New stores: 14 days of Protector freeFirst 25 stores.Uptime, daily scans, Slack and Telegram alerts. No card.See what’s included
Store Monitoring

The metrics worth watching when nobody’s watching the dashboard.

Most metric lists are written for someone whose job is to look at them. Yours is not — you run the store. So this list is split by what a number is for: a few you should be alerted on because they mean something is broken right now, a few you should review weekly because they move slowly, and the rest that only make sense in a monthly context.

Free plan, no credit cardRead-only connectionAlerts to email, Telegram, Slack or Discord

Alert on these — they mean something is wrong now

Each of these has a definite threshold, so it can be watched by a machine and does not need interpretation. If a metric here moves, something is broken.

MetricThe signalSensible threshold
Orders per period vs baselineCheckout, payment or availability failureZero in a period that normally has orders
Order volume vs expectedPartial funnel failure or regional outageSignificantly below the weighted baseline
Store HTTP statusOutage, DNS, redirect loop, password pageAny non-success, confirmed by a second check
SSL days remainingCertificate about to expireFewer than 30 days
Domain days remainingRegistration about to lapseFewer than 60 days
Variants at $0 or inverted sale pricePricing errorAny occurrence
Tracked variants at zero stockSelling something you cannot shipAny bestseller

Review these weekly — they move slowly and cost quietly

These do not need an alert, but a fortnight of drift is expensive and a quarter of drift is a project.

  • Largest Contentful Paint on mobile — how quickly the page becomes useful on a phone.
  • Cumulative Layout Shift — whether the buy button moves under a thumb.
  • Interaction to Next Paint — whether taps feel responsive.
  • Broken internal links and missing product images.
  • Pages with missing titles, missing meta descriptions or a broken heading hierarchy.
  • Product pages missing or with incomplete Product schema.
  • Median response time, which slides before it fails.

Read these monthly — they need context, not alerts

These are business metrics rather than health metrics. They are meaningless without a comparison period and they should never page you.

  • Conversion rate, segmented by device — mobile and desktop separately or the number hides a regression.
  • Average order value and its trend.
  • Revenue per session, which combines the two above and is harder to fool.
  • Refund and chargeback rate.
  • Repeat purchase rate and customer lifetime value.
  • Organic sessions and indexed page count.
  • Ad spend against revenue per channel.

One number for “how are we doing overall”

The Cassian Score™ is a 0–100 composite of technical health (30%), content quality (25%), translation accuracy (20%), platform consistency (15%) and LLM readiness (10%), with the translation layer dropped and the weights renormalised for monolingual stores.

It is deliberately subordinate to the alerts. A score is a summary of how healthy the store is in general; an alert is what is wrong right now, and right now always wins. Use the score to see whether the last quarter improved anything, not to decide what to do this morning.

What Cassian™ actually does

Specific checks, with the plan they run on.

Order counts are ingested hourly and compared against a weighted historical baseline, with zero-order periods raised as critical.

Every plan, including free.

PageSpeed Insights runs daily, recording Performance, Accessibility, Best Practices and SEO scores plus LCP, CLS and INP, separately for mobile and desktop, with 30-day trends.

Every plan, including free.

Response time is recorded on every uptime check, so a store getting slower is visible as a trend before it becomes an outage.

Protector and above.

The Cassian Score™ is a weighted composite across technical health, content, translation, platform consistency and LLM readiness, and the exact weights and severity deductions used for any historical score are stored alongside it.

Every plan.

Missing titles, missing meta descriptions, broken heading hierarchy and missing or incomplete product schema are detected deterministically on every scan, at roughly 2ms per page.

Every plan.

FAQ

Common questions

What are the essential metrics to track for a Shopify store’s success?
Split them by purpose. Alert on the ones with a definite threshold: orders per period against a historical baseline, store HTTP status, SSL and domain days remaining, variants priced at zero or with an inverted sale price, and bestsellers at zero stock. Review weekly: mobile LCP, CLS and INP, broken links and missing images, and pages missing titles, meta descriptions or product schema. Read monthly with context: conversion rate by device, average order value, revenue per session, refund rate, repeat purchase rate and organic sessions. Business metrics should never page you; health metrics should.
Which single metric matters most?
Orders in the current period against what that period historically produces. It sits downstream of every other failure — availability, checkout, payments, pricing, stock — so it moves whatever the cause. It is also the only metric where zero is unambiguous.
How is the Cassian Score™ calculated?
It is a weighted composite: technical health 30%, content quality 25%, translation accuracy 20%, platform consistency 15% and LLM readiness 10%. Each category starts at 100 and issues deduct by severity. Stores that sell in a single language have the translation layer excluded and the remaining weights renormalised, so a monolingual store is never penalised for translations it does not have. 80 and above is green, 60 to 79 amber, below 60 red.

Imagine if your store could tell you when something's wrong.

Free scan, results in minutes. Zero-order alerts on every plan, including the free one.