Your Guide to Tops Down vs Bottoms Up Target Setting
Why this is in the vault
A structured annual-planning how-to (tops-down guardrails, then bottoms-up envelope-building) with named-source CFO frameworks from Samsara and Confluent, worth keeping as the reference version of "how a real finance org sets budget."
The core argument
Budgeting runs top-down then bottoms-up. The CEO/CFO set three guardrails first — revenue, productivity (people-to-revenue ratio), profitability — triangulated against public comps, before anyone below builds a plan. CJ cites Samsara CFO Dominic Phillips: Samsara starts from a "1,000 day strategy" (three-year plan), lops it to one year, sets operating-leverage/efficiency/rule-of-40 guardrails, then splits the resulting envelope across functions (high-30s% of revenue to sales/marketing, high-teens-to-20% to R&D split 70/30 current-roadmap vs. longer-horizon bets, rest to G&A) — with the split shifting toward go-to-market as a company scales past Series B/C.
Bottoms-up is where department leaders build inside their slice. FP&A arms each leader with trailing-12-month baseline costs by named vendor (not just dollar totals), then the leader proposes new heads, new tools, and flex capacity. CJ contrasts three budgeting styles: pure incremental ("I spent X, we're growing Y%, so I need Z" — assumes last year's spend was already optimal), full zero-based budgeting (he ran one: six months, 85% of the old plan survived anyway, only 5-7% savings found), and a hybrid he attributes to Confluent CFO Rohan Sivaram — leaders must name what they'll stop funding (start the ask at ~5-10%, not 50%) before they get to argue for anything new, reallocating freed dollars into the next big bet. The gate on the whole rollup: every department's priorities must ladder 1:1 to the CEO's global priorities, and headcount must be presented as an actual org chart (current, in-recruitment, budgeted-unfilled, future-need), not a number.
Issue contents
- Numeric sponsor block (masthead "proudly powered by," financial-data-platform pitch — ERP-as-trial-balance vs. a reporting-first data layer)
- Looking for Leverage — CJ's second newsletter for PE-backed finance leaders (self cross-promo, not third-party)
- Run the Numbers podcast — episode with Zuora COFO (CFO+COO) Todd McElhatton on AI pricing and the Zuora take-private
- Quote I've Been Pondering — Jay-Z lyric, recurring closer feature
- Annual Planning Bible — CJ's own lead-magnet download (docsend link), not a third-party offer
Mapping against Ray Data Co
The Confluent "name what you stop before you get to ask for new" discipline is the same mechanism as RDCO's own feedback_design_system_before_iterating rule (no "optional" flourishes surviving past two design rounds — a CUT list comes first) and the spending-authority gate in project_rdco_spending_authority (new spend gets announced and capped, not stacked on top of existing recurring spend by default). Both are structurally the same move: force an explicit subtraction before authorizing addition, so growth doesn't default to incremental gross-up. Worth reusing the specific mechanic — "start the stop-ask at 5%, not 50%" — as language next time a CUT-list conversation needs a concrete number instead of a vibe.
The guardrail-sequencing point (top-line first, then profitability, with the gap between them literally defining the spend envelope) is also a clean frame for project_home_build_and_household_cashflow's still-open "3-scenario page" ask — set the top (income ceiling) and bottom (savings-rate floor) first, then let the build/household spend envelope be the gap, rather than building up from line items.
⚠️ Sponsorship
Numeric ran the masthead "proudly powered by" sponsor block (Financial Data Platform / ERP-replacement pitch, UTM-tagged fdp-launch campaign). This is the first time Numeric has appeared as a Mostly Metrics sponsor specifically — it was previously logged as the Mostly Metrics/CFO Secrets shared rotating-pool's 11th member via a CFO Secrets issue (2026-09-22), so this is a known advertiser crossing into Mostly Metrics' own inbox, not a brand-new third party. Treat it as unknown-until-scanned per issue going forward, same posture as the rest of the pool (Brex, Intuit, Samsara, Rivian, MLB, Abacum, Koyfin, Rillet, Tabs) — one appearance isn't a standing claim.
Separately, Looking for Leverage (CJ's own second newsletter) and the Annual Planning Bible lead magnet are self-promotion, not paid third-party content — same bucket as Mostly Talent.
Not Quote-to-Cash Part 2/3. This is a distinct "how-to" issue on annual target-setting/budgeting, unrelated to the Tabs-sponsored billing/quote-to-cash series (Part 1: 2026-10-01). Tabs did not recur here (expected — this issue isn't a series installment). The open question — whether Tabs is series-standing for Quote-to-Cash specifically — remains unresolved and can only be settled by Parts 2-3 directly.
Related
- [[2026-09-15-mostlymetrics-annual-planning-five-inputs]] — prior Mostly Metrics annual-planning framework, direct topical pair
- [[2026-07-07-mostly-metrics-multi-plan-budgeting-hyperscale]] — same sender's prior budgeting-process deep dive
- [[2026-10-01-mostlymetrics-quote-to-cash-part-1]] — the series this issue is confirmed NOT part of
- [[project_rdco_spending_authority]]
- [[feedback_design_system_before_iterating]]