How it works

How the Shopify performance audit works. Lighthouse-based.

Store Auditor isn't a heuristic or a "third-party scripts detected" warning. It's a Google Lighthouse 13 scan, run against your live Shopify storefront, with per-app attribution layered on top. Built by Defyn Digital.

The method

Three layers, in order.

  1. 01

    Discover your storefront pages

    On install, we read your products, collections, and theme configuration via read-only API calls. We assemble a representative page set: your homepage, a top product page, a top collection page, and the cart. Each is scanned on both mobile and desktop, 8 audits in total for a typical store.

    Scopes: read_themes, read_products Runtime: ~2s
  2. 02

    Run Google Lighthouse 13

    Each page is scanned via Browserless using the official Lighthouse CLI, the exact same engine that powers PageSpeed Insights and Chrome DevTools. We extract performance score, LCP, TBT, render-blocking resources, third-party impact, and the full main-thread breakdown.

    Engine: Lighthouse 13.x Strategy: mobile + desktop Runtime: 60–120s
  3. 03

    Attribute scripts to named apps

    This is the part nobody else does. We cross-reference every blocking script against a fingerprint database of Shopify apps: script host, Theme App Extension block name, global variable names, content patterns. Each script gets attributed to a specific app: Klaviyo, Judge.me, GA4, Tidio, whatever's actually on your store. Anything unmatched is labeled unattributed third-party so it's never silently mis-categorized.

    Sources: script host, TAE blocks, global vars Coverage: ~94% of installed apps
  4. 04

    Rank by severity, suggest fixes

    Each app's combined blocking time is ranked. We tag each entry critical / high / medium / low based on absolute and relative impact, then generate a specific recommendation: defer this script, move to web pixels, lazy-load below the fold, replace with the native Shopify channel, etc.

    Output: ranked, named, actionable

What we deliberately don't do

No magic. No false promises.

×

No write access

We never change your theme, products, or apps. We can't: our scopes are read-only.

×

No guaranteed sales lift

We don't claim a fix will lift conversion by X%. We show you the blocking time and severity. The lift depends on your store.

×

No automatic changes

Every recommendation is something you (or your developer) apply yourself. We never silently flip a switch on your storefront.

×

No PII collected

We do not read customer data. Aggregate sessions / AOV / conversion rate only, used for audit context, and the merchant can disable that.

Common questions

FAQ

Do you use synthetic Lighthouse or field data (CrUX)?
Synthetic Lighthouse runs, intentionally. CrUX field data only updates monthly and only covers Chrome users with sufficient sample size. Too slow and too coarse to debug a specific app. Lighthouse gives us deterministic, repeatable measurements that you can re-run after each fix.
Which Lighthouse audits drive recommendations?
Primarily third-parties-insight (Lighthouse 13's third-party impact analysis), main-thread-tasks, render-blocking-resources, unused-javascript, and bootup-time. We extract per-script entries from those audits and map them to specific apps via our fingerprint database.
How do you fingerprint apps?
Three signals, in order: (1) script host (e.g. klaviyo.com, static.judgeme.io); (2) Theme App Extension block name from the merchant's theme settings; (3) script content patterns and global variable names. When all three agree, we attribute the script to that app with high confidence.
How long does a scan take?
Typically 60–120 seconds for a full audit. We hit 8 page types (homepage, product, collection, cart × mobile, desktop) in parallel via Browserless. Larger themes can push the upper bound, but a scan should always complete in under 5 minutes.
What if my store uses a custom-built app?
Custom apps appear in the scan results as 'Custom app' or labeled by their script host. The blocking time and severity calculations are identical. Only the attribution label changes.

Run your first scan

See the methodology in action on your store.

Install free, run an audit, and decide. Pro and Plus include 14-day trials.