Trust Center
This page explains the current public trust boundary behind Metrivant: what is monitored, what AI does, how freshness should be read, and what billing and support paths are live today.
Need the product proof stack? Review verified competitor signals, methodology, and the public ledger.
What should buyers trust here?
Metrivant's public trust posture is evidence first: monitor public competitor surfaces, preserve proof, separate verified movement from interpretation, and show coverage or quiet-state limits instead of overstating certainty.
- Evidence-first product trust means every important read should have an inspectable source boundary.
- Monitoring health and coverage honesty are part of the claim; degraded or warming surfaces should not read as complete coverage.
- Quiet states mean no qualifying movement crossed the current proof boundary, not that nothing happened everywhere.
- This page does not claim SOC 2, GDPR, CCPA, public reviews, ratings, or third-party badges unless those assets become real and visible.
Last updated:
Owner: Metrivant product
Evidence and provenance
Metrivant starts from public competitor surfaces and preserves before/after evidence, timestamps, and typed signal context before any AI interpretation is layered on top.
Freshness and monitoring honesty
We distinguish between live, warming up, quiet, and degraded states. A quiet dashboard should not pretend to be broken, and a degraded feed should not pretend to be complete.
Access, isolation, and billing boundaries
Authentication runs through the app account layer, customer-facing access is scoped from there, and backend monitoring jobs use privileged server-side access to operate the pipeline. Billing runs through Stripe-hosted checkout, webhooks, and customer portal flows rather than a custom card vault.
Why this positioning holds up
Klue, Crayon, Kompyte, and similar platforms are strong at broad distribution, battlecards, and revenue-team enablement. Metrivant’s trust claim is narrower and simpler: preserve the evidence chain, explain it clearly, and deliver it in an operational cadence.
Deterministic detection
Metrivant starts from observed competitor page and feed movement, stable baselines, and typed signals. If a change is not verifiable, it does not become live intelligence.
Explainable AI synthesis
AI is applied after proof. It adds context, recommended action, and pattern synthesis on top of evidence that stays inspectable in before-and-after terms.
Operational delivery
Signals resolve into the dashboard, alerts, briefs, and workflow-ready outputs on a real cadence. Intelligence is meant to move from detection to decision in one loop.
Public-source monitoring only. Metrivant is not claiming access to private competitor data.
AI is used for synthesis after proof. It does not originate a signal without observed evidence.
Current product operations rely on Supabase for auth and primary application data, Vercel for hosting, Stripe for billing, Sentry for error monitoring, Google Analytics for product and site analytics, and email/onboarding providers where required.
Billing is routed through Stripe checkout and Stripe customer portal flows.
Support, billing, privacy, and product-truth links are available directly from the public site.
Metrivant does not currently claim SOC 2 certification on the public site.
Metrivant does not currently publish a DPA, custom data-residency promise, or enterprise compliance pack on the public site.
A newly created org may need warm-up time before signal history and briefs become meaningful.
Quiet competitors can be truly quiet; low activity is not automatically a product failure.
Coverage health can vary by monitored surface, so provenance and coverage status matter alongside the headline summary.
Operational retention is shorter for raw HTML and observability than for the derived evidence layers buyers see in-product.
AI interpretation boundary
Metrivant’s product story is evidence first, interpretation second. AI is expected to summarize and synthesize after observed movement exists. This page does not claim zero interpretation errors.
Billing and cancellation
Billing is routed through Stripe checkout and customer portal flows. Cancellation should preserve access until the active billing period ends. Billing questions can go to hello@metrivant.com.
Freshness and warm-up
A new org may need a warm-up period before weekly briefs and higher-order pattern surfaces become useful. A quiet dashboard can still be truthful if no qualifying public movement has been detected.
Data scope and isolation
Metrivant’s public-facing methodology is built around public pages and feed surfaces. Customer-facing access is tied to authenticated app accounts, while backend jobs still use privileged server-side access to operate the monitoring pipeline. It should be evaluated as a proof-first monitoring system, not as a claim to hidden internal competitor data access or customer-managed isolation controls.
Operational retention
Current operational retention windows are shorter for raw and observability artifacts than for derived evidence layers: raw snapshot HTML is retained for 7 days, extracted sections for 90 days, section diffs for 180 days, pipeline events for 90 days, and media observations for 30 days.
Service providers
The current operating stack includes Supabase, Vercel, Stripe, Sentry, Google Analytics, Resend, Klaviyo, and OpenAI. Their role is limited to the feature boundary involved: hosting, auth/data, billing, analytics, outbound lifecycle messaging, observability, or bounded post-evidence interpretation.
What an evaluator should be able to audit here
This is the compact audit pack for public evaluation. It is meant to answer four questions directly: what is validated, what citation layers are expected, where model behavior begins, and how often the monitored surfaces are expected to refresh.
Validation status
Detection is expected to stay deterministic up to the verified-change layer. Interpretation and narrative synthesis come after observed evidence exists. Public pages do not claim that every later-stage summary is human-reviewed before display.
Citation completeness
The public proof standard expects an attributable surface, before-and-after evidence, timing, classification, confidence context, and the interpretation layered on top. If one of those layers is missing, the buyer should treat the output as incomplete rather than inferred.
Model and system trace
The current boundary is system-level, not per-signal model telemetry on the public site: deterministic code detects movement, and OpenAI-backed stages may interpret or generate context after proof. Public pages do not claim per-entry model IDs today.
Verification cadence
Public trust should be read alongside published operating cadence: pricing and changelog pages are checked every 60 minutes, newsroom and blog pages every 30 minutes, and homepage and feature pages every 3 hours. Freshness labels still matter more than any generic promise.
For the stage-by-stage explanation, open methodology, the public ledger, and the pipeline.
Does Metrivant use AI to create signals on its own?
No. Signals are supposed to begin with observed public movement. AI is used after that point to interpret implications, summarize patterns, and draft workflow-ready context.
How should a buyer read a quiet dashboard?
A quiet dashboard means no qualifying verified movement has crossed the current threshold yet, not that monitoring is disabled. Coverage health and freshness cues should be read alongside the signal count.
How does billing and cancellation work?
Checkout, upgrades, and billing management run through Stripe-hosted flows. If a subscription is canceled, access should continue until the current billing period ends.
What third-party services are part of the current trust boundary?
The current operating footprint includes Supabase for authentication and primary application data, Vercel for hosting, Stripe for billing, Sentry for error monitoring, Google Analytics for analytics, Resend and Klaviyo for outbound lifecycle messaging where used, and OpenAI for bounded interpretation or generation tasks after observed evidence exists.
What trust claims is Metrivant making right now?
The current public claim is proof-first competitor monitoring with inspectable evidence and AI interpretation after proof. This page deliberately avoids unverified claims like certifications that are not yet published.