Skip to main content
FHAVI

Free resource

Every script you did not write

A Shopify store with eight apps typically loads between fifteen and thirty third-party scripts, and a meaningful share of them were added by an app you installed for one feature and forgot. Two costs follow: page speed, and a second copy of your tracking firing behind your back.

Inventory

Build the list before changing anything

01

Per script, record

  1. Source domain — not the script filename. Domains are what you can actually block.
  2. What installed it: an app, a theme snippet, a pixel, or the storefront itself.
  3. Loaded on which pages — every page, product pages only, or checkout. Checkout matters most and is easiest to miss.
  4. Size and main-thread time, from a performance trace rather than a guess.
  5. Whether it sets cookies, and which ones.
  6. Whether it fires any analytics event. This is how duplicates are found.

Duplicates

How the double-counting actually happens

An app that adds its own pixel snippet, plus your own implementation of the same pixel, produces two PageViews and two AddToCarts with different event ids. Meta counts two. The store still records one order, so the discrepancy hides in the ratio rather than the absolute number.

02

Where to look

  1. Search the page source for the pixel id. Two occurrences on one page is one too many.
  2. Search for the GA4 measurement id and for a second GTM container id.
  3. Check the checkout for scripts the storefront does not load — checkout extensions often inject their own.
  4. Look for legacy pixels: an old Facebook pixel id, a TikTok pixel from a test, a Hotjar snippet nobody removed.
  5. Confirm which copy is authoritative before deleting anything. Removing the working one is worse than leaving the duplicate.

Decide

Keep, defer, or remove

03

The three-way call

  1. Remove: anything tied to an uninstalled app, a completed test, or a feature nobody uses. This is usually a third of the list.
  2. Defer: chat widgets, review popups, personalisation and anything below the fold. Load them on interaction or after first paint.
  3. Keep synchronous: your own analytics, consent management, and the checkout itself. Nothing else earns blocking the first render.
  4. Never load a script before the consent decision if it sets advertising cookies — sequencing matters more than size.
  5. After removal, re-run the speed test on mobile and the add-to-cart test. Removing a script that was the tracking is a common, quiet disaster.

Keep going

The rest of the library

Every checklist here is free and ungated. The research pages carry the reasoning behind them.

All resources · Research library

Suspect a duplicate you cannot find?

Send the storefront URL. We will list every script, name the duplicate events, and tell you which copy to keep.

Prefer email? Write to md@fhavi.com