Product Proof · Live Status

The detection pipeline

Metrivant turns monitored competitor page movement into evidence-backed signals, then synthesizes those signals into strategic movement only when the pattern is real. This page shows the live core path and the crawlable text layer behind it.

Need the buyer-facing proof hub first? Open verified competitor signals. Need the broad software category first? Review competitive intelligence software.

Status indicators reflect a simplified live view of core pipeline health and refresh every 60 seconds.
Monitored Inputs

Pricing and changelog pages are crawled every 60 minutes, newsroom and blog pages every 30 minutes, and homepage and features pages every 3 hours.

Evidence Layer

Observed page changes are anchored against stable references so every later interpretation can be traced back to a real diff.

Confidence Gates

Signals are scored, deduplicated, and held behind gates before interpretation or movement synthesis can affect the dashboard.

Supporting Systems

Validation, feed pools, coverage repair, and watchdog checks run outside the core diagram to keep output reliable.

Proof-chain standard

Verified diffs first. AI interpretation second.

Observed movement becomes a verified diff, verified diffs become confidence-gated signals, and only then does Metrivant add strategic context and one decision-ready action.

Verified diffs

Observed public movement becomes an inspectable diff before Metrivant creates a signal, summary, or recommendation.

Confidence-gated signals

Typed signals stay behind deterministic scoring, deduplication, and confidence gates before they can influence the dashboard.

AI interpretation second

Strategic context is added only after the underlying movement is verified and attributable to a real source change.

Decision-ready output

Teams get reusable evidence, one recommended action, and movement context that can be inspected instead of trusted blindly.

System Status
Operational— 8 core stages
60s refresh
OperationalDelayedDegradedChecking

This diagram shows the core detection path only. Feed ingestion, ambient activity, validation, pressure scoring, movement narratives, and self-healing operate in adjacent loops.

→ scroll
Crawlable Reference Layer

Eight stages in crawlable HTML

The visual diagram is for humans. The section below states the pipeline in plain HTML so the core product claims, stage order, and evidence model are visible to search engines and AI systems too.

Stage 01

Capture

Metrivant fetches monitored competitor pages on repeated cycles. Pricing and changelog pages are checked every 60 minutes, newsroom and blog pages every 30 minutes, and homepage and features pages every 3 hours.

Stage 02

Extract

Raw page snapshots are segmented into stable sections so the system compares meaningful product, pricing, and positioning blocks instead of unstable page chrome.

Stage 03

Baseline

Each section is anchored against a stable reference so later comparisons resolve against an inspectable origin rather than a drifting summary.

Stage 04

Diff

Current content is compared against the reference layer. Cosmetic churn is filtered out so only real movement advances.

Stage 05

Signal

Validated diffs are classified into typed signals with deterministic confidence, deduplication, and confidence gates before they can influence the dashboard.

Stage 06

Intelligence

Signals are interpreted in context only after the underlying change is verified. Strategic implication and one recommended action are added here, not before.

Stage 07

Movement

Multiple signals are synthesized across time into strategic movement only when the pattern is coherent, corroborated, and worth surfacing.

Stage 08

Dashboard

Confirmed movement, validated signals, and context assemble into the live dashboard where teams inspect the evidence trail and act.

Deterministic

Every signal traces to an observed page change. If the diff is not verifiable, it does not become live intelligence.

Evidence-grounded

The evidence chain is preserved at every stage so a team can inspect the exact movement behind the interpretation.

Continuous

The system runs continuously across page classes and supporting pools so the dashboard reflects accumulating movement, not isolated snapshots.

Proof Standards

Locked claims on the pipeline page

Every signal traces to a verified page diff - not an AI inference.
Priority pricing pages are targeted on roughly 60-minute cycles when source cadence and runtime health permit.
One concrete recommended action per signal - not a list.
Deterministic detection first. AI interpretation second.
From $15/month. 5-day free trial. No credit card required.
Pipeline FAQ

Questions buyers ask about the detection pipeline

Simple answers to the core product questions.
01

How often does the pipeline refresh competitor data?

The capture layer runs on repeated cycles. Pricing and changelog pages are checked every 60 minutes, newsroom and blog pages every 30 minutes, and homepage and feature pages every 3 hours.

02

What makes the pipeline deterministic?

Every stage starts from observed page movement, stable section anchoring, and confidence-gated transitions. If a change is not verifiable, it does not become a live signal.

03

When does a signal become strategic movement?

Only after multiple corroborating signals cohere across time. Movement synthesis is a later-stage output, not an interpretation layer applied to isolated changes.

Next step

The pipeline page should build technical confidence, then move the reader into the broader software category, the proof stack, or the workflow that matches the monitored job.