From 200,000 sources to one actionable alert.
Here's exactly how Brandpathio surfaces the signals that matter — and why velocity-based scoring with entity disambiguation produces a meaningfully lower false-positive rate than keyword volume monitoring.
Four steps from raw signal to alert.
Ingestion
Continuous crawling of 200K+ sources across five source tiers. Tier 1 (major wire services, high-velocity social) is polled every 60–90 seconds. Tier 2 (industry trade press, regional outlets) is polled every 10–30 minutes. Tier 3 (newsletters, blogs, forums) is polled hourly with burst-rate detection for unusual velocity.
Deduplication logic prevents syndicated versions of the same story from triggering separate alerts. A Reuters article picked up by 12 regional outlets registers as one story with a reach metric, not 12 separate signals. This keeps alert volume manageable.
Extraction
Each ingested item passes through an NLP extraction pipeline that performs: (1) named entity recognition to identify brands, people, organizations, and products; (2) entity disambiguation to confirm that "Apple" in a health-focused newsletter means the fruit, not the technology company; (3) sentiment scoring at both the article level and entity level — a story can be negative overall but neutral toward your brand specifically.
The extraction layer also identifies source metadata: author, publication, estimated reach, topic cluster, and whether the framing constitutes a claim (factual assertion) versus an opinion. Claims are tracked separately because they propagate differently than opinion content.
Scoring
The crisis scoring model applies three factors: (1) velocity — the rate of new mentions per hour relative to your 30-day baseline; (2) source authority weight — a mention in the Financial Times carries more weight than a mention in an obscure domain-registered two weeks ago; (3) sentiment polarity weight — negative sentiment amplifies the velocity score; positive mentions de-amplify it.
The combined velocity score triggers thresholds that you configure during onboarding. Soft alert fires at 2x your baseline velocity. Hard alert fires at 5x. Teams on the Command tier can build custom scoring logic with weighted factors and custom entity rules.
Delivery
When a threshold trigger fires, an alert is assembled and delivered within 60 seconds to configured channels. The alert includes: velocity score and trend direction; top 5 sources driving the spike; sentiment breakdown (positive / neutral / negative percentage); a brief narrative summary of what the story is about; related entities flagged in the same cluster; and an estimated time-to-mainstream based on historical propagation patterns for similar stories.
Soft alerts route to a designated monitoring channel. Hard alerts route to a crisis channel and trigger @-mentions for configured contacts. Alert routing is configurable per brand and per alert type.
The crisis velocity formula, explained.
Transparency in scoring is a design principle, not an afterthought. You should be able to see exactly why something was flagged — not just that it was.
Where:
Δmentions / hour = current hourly rate minus 30-day baseline hourly average
source_authority_index = 0.1 (low-authority domain) to 3.0 (Tier 1 wire service)
sentiment_polarity_weight = 0.5 (positive) to 2.0 (negative)
Sends to monitoring channel. Indicates a story worth watching — not yet a crisis, but showing acceleration above normal. Most soft alerts resolve without escalating. Teams can configure a "watch" state that tracks without notifying the full crisis team.
Sends to crisis channel with @-mentions for configured contacts. Requires human review and a disposition decision within 30 minutes. Hard alerts indicate a story that has left early-phase territory and is propagating into secondary coverage. Response window is typically 4–8 hours before mainstream coverage begins.
of alerts rated actionable after 30-day tuning period (median team)
Signal quality: how we manage false positives.
Every new brand profile goes through a 30-day learning period where Brandpathio observes your normal mention volume before setting alert baselines. During this period, you'll receive alerts in review mode — you see them, can act on them, but we track your feedback to tune the model.
The tuning mechanism is transparent. You'll see why each alert fired and can flag false positives directly. We use those flags to adjust entity disambiguation rules, source authority weights, and threshold sensitivity for your specific profile.
The ~91% actionable rate is a median across teams after the 30-day tuning period. Teams with simpler brand profiles — single product, narrow industry, well-established brand name with little disambiguation ambiguity — typically reach 95%+ within two weeks. Teams with more complex entity landscapes — multiple sub-brands, executives whose names appear in other contexts, industry terminology that overlaps with unrelated sectors — typically land at 85–90% by day 30. We don't hide these ranges. The tuning period is real work, and you should know what to expect going in.
See what's in your monitoring gap.
Free trial, no card required. Most teams see their first actionable alert within 6 hours of completing their brand profile setup. The 30-day tuning period runs in the background while you work.