Coverage matrix

Ten categories. One number you actually care about: latency.

Tag failures hit Slack in under 2 seconds. Anomalies surface every 5 minutes. SOC 2 evidence reconciles daily. The full list of what TagDrishti watches, what each signal looks like, and how fast you hear about it.

# real-user data on the next pageview · no PS engagement · one developer ticket
# latency_at_a_glance

What gets reported, and when.

The same coverage list a scheduled crawler would offer — but with the time-to-detection clock honestly attached to each row.

CategoryChannelReported
Tag fire / no-fire (the core)Slack · PagerDuty · Teams · Jira · Webhook · Emailunder 2s on a healthy path
Consent & privacy complianceSlack · PagerDuty · Teams · Jira · Webhook · Emailon the first violating session
Security · Magecart · PCI DSS 4.0Slack · PagerDuty · Teams · Jira · Webhook · Emailseconds — runs on every real checkout session
Performance & Core Web VitalsSlack · PagerDuty · Teams · Jira · Webhook · Emailwithin 24 hours of install
Attribution & measurement qualitySlack · PagerDuty · Teams · Jira · Webhook · Emailcontinuous (multi-channel reconciliation runs hourly)
Anomaly detection (statistical)Slack · PagerDuty · Teams · Jira · Webhook · Emailevery 5 minutes per active workspace
GTM / TMS change correlationSlack · PagerDuty · Teams · Jira · Webhook · Emailwithin minutes of the publish (vs hours / days manually)
Vendor-side outagesSlack · PagerDuty · Teams · Jira · Webhook · Emailminutes
Synthetic journeys (Pro+ tier)Slack · PagerDuty · Teams · Jira · Webhook · Emailconfigurable; default 15 min
SOC 2 / compliance evidence (audit-log completeness)Slack · PagerDuty · Teams · Jira · Webhook · Emaildaily at 02:00 UTC
# the_full_list

Every category, spelled out.

The bullets aren’t marketing flourish — each one is an actual detector that fires in the dashboard, an alert template that goes to Slack, or a row in your BigQuery export.

Tag fire / no-fire (the core)

under 2s on a healthy path

Every tag, every real user session, every fire and every failure — captured in the browser, parsed, validated, scored.

  • 80+ vendor pixels auto-detected (GA4, Meta Pixel, Google Ads, TikTok, LinkedIn, Pinterest, Snap, Reddit, Bing, Hotjar, Amplitude, Mixpanel, Criteo, Taboola, Outbrain, and more — no mapping config)
  • Tag stops firing entirely after a deploy
  • Tag fires the wrong payload (missing transaction_id, wrong event name, malformed cart array)
  • Per-browser / per-OS failures (e.g. only breaks on iOS Safari)
  • Per-region failures (e.g. only EU traffic affected)
  • Tag fires when it shouldn't (e.g. before consent banner resolves)

Security · Magecart · PCI DSS 4.0

seconds — runs on every real checkout session

Every script that loads on a payment page is inventoried, SRI-checked, and diffed in real time. If a Magecart-style script appears, you know in seconds.

  • Unknown third-party script on a payment page
  • SRI (Subresource Integrity) hash drift on critical scripts
  • CSP (Content-Security-Policy) violations from real-user reports
  • Cross-origin script injection
  • Tracking-hygiene allowlist drift (an unfamiliar domain appears in tag inventory)

Performance & Core Web Vitals

within 24 hours of install

Per-tag attribution for LCP, INP, and CLS. Know which third-party script is eating your conversion rate.

  • Per-tag LCP, INP, CLS — which specific tag slowed the page
  • Render-blocking third-party scripts
  • Async load failures
  • Long-task attribution down to the firing tag
  • P75 figures (not lab synthetic) tied to real-device, real-network sessions

Attribution & measurement quality

continuous (multi-channel reconciliation runs hourly)

The gaps between what Meta reports, what GA4 reports, and what Shopify actually banked — surfaced before the finance reconciliation does.

  • GA4 ↔ Shopify backend revenue gap
  • Meta CAPI deduplication failures (fbp / fbc cookie timing)
  • _gl linker parameter failures cross-domain
  • Referrer / UTM loss
  • Server-side vs client-side event divergence
  • Conversion API event_id mismatches

Anomaly detection (statistical)

every 5 minutes per active workspace

A z-score on a 14-day rolling baseline, smoothed by EWMA, augmented by CUSUM for sustained drift, and decomposed seasonally so a Tuesday is compared to other Tuesdays.

  • Fail rate drift
  • Tag count drift (new tag appeared, old tag disappeared)
  • Latency drift
  • Revenue-at-risk drift
  • Adaptive thresholds per workspace — no manual tuning
  • The output is explained: the alert names which signal triggered, so you can debug the alert rule when it misfires

GTM / TMS change correlation

within minutes of the publish (vs hours / days manually)

When a tag breaks, was it a code change, a vendor outage, or a GTM publish? We tie the anomaly start time to the publish timestamp.

  • A GTM publish that broke a tag — surfaced by correlating anomaly onset with the publish event
  • Trigger condition changes that silently exclude traffic (e.g. URL contains "?utm" → tag now skips deep-link landings)
  • Tealium iQ / Adobe Launch / Segment / Piwik PRO equivalents on Agency Plus
  • Pre/post diff at the trigger and variable level

Vendor-side outages

minutes

Meta / Google / TikTok endpoints fail on their side more often than the public dashboards admit. We correlate across tenants and tell you when it's not your code.

  • Meta Conversions API outages
  • Google Ads pixel endpoint slowness
  • TikTok Events API rejections
  • Cross-tenant correlation: if 30% of TagDrishti customers see Meta CAPI 5xx in the same 30-minute window, that's a vendor outage, not your bug

Synthetic journeys (Pro+ tier)

configurable; default 15 min

Multi-step user flows replayed on a schedule — useful for zero-traffic critical paths (a niche product page that rarely converts but converts large).

  • Scripted checkouts on a schedule (default every 15 min, configurable)
  • Form submission validation
  • Behind-login flows where real users haven't completed the path yet
  • Pre-launch staging audits on URLs nobody visits
  • Cross-browser (Chrome, Firefox, Safari) and cross-region (US, EU, APAC, AU)

SOC 2 / compliance evidence (audit-log completeness)

daily at 02:00 UTC

Every privileged action against the platform is logged into a canonical audit_logs table. A daily agent diffs the source-of-truth against audit_logs and fires P0 if any gap exists. Built for Sprinto / Vanta / Drata pipelines.

  • Admin logins via Clerk (action_type: admin.login)
  • API key creation / rotation / revocation (api_key.mutation)
  • Tenant impersonation sessions (tenant.impersonation)
  • Billing changes via Paddle (billing.change)
  • Every /api/superadmin/* call (superadmin.action)
  • Daily completeness check at 02:00 UTC
# where_we_don_t_pretend

Three things we deliberately don’t do.

A coverage page that admits nothing reads as marketing copy, not analysis. Three honest gaps and how to handle them.

Zero-traffic pages.

Real-user monitoring only sees what real users hit. A niche product page with no traffic generates no data for us. Pair with the synthetic-journeys module (Pro+) for critical low-volume paths, or run a one-off audit via audit.tagdrishti.com on the specific URLs.

honest gap

Behind-login flows real users haven’t completed.

If 80% of your tracking lives post-checkout and you need pre-launch validation before any real customer walks the flow, a scripted synthetic crawler with credentials is the right tool. Our synthetic-journeys module covers some of this; an ObservePoint or Wappalyzer pass covers the rest.

honest gap

Quarterly governance binders.

If compliance needs a “we ran 1,200 URLs through a tag audit on March 31” PDF for SOX or internal governance, the scheduled-crawl format is what they want. We ship continuous monitoring + a free one-URL editorial audit, not a 1,200-URL crawl report.

honest gap
first signal on the next pageview

See your coverage on the next pageview.

One async script in <head>, installed once by your developer (not via GTM, by design). Real-user data shows up on the next pageview. The ten categories above start populating in minutes. 30-day free trial, no credit card. Starter from $99/mo.

30-day free trialno credit cardreal-user data on the next pageviewStarter from $99/mo