agentleFS
scumunnascumunna/programmatic-skills/skills/dv360-campaign-architecture/SKILL.md

dv360-campaign-architecture

Skill1 starsChanged 5 months agoMITView on GitHub

Structure a Display & Video 360 account the right way. Use when the user asks how to structure a campaign, insertion order vs line item, when to split insertion orders or line items, the DV360 hierarchy (partner, advertiser, campaign, insertion order, line item), naming conventions, or how to lay out a media plan in DV360.

The file

We split this skill into the rules it gives an agent

6 rules. About 0 are common, 5 are distinctive, and 1 is very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈5

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 1

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Start from the media plan, not the UI.

    line 58
  2. Very specific:

    Do not over-split. Every extra line item fragments conversion volume, which starves automated bidding (see dv360-bid-strategy on learning …

    line 115
  3. Distinctive:

    Use a single, consistent delimiter (an underscore works well).

    line 134
  4. Distinctive:

    Avoid special characters, leading/trailing spaces, and commas, which break CSV-based SDF workflows and report filters.

    line 135
  5. Distinctive:

    Keep values from a controlled vocabulary (fixed market codes, channel codes, stage codes) so filters match.

    line 139
  6. Distinctive:

    Read this when you are laying out a real plan and want a checklist to work through.

    line 156

≈ “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

DV360 campaign architecture

Decide how to map a media plan onto the Display & Video 360 hierarchy: how many insertion orders, where to split line items, and how to name everything so reporting and automation hold up. Good structure is the single biggest lever on whether bidding, pacing, and reporting behave later. Bad structure is expensive to unwind once a campaign is live.

This skill assumes you know what a CPM, a flight, and a funnel stage are. For KPI definitions and shared math, see the programmatic-foundations skill.

When to use this skill

  • "How should I structure this campaign in DV360?"
  • "Insertion order or line item for this?" / "Should this be one IO or several?"
  • "When do I split line items?" / "Should video and display live in the same line item?"
  • "What is the DV360 hierarchy?" / "Where does Floodlight sit?"
  • "What naming convention should I use?"

Boundaries with sibling skills:

  • Bid strategy choice per line item: hand off to dv360-bid-strategy.
  • Targeting and audience construction: hand off to dv360-targeting-and-audiences.
  • Deal and PG/PMP setup: hand off to dv360-deals-and-inventory.
  • Budget pacing and in-flight optimization: hand off to dv360-pacing-and-optimization.
  • Account-level governance, partner/advertiser provisioning, taxonomy ownership: hand off to dv360-account-setup-and-taxonomy.
  • Pre-launch verification: hand off to dv360-launch-qa.

Quick reference

The hierarchy, top to bottom:

Partner            agency / trading desk; partner-level settings inherit down
  Advertiser       one real business; owns creatives, audiences, Floodlight, deals
    Campaign       optional planning container; overall objective and total budget
      Insertion Order   the budget + flight + pacing + KPI container
        Line Item       the targeting + bidding + creative-assignment unit
          Ad / Creative assigned at the line item

Floodlight (conversion tracking) lives at the advertiser level and is shared from Campaign Manager 360 across Google Marketing Platform, not inside an individual IO or line item.

What each level decides:

LevelOwnsYou change it when
Insertion orderBudget, flight dates, pacing, frequency caps, KPI/performance goalThe money, the dates, or the goal differ
Line itemTargeting, bid strategy, creative assignment, formatThe audience, bid, format, or creative set differ

Core process

  1. Start from the media plan, not the UI. List every distinct combination of objective, market, channel, buying type, budget, and flight. Each distinct budget-and-flight cell is a candidate insertion order, because the IO is the only level that holds a budget and dates.
  2. Group those cells into insertion orders using the IO split rules below. One IO carries one budget, one flight, one pacing setting, and one KPI, so anything that needs its own budget, schedule, or success metric needs its own IO.
  3. Inside each IO, break out line items using the line item split rules. The line item is the bidding and targeting unit, so any axis you want to bid, target, optimize, or report on independently becomes its own line item.
  4. Pick the line item type per line item (display, video, audio, and YouTube handled separately). Type drives available formats, bidding, and inventory.
  5. Name every object with the structured convention below before you build, so reporting, filtering, and Structured Data File (SDF) automation work from day one.
  6. Hand off bidding to dv360-bid-strategy, targeting to dv360-targeting-and-audiences, and run dv360-launch-qa before going live.

Decision rules and thresholds

Split INSERTION ORDERS by

Each of these forces a separate budget, flight, pacing, or KPI, which only an IO can hold.

  • Funnel stage. Awareness, consideration, and conversion carry different KPIs and budgets, so keep them in separate IOs. This also keeps frequency management and reporting clean.
  • Market or geo. Separate IOs per country or major region when budgets, flights, currency pacing, or reporting are managed independently.
  • Channel or format. Display, video, CTV, and audio behave differently on pacing, viewability, and benchmarks. Separate IOs keep budget control and reporting honest per channel.
  • Buying type. Open auction versus programmatic guaranteed or PMP. Guaranteed and reserved deals have fixed commitments and fixed pricing, so isolate them from biddable open auction.
  • Bid strategy. If two sets of line items need different optimization goals at the IO level, separate the IOs so one goal does not constrain the other.
  • Verification vendor. Separate IOs when a third-party measurement or brand-safety vendor differs, so the cost and reporting attach cleanly.
  • Distinct budgets or flights. Any line items that must not share a budget pool or start/end date belong in different IOs. Line items inside one IO share its budget and dates.
  • Reporting needs. If finance or the client needs a clean budget line for a slice of the buy, give that slice its own IO.

Split LINE ITEMS by

Each of these is an axis you want to target, bid, optimize, or report on independently.

  • Audience strategy. Prospecting versus retargeting, or distinct audience segments, so bids and budgets do not bleed across intent levels.
  • Creative set. A line item serves its assigned creatives, so a different message, offer, or campaign concept needs its own line item to keep reporting and rotation clean.
  • Format or size. Separate display sizes, or static versus rich media, when you want to read performance or pace them differently.
  • Device. Split mobile, desktop, CTV, and tablet when bids, creatives, or performance differ by device.
  • Frequency rule. A distinct frequency cap or exposure goal for a slice of the audience.
  • Optimization axis. Any dimension you want the bid strategy to optimize or that you want to isolate in reporting (for example, a test cell).
  • Deal versus open auction. Keep a specific deal in its own line item, separate from open auction, so deal delivery and pricing are visible and controllable.

Do not over-split. Every extra line item fragments conversion volume, which starves automated bidding (see dv360-bid-strategy on learning periods). Split only where you will act on the distinction. If two cells would always get the same bid, targeting, and creative, merge them.

Line item types

Pick the type that matches the buy: display (image, HTML5, native), video, and audio. YouTube and Demand Gen line items are a separate family with their own formats and bidding; treat them apart from standard display/video/audio line items.

Naming convention

Consistent, structured names are what make reporting, filtering, and SDF automation work, because DV360 reporting and bulk edits key off the name string. Use a delimited template and keep it identical across every object.

Recommended fields, in order: client, market, channel, objective, funnel stage, flight.

Rules:

  • Use a single, consistent delimiter (an underscore works well). Pick one and never mix.
  • Avoid special characters, leading/trailing spaces, and commas, which break CSV-based SDF workflows and report filters.
  • Encode the same fields in the same order on every IO and line item so a wildcard filter isolates exactly the slice you want.
  • Keep values from a controlled vocabulary (fixed market codes, channel codes, stage codes) so filters match. Free-text values defeat the purpose.

Example IO name: ACME_US_CTV_Awareness_UpperFunnel_2026Q3

Example line item names under it: ACME_US_CTV_Awareness_UpperFunnel_Prospecting_15s_2026Q3 ACME_US_CTV_Awareness_UpperFunnel_Retargeting_30s_2026Q3

For the controlled vocabulary, governance, and who owns the taxonomy, see dv360-account-setup-and-taxonomy.

Reference material

  • references/structure-decision-tables.md: the full IO-split and line-item-split checklists, a worked multi-market plan mapped to IOs and line items, and a naming-field cheat sheet. Read this when you are laying out a real plan and want a checklist to work through.

Templates and examples

A two-market, two-funnel-stage prospecting-plus-retargeting video plan:

Advertiser: ACME (Floodlight shared from CM360 at this level)
  Campaign: ACME_2026Q3_Brand
    IO: ACME_US_OLV_Consideration_MidFunnel_2026Q3      (budget, flight, even pacing, CPV KPI)
      LI: ..._Prospecting_InStream_2026Q3
      LI: ..._Retargeting_InStream_2026Q3
    IO: ACME_US_OLV_Conversion_LowerFunnel_2026Q3        (separate budget, tCPA KPI)
      LI: ..._Retargeting_InStream_2026Q3
    IO: ACME_CA_OLV_Consideration_MidFunnel_2026Q3       (separate market = separate IO)
      LI: ..._Prospecting_InStream_2026Q3

Why it splits this way: US and CA are separate IOs because their budgets, flights, and reporting are managed independently. Consideration and conversion are separate IOs because they carry different KPIs and budgets. Prospecting and retargeting are separate line items because they need different audiences, bids, and creative, and you want to read them apart.

Common pitfalls

  • Putting display and video in one line item. They cannot share a creative format cleanly and the blended reporting hides what is working. Split by channel/format.
  • One mega-IO for the whole plan. You lose independent budget control, pacing, and per-goal KPI reporting. Split by funnel stage and market first.
  • Splitting line items so finely that no single line item gets enough conversions for automated bidding to learn. Consolidate test cells you will not act on.
  • Free-text or special characters in names. They break SDF round-trips and report filters. Lock a delimited template and a controlled vocabulary before building.
  • Expecting Floodlight to live inside an IO. It is configured at the advertiser level and shared from Campaign Manager 360; conversions then attribute to line items.

Sources

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.