public URL that writes a database row will be requested by software that is not a shopper, and a wishlist button rendered as a plain link is exactly that URL. You can confirm it from your own logs, decide in monitor mode, and enforce before the wishlist plugin runs.

If you want the four layers above in one plugin, without changing your wishlist plugin or its templates, Wishlist Bot Shield is free to download and use. It includes the missing/malformed nonce guard, the crawler block, the signed-cookie challenge, rate limiting, the blocked-request log, monitor mode and the robots/noindex/canonical cleanup. It works with wishlist plugins that use a query-argument action URL; check yours before installing. No account, no key, no time limit.

Download it, run monitor mode for two days, and read your own log before you enforce.

Comments

Leave a Reply

Your email address will not be published. Required fields are marked *