Flagship Skill · Paid media strategy

The paid media strategy skill.

Run paid media that does not light money on fire.

A senior performance marketer's playbook. Hypothesis discipline for spend, channel selection, budget allocation, audience targeting, bid strategy, campaign types, what NOT to spend on, creative testing, frequency capping, attribution reality, and the failure modes that produce expensive lessons. Built for performance marketers, growth leads, and founders making early paid spend decisions.

Audience: in-house performance marketers and agencies. Adjacent: founders, growth leads, and PMs allocating spend across channels.

View the skill on GitHubBrowse paid media platforms

What this skill is for

The default state of paid media is wasted spend.

Most accounts have campaigns running because they always have, audiences targeting because the rep suggested it, bid strategies on auto because manual is hard, and creative not refreshed because there is no system. The cost compounds. A 20% efficiency gain on a $500K-per-year account is $100K back to the business. A 50% gain on a $5M-per-year account is $2.5M.

This skill is the discipline that produces those gains. It assumes you have a paid media platform connected, a working analytics layer, and conversion tracking that fires. The hard part is the strategic discipline behind the spend, and that is what is here.

The output is a defensible plan. A primary channel with a falsifiable hypothesis, a budget shape that matches the stage of the business, audience strategies treated separately, a bid strategy matched to the data state, a pre-committed scale-or-kill rule. Three answers from the framework: scale, hold, or kill.

What is in the skill

Twelve sections covered in the body.

The SKILL.md spans the full strategic surface of paid media. Each section names a decision the team will face and the discipline that produces a defensible answer.

  1. 01

    Hypothesis discipline for paid spend

    Five-part anatomy: audience, offer, channel, outcome metric, magnitude. Pre-commit the falsification rule before scale. Bad reasons ("we need to scale Google Ads") fail because they have no audience, no outcome, no magnitude. Good hypotheses survive scrutiny.

  2. 02

    Channel selection: when to use which

    Search for high-intent demand capture. Meta and TikTok for demand creation. LinkedIn for B2B precision. Reddit, Pinterest, Snap, X as scale-out channels after primary is proven. The decision rule: start where intent matches offer; do not run all of them at once.

  3. 03

    Budget allocation

    Four splits operating at the same time: brand vs performance, baseline vs test, primary vs secondary channel, daily vs lifetime. Resist the equal-channel-split temptation; one channel does the heavy lifting at 60 to 70% until two are independently proven.

  4. 04

    Audience targeting

    Three audience types treated as separate strategies. Prospecting for new demand, retargeting for engaged-but-not-converted, exclusion for current customers and never-converters. Lookalike from top-LTV, not all customers. Window splits for retargeting. Exclude employee and customer lists.

  5. 05

    Bid strategy

    Manual to maximize conversions to tCPA to tROAS as data accumulates. Each strategy fits a data state; switching too often resets the learning phase. Common mistakes: tCPA before 30 conversions, tROAS too aggressive, switching every week.

  6. 06

    Campaign types

    Per platform: Google (Search, Shopping, PMax, Display, Video, Demand Gen), Meta (Sales, Leads, Awareness, Engagement), LinkedIn (Sponsored Content, Lead Gen Forms, Message Ads), TikTok (In-Feed, Spark Ads, TopView). Match type to goal. Spark Ads outperform pure paid creative when organic exists.

  7. 07

    What NOT to spend on

    Branded keywords beyond defensive, untargeted display, geos you do not serve, hours you cannot service, devices that do not convert, audiences who never convert, creative that is tired. Audit settings drift quarterly; default settings produce most of the waste.

  8. 08

    Creative testing and frequency

    Within-campaign rotation for low-overhead testing. Across-campaign A/B for high-learning-rate testing. Refresh top creative every 30 to 60 days at scale. Set explicit frequency caps (3 to 4 awareness, 6 to 8 direct response). Rotation plus capping is the strongest pattern at scale.

  9. 09

    Attribution reality

    Platform-reported conversions are inflated by view-through, generous click windows, and self-attribution. Google Ads reports 1.3 to 1.5x what GA reports. Meta reports 1.5 to 2.5x. Two platforms claim the same conversion routinely. Single source of truth in the warehouse; platform metrics for in-flight tuning only.

  10. 10

    Common failures

    Twelve patterns: scaling but CAC up (saturation), platform-fine-revenue-bad (quality drift), 5% A/B winner (within noise), turned off underperformer and total dropped (view-through), frequency 8 (refresh creative), 5x scale jump (phase it), wrong-channel-for-offer, tROAS too aggressive.

  11. 11

    The framework: 11 considerations

    Hypothesis, channel fit, budget shape, audience strategy, bid strategy, campaign type, what not to spend on, creative testing, frequency capping, attribution reality, decision rule. Output: scale, hold, or kill at pre-committed thresholds.

  12. 12

    When in doubt, kill

    Asymmetric risk. The cost of a marginal campaign is real money daily. The cost of killing a marginal campaign is the inconvenience of restarting. Default to kill. Symmetric: when the hypothesis is clearly confirmed at the magnitude that matters, scale aggressively.

Reference files

Seven references that go alongside the SKILL.md.

The references hold the matrices, templates, and pattern catalogs the SKILL.md cites. Each is a self-contained doc the team can lift into a project without reading the rest.

  • references/channel-decision-matrix.md

    Context-to-channel matrix with worked examples across 10 common business contexts: pre-PMF B2B SaaS, high-AOV B2B, low-AOV consumer, DTC visual brand, local services, enterprise SaaS, e-commerce catalog, subscription consumer, marketplace, pre-launch waitlist. Five worked examples covering Series B SaaS, DTC soda, local services, e-commerce, and pre-PMF.

  • references/budget-allocation-templates.md

    Four split templates: growth-mode (70/20/10), steady-state (50/30/20), brand-heavy (60/40), performance-heavy (20/80). Pacing patterns for testing weeks, seasonality, launch, and steady-state. Daily vs lifetime budget guidance. Reallocation triggers.

  • references/audience-segmentation-patterns.md

    Prospecting, retargeting, and exclusion templates. Lookalike sourcing patterns (top-LTV seed). Retargeting window splits (0 to 7, 8 to 30, 31 to 90 days). Cross-platform reconciliation. Anti-patterns: too narrow, too aggressive, no exclusions, wrong lookalike source, ad-set fragmentation.

  • references/bid-strategy-reference.md

    Each strategy's fit and migration: manual, max conversions, tCPA, tROAS, max conversion value, enhanced CPC. The progression as data accumulates (manual or max conversions to tCPA to tROAS). Three signals to NOT switch strategy.

  • references/campaign-type-reference.md

    Per-platform campaign type guide. Google (Search, Shopping, PMax, Display, Video, Demand Gen, Discovery, App). Meta (Sales, Leads, Engagement, Awareness, Traffic, App). LinkedIn (Sponsored Content, Message Ads, Conversation Ads, Lead Gen Forms). TikTok (In-Feed, TopView, Spark Ads, Branded Hashtag).

  • references/ads-platform-comparison.md

    Per-platform reporting quirks and attribution differences. Google DDA defaults, Meta iOS impact and Conversions API, LinkedIn longer windows, TikTok video-completion attribution. Cross-platform interference: two platforms claiming the same conversion. Three-layer single-source-of-truth pattern.

  • references/common-failures.md

    Twelve failure patterns with name, symptom, root cause, fix, and prevention for each. Saturation, attribution mismatch, marginal A/B wins, view-through, fatigue, scaling-too-fast, wrong-channel-for-offer, aggressive tROAS, impression-share drops, PMax black box, lookalike expansion, brand search lift.

Browse all reference files on GitHub

Where to use it

Four moments where the skill earns its place.

Designing a paid media plan from scratch. Walk the 11-consideration framework, pick the channel where intent matches offer, set the budget shape, write the hypothesis with falsification rules. The output is a plan that survives a CFO review.

Auditing an existing account. Use the "what NOT to spend on" section as the audit checklist. Branded over-spend, untargeted display, wrong geos, off-hours, low-converting devices, never-convert audiences, tired creative. Most accounts find easy savings within an hour.

Deciding whether to scale or kill. Pre-commit the falsification rule before scale. Use the attribution-reality section to make sure you are evaluating against the right number. Default to kill on marginal results; the asymmetric risk favors it.

Adding a new channel. Start with the channel decision matrix. Match channel to offer. Phase the budget; do not equal-split across an unproven channel and a proven one. The 80% test rate is expensive learning if every test happens at full scale.

Pairs with these platforms

Five paid media platforms in /integrations.

The skill is platform-agnostic for the strategy. Open the integration microsite for the platform you are running to see its MCP commands, auth setup, example prompts, and the platform-specific tactical detail. The skill is the strategic layer; the microsite is the tactical layer.

Where this skill fits in the suite

The first skill in the marketing suite.

The marketing suite covers the paid media discipline across three skills. paid-media-strategy is the foundation; the other two follow in subsequent dispatches.

ads-creative-development covers the creative production layer: brief writing for ad concepts, format-specific creative direction (vertical video for TikTok and Reels, single-image for Search, carousel for Meta), and the iteration loop with the platform's algorithm. Skill landing page lands when the SKILL.md ships.

ads-performance-analytics covers the result-interpretation layer: which conversions are real, what the platform inflated, how to talk about results without losing stakeholder trust, the multi-touch attribution playbook, and the marketing-mix-modeling primer. Skill landing page lands when the SKILL.md ships.

Together the three cover the full paid media lifecycle from strategy through creative production to result interpretation. The integrations catalog at /integrations covers the platform-specific tactical layer underneath.

Open source under MIT

Read the SKILL.md on GitHub.

The skill source lives in the rampstackco/claude-skills repository alongside dozens of other skills covering the full lifecycle of brand and product work. MIT licensed.

View SKILL.mdBrowse the full catalog

Frequently asked questions.

How is this different from a vendor blog post about paid media?
A vendor blog post lists features and tells you to use the vendor. This skill is platform-agnostic and decision-shaped. Eleven considerations cover the full strategic surface: hypothesis, channel fit, budget shape, audience strategy, bid strategy, campaign type, what not to spend on, creative testing, frequency, attribution, decision rule. The output is a defensible scale, hold, or kill recommendation, not a feature pitch.
Why does the skill default to kill when in doubt?
Asymmetric risk. The cost of a marginal campaign that is not clearly working is real money flowing out the door every day. The cost of killing a marginal campaign is mostly just the inconvenience of restarting it later. Most paid media accounts have legacy campaigns that should be killed but never are. Defaulting to kill is the cheap, reversible answer; defaulting to keep running is the expensive irreversible one over time.
How does this pair with the integrations catalog?
The skill names the strategy. The /integrations pages name the platform-specific MCP commands, auth setup, and example prompts. Open the skill for the strategic decision (which channel, what budget split, which audience, what bid strategy). Open the integration page for the platform-specific tactics (GAQL queries for Google, lookalike sources for Meta, app review timeline for LinkedIn, Spark Ads for TikTok, cross-platform reporting for Synter).
What about the attribution chapter?
Most paid media decisions get made against the wrong number. Platforms inflate their own conversions via view-through, generous click windows, and self-attribution bias. Two platforms claim the same conversion routinely. The skill names the trap and points at the single-source-of-truth pattern (warehouse multi-touch attribution plus marketing-mix modeling) without taking on the deeper interpretation work, which belongs in the forthcoming ads-performance-analytics skill.
Why is the framework 11 considerations and not 5 or 20?
Eleven is the count where you cover the full strategic surface without producing a checklist that becomes ritual. Cover hypothesis, channel, budget, audience, bid, campaign type, what not to spend on, creative testing, frequency, attribution, and the decision rule. Skip any one and a real failure mode goes uncaught. Add more and the framework drifts into trivia.
What about the marketing skill suite?
Three skills planned. paid-media-strategy (this skill) covers strategy and operations. ads-creative-development covers creative production: brief writing, format-specific creative direction, the iteration loop with the platform's algorithm. ads-performance-analytics covers result interpretation: which conversions are real, what the platform inflated, how to talk about results without losing stakeholder trust. The latter two land in subsequent dispatches.