Competition Hub · Competitor 360
Their strategy, decoded on demand.
Competitor 360 turns the sources you choose into a scored, predicted, counter-planned profile of every competitor — so you see the move while there's still time to answer it.
One continuous pipeline: the highlighted signal travels from a source you chose, through the engine and the graph, into the models and the score — and out the other side as a prediction with a prescribed counter-move. Hover any station — the watchlist, an engine stage, the leadership models, the gates — for what's happening inside.
Every source, category, model, and mechanism in this walkthrough ships in the product today — the core watchlist, the ten signal categories, the five-tier entity ladder, the H2O model targets, and the eight-dimension score are all named as they exist in the platform.
Step 1 · The watchlist
Core sources on by default. A catalog of 1,176. And yours.
Competitor 360 doesn't watch “the news” — it watches named, versioned sources with a defined cadence, and you decide the mix per competitor.
Core set — always on
The always-running spine, tuned for competitive movement — including CWCS, the competitor website classification system: we crawl each competitor's entire site, classify every page by surface (pricing, products, customer proof, careers…), snapshot it daily, and diff it — so a quiet pricing edit or an unannounced product page becomes a signal.
CWCS full-site crawl & diffs · SEC EDGAR filings · Crunchbase · USPTO patents & trademarks · public ATS job boards (Greenhouse / Lever / Ashby / Workday) · WARN Act layoff notices · GDELT global news · GitHub & open-source telemetry · Hacker News & Stack Overflow · status pages · CVE / security advisories · OpenCorporates
Add from the catalog — 1,176 sources, ten categories
Every source classified into one of the ten signal categories — browse by category and switch on what your market needs.
capital · talent · technology · commercials · marketing · sales · customers · ecosystem · observers · governance
catalog depth today: governance 346 · commercials 228 · observers 191 · technology 131 · ecosystem 126 · customers 96 · talent 33 · capital 23 · marketing 2
Configure custom — bring your own
Any URL, RSS feed, or API — with your keywords, your credentials, aimed at chosen competitors. Set cadence or cron, daily cost caps, and exactly which surfaces it feeds.
include in → competitor refresh · battlecard refresh · customer refresh
Step 2 · The signal engine
How an article becomes a signal.
Raw pages aren't intelligence. The engine reads each item the way an analyst would — then does it thousands of times a day, with provenance kept at every step.
| Stage | What happens | In the product |
|---|---|---|
| 1 · Fetch & parse | Pull the article, filing, job post — or a page from the competitor's own site via the CWCS crawl — and strip it to clean text. API-first, certified scrape fallback. | per-source query strategies · CWCS surfaces |
| 2 · Extract | Source-tuned prompts pull every distinct event out of the prose — one funding article can carry a raise, hiring plans, and an upmarket push at once. | signal extractor + per-source prompts |
| 3 · Categorize | Each extracted signal is classified into one of ten MECE categories by which actor caused it, grouped Resources / Offering / Go-to-market / Market. | capital · talent · technology · commercials · marketing · sales · customers · ecosystem · observers · governance |
| 4 · Ground | Anchored to the graph's taxonomies — category, industry, country, job — so each signal knows its market context. | taxonomy grounding |
| 5 · Resolve | An entity ladder from exact identifiers up to Tier-5 LLM ballots. The mention becomes a hard link to the company and product node — or it doesn't ship. | five-tier resolution · confidence floor |
| 6 · Score & dedupe | Every signal scored 0–10 with critical/high bands. Then the merge: near-duplicates across sources collapse into one signal with many citations. | signal scoring service |
One engine, one contract: whether the input was an SEC filing, a job post, a CWCS website diff, or your custom RSS feed, what comes out is categorized, entity-linked, scored, deduplicated signals. The engine splits and merges as the evidence demands — one article can carry several signals (a raise + hiring plans + an upmarket push), and several articles carrying the same signal collapse into one with many citations. That discipline is why everything downstream can trust it.
Steps 3–4 · Graph + models
The model bench: trajectory, intent, and the humans in charge.
Trajectory models (H2O AutoML)
Trained on the graph's own history, scored per competitor.
Leadership assessment
Executive-disposition profiling from their own words and filings.
Talent assessment
Hiring velocity and shape read as strategic commitment — a sales-heavy quarter and an ML-heavy quarter predict different futures.
Classification
Dual-output strategic phase & growth posture — is this competitor consolidating, expanding, or pivoting?
On assessing people: leadership reads are peer-reviewed constructs computed from public words and filings — surfaced as hypothesis inputs with confidence bands, never verdicts about a person. Thin evidence yields low confidence, not false precision.
Step 5 · The score
The Competitive Score framework.
Eight dimensions, each scored 1–10 from its own evidence. The overall is confidence-weighted — thin evidence counts less, strong suits lead, and lopsided profiles pay an imbalance penalty. Scores snapshot daily, so every number has a trajectory.
| Dimension | What it reads | Fed by |
|---|---|---|
| Products | Capability breadth, feature velocity, catalog strength | product graph · releases |
| Markets | Category coverage and momentum where you both sell | categories · signals |
| Financials | Fuel: funding, metric trends, filing deltas | filings · funding feeds |
| Positioning | Analyst standing & narrative strength | MQ · MarketScape · Wave |
| Distribution | Go-to-market reach and channel motion | signals · hiring shape |
| Ecosystem | Partners, integrations, platform gravity | ecosystem map |
| Operations | Execution capacity — hiring, leadership stability | talent + leadership models |
| Customers | Sentiment, engagement, review momentum | 2.4M reviews · engagement |
Two lenses, same framework. The market lens is the public baseline every tenant sees. Connect your data and your own lens diverges — your win/loss record and CRM reality re-weight what the market can't know. Until you do, the two are identical by design.
Step 6 · Fan-out
One signal, every surface that needs it.
Competitor 360
Lands in the Pulse feed and re-scores the dimensions it touches; predictions update.
Battlecards
Event-triggered refresh: high-confidence capital, technology, commercials, and governance signals rebuild the card within the day.
Competitive Matrix
Posture and evidence feed the vendor-scoring grid your buyers see.
Newsletter & digests
The week's material moves, compiled per competitor and shipped to the team.
Win/Loss & deal risk
Live signals join deal reviews and risk alerts — “they repriced mid-cycle” arrives before the loss does.
Alerts & webhooks
Rules route critical signals to Slack, email, or your systems via tenant webhooks.
Bring one hard competitive question.
We'll show you the answer in your category — scored, cited, and counter-planned — in a live session on your competitors, not a canned deck.