HG Insights

Skill: SDR Trigger Prioritization

A 200-account prospecting list ranked to 25, each row led by its strongest dated trigger.

Overview

Rank an SDR's outbound triggers the way a strong rep would. Claude works across six trigger families (buying intent, momentum, incumbent tenure, tech-stack add, function-led and SEC event), applies the recency window for each, leads with the dominant family when several fire, downgrades contradictory patterns, and compresses a 200-account list into a 25-row ranked output with the evidence, date and source for every signal.

Use cases

  • Monday morning ends with 25 ranked accounts

    Claude scores the list against the dominance order: corroborated intent on a resolved topic in the last 30 days leads, an 8-K in the last 90 days ties outreach to a public event, a recent competitor add follows. Each row names its lead trigger with the value, date and source tool, and a hook ready for the first touch.

  • Strong-looking signals that are really churn noise get downgraded

    Rising intent while the incumbent's usage falls, several competitor adds in six months, or HIGH intent on a generic topic at a large company move the account to 'research before opening' rather than the top of the list.

  • Renewal questions answered honestly

    Long incumbent tenure is labelled a weak heuristic, never a renewal date, and placeholder first-verified dates are excluded from tenure, so your reps don't promise a renewal window the data can't support.

View full skill

SDR Trigger Prioritization

Rank triggers so that a 200-account prospecting list compresses to a 25-row ranked output.

When to use

  • An SDR/BDR workflow is compressing a long prospecting list into a short ranked one.
  • A workflow must decide which trigger leads on an account when several fire.
  • An author is about to inline trigger-ranking rules into an outbound workflow. Reference this skill instead.

Tools you'll touch

  • search_companies: build the cohort. Filter on installs.products[] (ids, inclusion_method, intensity{min,max}, install_age / install_age_months, fai), intent.topics.ids with intent.signal_score, and firmographics (set company_level to avoid family duplicates). There is no momentum or trend filter. Cost: 1 per 100 companies returned.
  • list_intent_topics: resolve topic keywords to ids. Cost: free.
  • company_intent: per-account intent on resolved topics. Cost: 0.1 per call.
  • company_install_time_series: adoption trajectory per product. Always pass product_ids. Cost: 3 per product row returned.
  • company_technographic: current installs with product_first_verified_date and product_last_verified_date. Cost: 2 per company.
  • company_fai: which function operates a product, with decision-maker titles. Cost: 2 per call.
  • company_firmographic: industry and size gate, not a trigger. Cost: 0.1 per company.
  • sec_filing_section / sec_full_text_search: filing events. Tickers only. Cost: 1 per call, not cached.

The six trigger families

#FamilyTool And FieldsWindow
1Buying intentcompany_intent with fields:["topics"], granularity:"global", resolved topic_ids, signal_level:["HIGH"], signal_date.from; read score, last_seen_at, trend, buyers_journey_names≤ 30 days on last_seen_at
2Momentum / adoptioncompany_install_time_series: intensity_momentum and monthly detection days; cohort via search_companies install intensity and install_age filterslast 3 complete, loaded months
3Incumbent tenure (weak heuristic)company_technographic.product_first_verified_date on the incumbenttenure ≥ 24 months
4Tech-stack addproduct_first_verified_date inside the lookback; cohort via install_age_months{max}≤ 6 months leads; 12 to 24 months is context
5Function-ledcompany_fai department_usage_share (deduped by department) plus a recent product add owned by that functionfunction-owned add ≤ 12 to 24 months
6SEC eventsec_filing_section (companyTicker, filingType:"8-K", e.g. section:"5.02" officer changes) or sec_full_text_search (tickers, formTypes, startDate)≤ 90 days (8-K); ≤ 365 days (10-K)

Family rules on V2

  • Intent. Only resolved topics count; large companies have thousands of HIGH topics at score 100. Evaluating in buyers_journey_names is medium confidence (page, pricing or comparison activity), and most topics list both stages, so the stage rarely separates accounts. There is no default date window: pass signal_date.from. A relevant TrustRadius activity (fields:["activities"], filtered by vendor_ids or product_category_ids) upgrades the trigger. No topics at a smaller company means no coverage.
  • Momentum. intensity_momentum ≤ -20 = declining; ≥ +20 = growing; -10 to -20 = early decline (watch); otherwise stable. Adoption strength: average monthly detection days over the last 3 complete, loaded months (drop the current month and any month after product_last_verified_date). 20 or more = heavy, 5 to 19 = regular, under 5 = sparse. Search filters can only narrow the cohort by intensity and install_age; fetch the time series per survivor to read the trend. Intensity is unbounded: compare within a company, never across.
  • Incumbent tenure. Long tenure suggests a renewal cycle may be near, but no tool returns product renewal dates. Label it as a heuristic. A product_first_verified_date of 2000-01-01 means "before tracking began", not a real adoption date: don't compute tenure from it. company_contracts covers GSI/ICT outsourcing and federal awards only, not SaaS renewals.
  • Tech-stack add. Require product_last_verified_date within 90 days so the add is current. Ignore 2000-01-01 first-verified dates.
  • Function-led. FAI is point-in-time; there is no hiring or usage delta. The signal composes FAI with a recent function-owned add. It picks the contact, not the account.
  • SEC. Resolve the ticker first. Foreign filers (20-F, 6-K) go through sec_full_text_search.

company_firmographic is a gate (industry, size). An ICP-matched account with no trigger goes on the research backlog, not the ranked list.

Dominance order when multiple triggers fire

Lead the message with the dominant family:

  1. Buying intent ≤ 30 days on a resolved topic at HIGH, corroborated by a relevant TrustRadius activity or by Evaluating plus a rising trend (≥ +1). Uncorroborated HIGH intent drops to position 5.
  2. Incumbent tenure ≥ 24 months on the displacement target. Weak heuristic: lead with it only when a second family supports it.
  3. SEC event ≤ 90 days (8-K). The message ties to a public event.
  4. Tech-stack add ≤ 6 months. A recent competitor or adjacent install.
  5. Momentum decline or rise. Shifts the conversation but rarely leads.
  6. Function-led signal. Modulates which contact leads, not which account.

The order is ordinal, not additive. Two strong signals stacked beat one alone, but the opening line carries the dominant family, not a paragraph of all of them.

Recency cutoffs

Defer to hg-recent-signals for general thresholds. SDR additions:

  • The opening-line decay window is 7 to 14 days. Older signals are research context, not the lead.
  • Corroborated intent last seen ≤ 14 days is the highest-decay signal: lead today.
  • Tech-stack add ≤ 6 months is the window for "recent add"; beyond that, frame it as incumbent context.

Contradiction patterns (downgrade the account)

PatternWhy Downgrade
Rising intent (trend ≥ +1) + incumbent intensity_momentum ≤ -20Churn-driven research, not a funded buying motion
Multiple competitor adds in 6 monthsTool sprawl; the exploration cycle hasn't converged
High function-led signal + 10-K layoff disclosure within 90 daysReorg, not growth; lead with the restructuring, not expansion
SEC 8-K + intent on the categoryConfirm the 8-K is on topic; an officer change is not a security buying signal
HIGH intent at a large company on an unresolved or generic topicSaturation noise, not a trigger

These are not disqualifiers. They downgrade the lead from "open with this" to "research before opening".

Required content

Layout is up to the formatting skills (tabular-summary-discipline, executive-prose-conventions). Whatever the layout, the answer contains:

For each ranked account

  • The lead trigger family, named exactly as one of: Buying intent, Momentum, Incumbent tenure, Tech-stack add, Function-led, SEC event.
  • Its evidence: one value and the date of that value.
  • At most 2 supporting signals, each with its value and date.
  • The source tool in brackets after every value, one per signal (hg-citation-discipline). Name the tool, not the data source: TrustRadius activity comes from [company_intent].
  • A one-line hook that follows opening-line-discipline and the wording in intent-data-calibration (never "you are evaluating").

Accounts with no firing trigger are listed apart as backlog, not ranked.

For a single-account question

  • The lead trigger family with its value, date and source tool, or "no trigger" with the reason and the tools checked.
  • When the user asks about renewal: tenure is a weak heuristic for renewal timing, never a renewal date. Say so.

Common pitfalls

  1. Leading with a stale trigger. A competitor add 13 months ago belongs in the body.
  2. Treating Researching or bare HIGH as buying intent. Researching is education-stage; HIGH alone is saturated at large companies.
  3. Calling Evaluating a trial. It is medium-confidence page activity.
  4. Inventing an FAI hiring delta. FAI has no trend series.
  5. Summing weights. Three weak signals don't add up to one corroborated intent trigger.
  6. Computing tenure from 2000-01-01. It is a placeholder.
  7. Expecting a momentum filter in search_companies. It doesn't exist; read the time series per account.
  8. Passing a domain to SEC tools. They take tickers only.

Reference