HG Insights|Phoenix

Skill: HG Spend

Spend numbers framed honestly — modeled estimates labeled, benchmarks contextualized, no false 'risk' calls.

Overview

Teach Claude to interpret HG's IT-spend numbers as the modeled estimates they are, and to read security-spend benchmarks the way a CRO would. The agent labels modeled values explicitly, splits internal staff time from external vendor spend when the conversation needs it, and avoids calling 'risk' on a low security number that's actually outsourced to an MSSP.

Use cases

  • Briefs that survive a 'how do you know that' challenge

    Every spend figure in the output reads as 'HG modeled $50M' rather than 'they spend $50M'. When a CRO challenges the source, the brief tells them upfront it's a model — not an invoice extract.

  • Security-spend reads grounded in industry context

    Claude knows the 8–12% benchmark for critical-infrastructure industries and the wider band elsewhere. A 2% reading is treated as a signal to investigate (outsourced? embedded?), not as a confident 'they're underspending' headline.

View full skill

HG Spend

When to use

  • A workflow needs an order-of-magnitude IT-spend number for benchmarking, sizing, or peer comparison.
  • A prompt is about to claim "their X spend is high/low" and needs a defensible comparison set.
  • An author is about to inline a security-spend benchmark — reference this skill instead.

Tools you'll touch

  • company_spend — modeled total IT spend, broken out by HG's category tree

What HG actually returns

company_spend returns modeled spend in USD across HG's category taxonomy. Important framing:

  • It is not invoice data, ERP extracts, or actual purchase orders. It's a model HG fits to firmographic + technographic + cloud-cost signals.
  • The total reflects what HG estimates the company spends operating the IT footprint, including services. It is not just license fees.
  • Categories nest. "Software" is a top-level category; "Customer Relationship Management" and "Data Warehousing" are children. Do not double-count when summarizing.

The Phoenix-side credit cost is 3 credits per call (pricing.ts).

How to read it

Internal vs. external split — HG estimates how much of the spend is "internal" (the company's own staff time + infra) vs. "external" (vendor + service-provider invoices). For benchmarking against budget, external is usually the more comparable number. For sizing the team, internal matters more.

Security-spend benchmark: for industries with critical-infrastructure exposure (financial services, healthcare, public sector), a normal security spend lands at 8-12% of total IT spend. Outside those industries the band is wider, often 4-8%.

Why low security spend is a signal, not a gap: a 2% reading can mean (a) the company outsources security to an MSSP and the spend lives in services, (b) security is embedded in cloud-platform fees rather than a separate budget line, or (c) genuine under-investment. Use the technographic stack to disambiguate before claiming "underspending".

Common pitfalls

  1. Treating modeled spend as invoiced. "They spend $50M on cloud" is a modeled estimate; "they have $50M of cloud invoices" is something HG cannot tell you. Hedge with "modeled" or "estimated".
  2. Comparing across industries without normalization. A 6% security spend at a SaaS company and at a regional bank read very differently.
  3. Summing top-level + child categories. Read the tree structure before adding numbers.

Citation rules

company_spend should always be cited as "HG modeled spend" the first time it appears in a deliverable. After the first mention, "modeled" is enough. Do not repeat "modeled" on every line — it gets noisy. See hg-citation-discipline for source-priority and density rules.

Reference