KB 03 — B2B Budget Allocation & Bidding

Serves decisions: D5 (where the budget goes), D6 (when a channel decision is safe). Dated: 2026-07-17. Cites 01-CANON.md. B2B-weighted (industrial/SaaS, long cycle, high ACV).


1. The first principle: measure paid on pipeline, not platform conversions

For B2B, platform-reported conversions (form-fills) are a vanity denominator. The correct denominators are cost-per-Opportunity, cost-per-SQL, CAC, and pipeline/revenue influenced. The single highest-leverage upgrade is importing offline conversions (MQL→SQL→Opp→Closed-Won) back to the platforms so bidding optimizes to revenue, not cheap leads. GROAS bidding 2026; NAV43 PMax/B2B. Until MSAI's tracking is fixed (KB 01/02), no allocation decision is safe — you'd be optimizing on broken data.

2. The core B2B budgeting error (avoid it)

Spending ~90% on demand capture (bottom-funnel, competitor/brand search) and ~10% on demand creation (awareness) works until the in-market pool is exhausted — then ROI falls off a cliff. Balance Brand (creation) / Demand (capture) / Expand (customers). Refine Labs.

3. A concrete allocation frame (growth-stage B2B)

Exit Five 2025-04-07:

4. Channel economics (current benchmarks, cite)

5. CAC by channel (the reality check)

First Page Sage, updated 2025-06-18: B2B LinkedIn Ads CAC $982, PPC/SEM $802, Thought-Leadership SEO $647; paid avg $1,907 vs organic $942. Use to sanity-check any channel's cost-per-customer once measured.

6. Bidding: match the strategy to conversion volume

GROAS 2026:

7. Unit-economics vocabulary (the decision language)

Fiscallion 2025; GA Connector:

8. The decision gate (D6) — when is a channel decision safe?

Encodes Alecia's Phase-0 freeze as a reusable rule. Do not kill or scale a channel until:

  1. Tracking validated by a live test conversion landing in the CRM with correct source (KB 01/02 Phase 3).
  2. A volume/time floor met: ≥N tracked conversions, or ≥$5K spend over ≥60 days.
  3. Cost-per-Opportunity computed on pipeline, not platform-reported conversions.
  4. The number placed against a cited, dated benchmark with ACV/cycle context applied. Any decision missing these is "unsafe on broken data" — flag, don't recommend.

9. Creative note

Native > polished; pain-first hooks, value in the first line, social proof, clear differentiation. LinkedIn TLAs in first-person voice, 1,000-1,500 chars. Reddit especially wants organic-looking, non-salesy creative. BOL Agency; ZEN ABM TLAs.