agentleFS
slilbudgetslilbudget/knowledge-delta-skills/skills/ads-core/senior-buyer-ops/SKILL.md

senior-buyer-ops

Skill3 starsChanged 49 days agoMITView on GitHub

Portfolio layer for 'what to kill, watch, or scale across accounts': day-1 operating contract, test/scale/reserve allocation, kill/watch/scale rules with marginal scaling, creative pipeline and funnel QA. Delegates buy/survive/count/measure mechanics to platform skills; owns cross-account decisions, not single-ad-set diagnosis.

1 other project uses a version of this.

Adopted by

Most recently taken up by

The file

We split this skill into the rules it gives an agent

2 rules. About 0 are common, 0 are distinctive, and 2 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈0

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 2

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    Before you scale or kill on a difference, confirm it's real, not noise (measurement-experimentation-ops): the kill/scale ladder and …

    line 34
  2. Very specific:

    When the platform is Google, the contract above gets three more items (conversion-action hygiene, offline-import availability, per-geo …

    line 65

≈ “Common” is an estimate: at least 1 in 5 of the 364,980 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

Senior Buyer Ops

The layer above the four adapters (buy / survive / count / measure): they tell you HOW; this tells you WHAT to decide as the person accountable for a portfolio and a team's numbers. Call the adapters in order; this file owns the decisions between them. Status: unmeasured — no baseline task has been run for this skill; treat sections as unvalidated until one is.

Route by platform, then by layer:

LayerMetaGoogleTikTok
Buy mechanicsmeta-adsgoogle-adstiktok-ads
Infra / launch / executionmeta-grey-opsgoogle-grey-opstiktok-ops
Retail data layer—google-feed-opstiktok-ads/10

Scope: Meta (FB/IG) + Google Ads + TikTok Ads — no Microsoft Ads or other networks.

Platform-agnostic: trackers/metrics → tracker-ops · is-a-result-real → measurement-experimentation-ops · portfolio & cadence → references/01-portfolio-and-cadence.md · creative production → references/02-creative-ops.md · funnel QA → references/03-funnel-ops.md · automated kill/scale rules that survive small samples → references/04-automated-rules.md · spy intel via Tyver API (CLI scripts/tyver.py, join with team stats) → references/07-tyver-spy-api.md · which automation pipe (API/MCP/CSV/Sheets) for agent vs human → references/05-automation-channels.md · platform is Google → also read references/06-google-lane.md (contract additions, structure-doctrine rationale, channel detail).

Do not port structure doctrine between platforms — Google splits by intent and unit economics, not by campaign count; merging price tiers/funnel stages the way Meta rewards fails there (mechanics: references/06-google-lane.md). TikTok differs again: delivery is creative-led, the documented learning signal is ~25 results (not 50), objects are created enabled by default, and there are only two bid strategies. The widely repeated "3-5 ad groups x 3-5 creatives, ABO then CBO" TikTok doctrine is an untested Meta port (tiktok-ads/11).

Before you scale or kill on a difference, confirm it's real, not noise (measurement-experimentation-ops): the kill/scale ladder and marginal-scaling math below assume a MATURED, non-contaminated result.

Operating contract (pin on day 1, in writing — before any spend)

The single most expensive mistake is optimising against the wrong definition. Get these from the TL explicitly, don't infer:

  1. Payout event — WHICH tracker status/measure pays (lead? reg? FTD? qualified FTD? confirmed COD?). Everything downstream is priced on this. CPA pays per qualified FTD; redeposits are unpaid on CPA (quality KPI only), paid only on RevShare/Hybrid (FIELD 2026-09-27).
  2. Downstream stages + conversion lag — how long until the payout event matures (call-center confirm / deposit / KYC). Sets your judging window.
  3. Caps — daily/total offer caps; traffic past a cap is unpaid.
  4. KPI + targets — target CPA/CPL, and the QUALITY metric the advertiser judges later (reg→FTD, confirm %, scrub rate).
  5. Timezone and currency of every ad account — the tz used for every daily number and the sheet. Google serving-account tz/currency are permanent (google-grey-ops/07). Meta change closes the account and opens a new act ID (meta-grey-ops/08). A mismatch is overhead for the life of the seat.
  6. Naming + mapping — the campaign-name convention and the Meta-macro→tracker- param mapping that makes it split (tracker-ops mapping contract).
  7. Reporting source of truth — which sheet/dashboard is authoritative and by when each day.
  8. Responsibility zones + escalation — who owns accounts/creatives/tracker/ payments; who to ping on a ban wave, a tracking break, a billing hold.

Missing any of 1-8 = you are flying blind; resolve before launch.

When the platform is Google, the contract above gets three more items (conversion-action hygiene, offline-import availability, per-geo certification) — references/06-google-lane.md.

Two 2026 facts that change portfolio decisions

Moved to references/01-portfolio-and-cadence.md (§ 2026 market facts) — they are the most perishable claims in this skill; re-verify there by 2026-10-01 (post-Sept-1 AI Max migration).

Cadence

  • Daily: spend→cost sync (yesterday, per FB campaign, account tz; re-push at end of day — tracker-ops/03) → per-account CPL vs target → kill/watch/ scale (01) → update watchlist → report before the deadline.
  • Weekly: portfolio review — winner migration, replacement queue, creative backlog health, buyer/stock comparison (01).

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.