Start here
The smallest coherent product
The company research names six steps. The durable advantage is not the number of pages PostMoney can scrape. It is the investor-specific, evidence-backed chronology that connects external developments to everything the team already knows about the portfolio company.
Build and confirm the watch profile
Official domain, newsroom, blog, feeds, careers page, founder names, and public handles. PostMoney proposes it from the company domain and existing portfolio data; the investor approves uncertain identities.
Collect a few dependable sources
First-party publishing, selected page changes, broad news mentions, and a few dependable structured sources. Not everything, just what can be trusted.
Resolve, deduplicate, show your work
Resolve identity, deduplicate observations into events, and show source, confidence, and freshness on every one.
Place events in an external lane
A clearly external lane beside founder-supplied updates. Same timeline, strong boundary.
Deliver a weekly brief
A weekly "since the last update" brief with questions to ask and helpful actions to take.
Let the investor correct, tune, and promote
Correct matches, tune sources, and promote events into reports or tasks. The stream gets better because the investor uses it.
Recommended direction
Three durable artifacts
The market research is explicit that the compelling version is not "Google Alerts inside PostMoney." It is a layer that remembers the investment thesis, watches the environment through several evidence lenses, and explains meaningful change in the context of each portfolio company. It should emphasize three artifacts.
An editable market thesis and ecosystem map
The investor's own market definition and beliefs, and the incumbents, substitutes, and competitors around the company. Editable, not generated once.
A cited "what changed" briefing
Meaningful change since the last look, with observation and inference kept visibly separate and every important sentence traceable to evidence.
A point-in-time market report
Frozen as of a date, with transparent TAM assumptions and the evidence available at the time, so it can be reproduced later.
Funding, news, hiring, demand, regulatory, customer, and technical sources become lenses feeding those artifacts. That keeps the experience centered on investor understanding and decisions instead of turning PostMoney into another undifferentiated feed reader.
Common ground
Shared building blocks
Company Pulse and the Market Dossier ask different questions, but they stand on the same foundations. These are the pieces both documents keep returning to. None of them is designed yet.
Watch profile and market thesis profile
Two editable profiles per company, proposed by PostMoney and corrected by the investor: aliases, exclusions, and identifiers on the company side; lenses, thesis statements, and disconfirming conditions on the market side.
Monitoring begins with a living profile, not just a company name. User-curated URLs and identifiers outrank inferred matches.
Entity resolution with a review queue
Matching combines multiple independent clues: identifiers, domain links, official handles, founder names, product phrases, geography, and negative evidence such as a namesake domain.
Entity confidence, claim confidence, and materiality are three separate answers. Low-confidence matches are quarantined, never silently added.
Observation, event, and the four layers
The company event is the core object, with many observations beneath it. Source artifact, observation, derived measure, and interpretation stay visibly distinct in the interface.
External observations and estimates must not silently become canonical founder-reported datapoints. They contextualize and challenge; they do not overwrite.
Deduplication and syndication clustering
One press release republished many times is one event with reach, led by the original source. Syndicated copies are counted, not repeated.
"Fourteen articles" is useful as reach. It is not fourteen developments.
Provenance and as-of snapshots
The provenance fields the market doc lists (source, dates, match rationale, extraction method, units, reported versus estimated versus inferred) plus report snapshots that preserve evidence and assumptions at the time.
Sources edit headlines, amounts, people, and dates. Corrections are first-class events that propagate to alerts, summaries, and drafts.
Source health and coverage indicators
Per source: last checked, last succeeded, last produced an item, and whether authorization expired. A LinkedIn gap says "not connected" rather than appearing as an empty feed.
"No news" is meaningful only if the monitors are working. Coverage bias is shown where the system is blind.
Alert materiality and digest cadence
Investor-defined thresholds and a cadence spectrum: immediate, daily, weekly, Monday radar, or quiet. The system learns from dismissals and batches weak signals.
More collection can make the product worse. Every alert should explain why it matters.
Organization scoping over reusable evidence
Global company and market evidence can be collected once and reused. Watchlists, annotations, thesis judgments, relevance decisions, and report usage stay private to each organization.
Shared facts, private judgment. One of the ten principles, and a hard boundary.
Rights and collection policy
Source-by-source evaluation of licensing, retention, display, and training rights. Permission-aware collection that respects robots directives, site terms, rate limits, and authentication boundaries.
"Publicly visible" does not mean "licensed for product ingestion and redistribution." Citations do not cure an unlicensed ingestion model. What may be shown to customers, founders, and LPs is unresolved.
Ties into existing PostMoney
Both documents name the same connection points: updates and extracted facts, Desk, reports and report items, tasks, tracked metrics, health signals, and reporting cadence.
This is what a standalone research tool cannot do. External evidence connected to the investor's actual exposure.
Once the stream is trustworthy
Blue-sky extensions
Both documents list richer experiences that only become possible after the evidence base exists. They are listed here as direction, not commitment.
Company
- Claim follow-through. A founder update says an FDA submission is planned in Q3; the monitor watches for the public milestone and links it when it appears.
- Narrative drift. How homepage positioning, target customer, product language, and founder messaging changed over a year.
- Launch echo. Whether a launch generated independent coverage, customer discussion, integrations, reviews, and hiring, not just a press release.
- Customer proof graph. Confirmed customer and partner evidence across both parties' sites, and when a relationship is newly announced, expanded, or removed.
- People movement context. A leadership announcement connected to a team-page diff, founder post, hiring search, and later coverage, without speculation about personal motives.
- Quiet-company check. Not merely "no press" but a broader absence across updates, releases, hiring, site changes, and official sources, followed by a suggested check-in rather than a declaration of distress.
- Help moments. A launch, executive search, conference appearance, fundraise, or customer expansion turned into a timely investor action.
- Source-aware Ask AI. "What has the company itself claimed about enterprise demand, and what independent evidence supports it?"
- Portfolio communication calendar. Upcoming conferences, webinars, launches, regulatory dates, and maintenance alongside update requests and meetings.
- Founder opt-in data room. A founder connects official publishing, app-store, analytics, social, or status accounts to improve coverage and correct identity, with investor notes kept private.
- Market handoff. When a company event points outward, open a related market-research thread rather than overloading the company feed.
Market
- Thesis ledger. A timeline of thesis statements, evidence for and against, confidence, and changes of mind: "Two assumptions strengthened, one weakened, and one remains untested."
- Counterfactual competitor. "If this company did not exist, what would customers use?" Maps of workflows, spreadsheets, internal tools, and incumbent bundles, not just similar startups.
- Emerging-cluster detection. Companies, job descriptions, papers, patents, and funding announcements beginning to use similar language before a category has a name.
- Market analogs. A young category compared with earlier adoption curves, comparing mechanisms and caveats rather than overlaying attractive charts.
- Thesis-to-portfolio gap finder. Sectors or value-chain layers the fund believes in but does not own, and who is appearing there.
- Relationship-aware intelligence. "A co-investor knows this competitor's former VP Sales," respecting connected-system permissions.
- Benchmark hypotheses. Comparable observations that suggest, not assert, benchmarks for sales efficiency, pricing, hiring mix, growth, or funding cadence, with cohort assumptions exposed.
- Expert and channel-check memory. Interview notes, conference takeaways, and consultant reports reconciled with public evidence while preserving access restrictions and licensing.
Before building
Questions to validate with investors
The market research closes with ten questions. They are quoted as written.
- Which recurring decision should this serve first: board preparation, quarterly review, reserve allocation, founder support, LP reporting, or sourcing?
- How often do investors revisit market maps and TAM assumptions after investing?
- Which signals have changed a board conversation or follow-on decision in practice?
- Do investors want broad automated discovery, tightly curated watchlists, or both?
- Who owns competitor classification and correction inside the fund?
- How much source-review work is acceptable before a report is trusted?
- Which verticals justify specialized regulatory, scientific, commerce, or procurement sources?
- What alert thresholds feel material for different stages and company types?
- Which licensed datasets do target customers already buy, and may PostMoney use their customer-provided credentials or exports?
- What may be shared with founders or LPs, and what must remain internal investment-team research?
Loose ends
Things to flesh out
Items each document raises and leaves open. Not a spec, a list of what the research did not settle.
Company Pulse
- Which page regions matter for diffs: pricing, team, partners, release notes? What counts as a change a human would care about?
- When to prefer feeds and when to fall back to watching a page, and what the defaults are.
- The default scope for founder monitoring: posts that mention the company, or all public professional posts with an explicit opt-in.
- How LinkedIn and app-review evidence arrives when generic access is not available: user-supplied links, approved integrations, founder-connected accounts, or licensed data.
- How namesakes are handled: negative keywords, exclusions, and the review path for "Possible namesake match" alerts.
- Retention for deleted posts and changed pages: marked deleted, history retained, within platform policy.
Market Dossier
- Which lenses come first for which verticals: regulatory, scientific, commerce, procurement.
- Which licensed datasets customers already buy, and whether their credentials or exports may be used inside PostMoney.
- The UX for TAM scenarios: ranges, sensitivity, and how assumptions are edited and compared.
- How many alternate market maps a company can carry, and how they are kept in sync.
- The workflow for a canonical funding observation when sources conflict on amount, date, or investors.
- Alert thresholds by stage and company type.
Undecided
Open product questions
The decisions that shape everything else, and that neither document tries to make.
- Which recurring decision comes first: board prep, quarterly review, reserves, founder support, LP reporting, or sourcing?
- Broad automated discovery, curated watchlists, or both?
- Who owns competitor classification and correction inside the fund?
- How much source review is acceptable before a report is trusted?
- What may be shared with founders or LPs, and what must remain internal?
- Is the event stream company-first or market-first in v1?
Where this comes from
Sources
Every claim on this site traces to one of two research documents in the PostMoney repository:
docs/research/company-news.md: Company News and External Monitoring. The watch profile, seven monitored layers, the company event, Company Pulse, trust boundaries, blue-sky extensions, and the first product shape.docs/research/marketing-research.md: Market Intelligence Around a Portfolio Company. The thesis profile, the living dossier, product experiences, the evidence model, risks, principles, investor questions, and the recommended direction.
Both are exploratory. Availability, licensing, retention rights, quotas, and prices should be re-checked before any implementation decision.