Overview

Make 'recent' mean the same thing in every brief. Claude applies one set of recency thresholds per signal: installs verified within 90 days are current, intent seen within 30 days is active, contracts announced within six months are recent. It treats M&A, restructuring or a new CIO as a reset that ages older signals, so stale data never ships as news.

Use cases

Outreach that doesn't cite last year's activity
An SDR is about to open with an intent spike. Claude checks the last-seen date: a top score last seen months ago is historical, not active, and the email says so or moves on.
'Runs X' only when it's true today
An install verified 11 months ago becomes 'used X as of last spring', not 'currently uses X', and an install verified more than two years ago is never the lead line.
Strategic events reset the clock
When a filing or press report shows an acquisition or a new CIO, Claude treats older installs as historical until a newer verification confirms them, because the new leadership may already be changing the stack.

View full skill

HG Recent Signals

When to use

A workflow is filtering for "recent" or "current" signals and the threshold isn't obvious.
A prompt is about to write "recently", "currently" or "this year": check the canonical threshold first.
An author is reviewing output that conflates a 30-month-old install with a 3-month-old one.

Canonical thresholds

Signal	Field	Current Or Active	Recent	Historical	Stale
Install in use (`company_technographic`)	`product_last_verified_date`	within 90 days	3 to 12 months	12 to 24 months	over 24 months
New install	`product_first_verified_date`	n/a	first seen within 24 months (a new add)	over 24 months	n/a
Usage trend (`company_install_time_series`)	`is_active`, last loaded month	`is_active:true` (verified within 90 days)	n/a	n/a	n/a
Intent topic (`company_intent`)	topic `last_seen_at`, `summary.latest_signal_date`	within 30 days	31 to 90 days	over 90 days	n/a
Contract (`company_contracts`)	`announcement_date`, `start_date`	n/a	announced or started within 6 months	older	n/a
SEC filing	`filedAt`	n/a	within 90 days	older	n/a
Web or press	publication date	n/a	within 90 days	older	n/a

These are working defaults. A workflow may tighten them; don't loosen them without stating why.

How to read it

Installs.

Current (90 days or less) → lead with it: "runs X".
Recent (3 to 12 months) → "used X as of ", not "currently uses".
Historical (12 to 24 months) → context only, with the verification month.
Stale (over 24 months) → don't lead with it; pair with a recent corroborating source or drop it.

Placeholders. A first-verified date of 2000-01-01 means "before tracking began", not a real adoption date. Never count it as tenure or as a new add. Treat 1970-01-01 and -1 as "no value" too.

Time series. The last complete month can be partly loaded (detection stops at product_last_verified_date) and look like a collapse. Drop the current month and any month after product_last_verified_date before reading a recent trend. See hg-momentum-signals.

Intent.

signal_date has no default window: pass signal_date.from (e.g. 90 days before today) whenever the question is about recent intent, or older signals mix in.
Use each topic's last_seen_at for its recency and summary.latest_signal_date for the company's most recent signal.
Intent decays fast. A score of 100 last seen over 90 days ago is historical, not active.
Resolve topic_ids first and pass fields:["topics"] with granularity:"global"; unfiltered topic lists are saturated. See hg-intent.

Operating signals. Mentions only cover the most recent ~90 days; "no-signal" doesn't mean never.

Cache. Company tools, intent included, are cached for 30 days with no way to force a refresh. A re-call inside the window returns the same payload. Don't promise data fresher than that.

Strategic-signal classification

Some events rewrite the account picture instead of just being "recent". HG tools don't report them: source them from SEC filings (sec_full_text_search, sec_filing_section for US filers) or web_search, and cite that source. company_firmographic can corroborate an ownership change (a new global_hq_id or corporate_parent_id).

Event	Strategic?	Why
M&A: acquired	Yes	New parent, procurement integration window
M&A: divested	Yes	Standalone now; original tooling may not survive separation
Restructuring (org-wide reduction in force)	Yes	Budget freeze likely; champion churn likely
Leadership change (CIO or CTO)	Yes	A full roadmap re-evaluation is normal in the first 6 months
Funding round (over $50M)	Yes	New spend appetite
Local team expansion	No	Noise unless the team aligns with the workflow's category
Press release on a partnership	Often no	PR is not active intent; check intent and installs

A strategic signal resets the recency clock. A leadership change last month makes an install verified 14 months ago effectively stale, because the new leader may already be changing the stack. After a strategic event, treat anything older than the event as historical until a newer verification confirms it.

Common pitfalls

Calling an install verified 11 months ago "current". Current means verified within 90 days.
Reading 2000-01-01 as a real first-seen date. It means "before tracking began".
Reading intent without a signal_date.from. Old signals mix in silently.
Treating press releases as intent. A partnership announcement is marketing, not a buying signal.
Ignoring a leadership change as a recency reset. Recommendations built on pre-change installs go stale.
Sourcing strategic events from HG tools. They come from SEC or the web; cite those.

Reference

V2 tool docs: https://phoenix.hginsights.com/docs/mcp-tools/v2/company-technographic
V2 tool docs: https://phoenix.hginsights.com/docs/mcp-tools/v2/company-intent
Full docs for LLMs: https://phoenix.hginsights.com/docs/llms-full.txt
Related skills: hg-mcp-calling-rules, hg-technographic, hg-intent, hg-momentum-signals, hg-fallback-to-web-search

Skill: HG Recent Signals