Use when a PAID MEDIA plan is needed — work backward from revenue to max CPA and required budget, split spend across channels, funnel stages, and campaign types, build a ramp-up timeline, set KPI targets per channel, and prepare a contingency plan. Trigger on 'media plan', 'paid media budget', 'how much should I spend on ads', 'channel mix for the campaign', 'plan the ad budget', 'we have 10k a month and no plan'. Also use when a budget number exists and nobody has decided where it goes. Not for — the pure revenue-to-budget math, see `10-reverse-kpi-global`; the whole marketing budget including non-paid, see `61-budget-planning-global`; the next period plan built from results, see `57-next-ads-plan-global`.
Installer avec Codex ou Claude Copiez ce prompt, collez-le dans Codex, Claude ou un autre assistant, puis laissez-le vérifier la page du skill et l'installer pour vous.
Une commande directe contourne le prompt de vérification. Examinez la source avant de l'exécuter.
Use when a PAID MEDIA plan is needed — work backward from revenue to max CPA and required budget, split spend across channels, funnel stages, and campaign types, build a ramp-up timeline, set KPI targets per channel, and prepare a contingency plan. Trigger on 'media plan', 'paid media budget', 'how much should I spend on ads', 'channel mix for the campaign', 'plan the ad budget', 'we have 10k a month and no plan'. Also use when a budget number exists and nobody has decided where it goes. Not for — the pure revenue-to-budget math, see `10-reverse-kpi-global`; the whole marketing budget including non-paid, see `61-budget-planning-global`; the next period plan built from results, see `57-next-ads-plan-global`.
metadata
{"version":"1.0.1","category":"performance"}
license
MIT
triggers
["media plan","ads plan","paid media budget","budget allocation","how much should I spend on ads","channel mix","plan a campaign budget","paid media strategy"]
output
File .md — full media plan: reverse-calculated KPIs, channel and funnel allocation, campaign-type budget split, ramp-up timeline, KPI targets per channel, and a contingency plan
A plan without hard numbers cannot tell you whether you are winning. Every figure in the plan must trace back to a revenue target. Run 10-reverse-kpi-global for max CPA and required budget before allocating anything.
Information gathering
Read .agents/product-marketing-context-global.md plus any output from 51-audience-research-global and 10-reverse-kpi-global. If information is missing, ask up to 4 questions:
Which offer, and what is the revenue target? Target revenue for the period, AOV, and gross margin percentage.
What is the budget? A hard cap, or a number derived backwards from the revenue target?
Which channels and markets? Meta / Google Search / TikTok / YouTube / Pinterest / LinkedIn — and which countries. Primary objective?
What is the timeline? Number of weeks, plus any launch date or seasonal moment that cannot move.
Principles
Work backwards from revenue, never forwards from "the budget we have". Revenue -> orders -> leads -> max CPA -> budget. If available budget is below required budget, lower the target or lower the CPA. Do not pretend.
Market tier is a planning input, not a detail. Per references/benchmarks-global.md, Tier 1 (US, Canada, Australia, Western EU) costs 6-7x Tier 2 (SEA, LATAM). A campaign profitable at 3x ROAS in a Tier 2 market may need 5-7x in the US for the same unit economics. Plan and report per market.
Season changes the math. Q1 is the cheapest window (January median CPM $15.74, about 21% below average). November peaks at $25.22, roughly 27% above average. A Q4 plan built on Q1 costs is already wrong.
Spread across the funnel. Running only bottom-funnel from day one produces high CPA because no trust exists yet.
No blank KPIs. Every channel gets its own CPA and efficiency target.
Test first, scale second. Weeks 1-2 are tuition. Do not judge ROAS in week one.
Review weekly and adjust. The plan is a hypothesis; live data decides.
Workflow
1. Reverse-calculate the KPIs (run 10-reverse-kpi-global)
Target revenue / AOV = Orders needed
Orders / close rate = Leads needed
Max CPA = (Revenue x gross margin %) / Leads needed
Break-even ROAS = 1 / gross margin %
Target ROAS = Break-even ROAS x 1.5
Required ad budget = Leads needed x target CPA
These must exist before step 2: leads needed, max CPA, target CPA, required budget, target ROAS. Build three scenarios (pessimistic / base / optimistic) and plan against base.
Sanity-check the result against references/benchmarks-global.md: global Meta medians are CPA $18-38 and ROAS 1.93-2.79. Healthy ROAS rule of thumb is 3x at roughly 30% gross margin, 5x+ for commodity margins, and 1.5x can work for high-margin digital products. If your required CPA sits far below the bottom quartile for your industry and market tier, the plan is not achievable as written.
2. Regional cost baseline
Set the expected cost floor per market before allocating budget:
Market tier
Examples
Meta CPM
Meta CPC
Tier 1 premium
United States
$15-25
$1.40-1.85
Tier 1
Canada, Australia, UK
$10-22
$1.10-1.80
Tier 1 Western EU
Germany and Western EU, Japan
$5-15
$0.90-1.50
Tier 2 SEA lead
Singapore
$6-12
$0.70-1.20
Tier 2
Thailand, Malaysia, Indonesia, Philippines
$2-7
$0.20-0.80
Tier 2 LATAM
Brazil and Latin America
$2-6
$0.30-0.70
Full table in references/benchmarks-global.md. Never blend Tier 1 and Tier 2 into one reported CPA — the average describes no real campaign.
3. Seasonality adjustment
Period
Median CPM
vs average
January
$15.74
-21%
Q1 average
$18.29
baseline
Q2-Q3
$18-21
rising
October
$22
+10%
November (peak)
$25.22
+27%
December
$22.04
+11%
Q4 average
$22.98
+26% above Q1
Planning implications: front-load acquisition into Q1 if cash flow allows; reserve at least 30% more budget for the same volume in Q4, or shift brand spend into Q3. In US election years, political spend displaces commercial inventory and adds roughly 20-40% to CPM in the last 30 days before the vote.
4. Channel allocation
Channel
% Budget
Amount
Objective
Primary KPI
Meta
[%]
[amount]
[Lead/Conversion]
CPA, CTR, ROAS
Google Search
[%]
[amount]
High intent capture
CPA, conversion rate
TikTok
[%]
[amount]
Video-first prospecting
CPA, view rate
YouTube
[%]
[amount]
Awareness, consideration
CPM, view rate
Pinterest
[%]
[amount]
Discovery for home/fashion/DIY
CPA, saves
LinkedIn
[%]
[amount]
B2B targeting
CPL, MQL rate
Total
100%
Starting archetypes, to be adjusted by real data:
DTC ecommerce: Meta 45% + Google (Search plus Shopping) 30% + TikTok 25%
Lead gen B2C services: Meta 50% + Google Search 30% + TikTok 20%
B2B SaaS: Google Search 45% + LinkedIn 30% + Meta 15% + YouTube 10%
Home, decor, fashion: Meta 45% + Pinterest 25% + Google 30%
Rule: a channel with proven performance takes 60-70%; a new channel gets a 15-20% test allocation. Do not spread evenly across channels you have never run.
LinkedIn changes the math for B2B. At $30-100+ CPM versus Meta's $13-20, LinkedIn costs roughly 2-5x more per impression. It is often still correct for B2B because the targeting precision and deal size justify it — but B2B/SaaS CPL benchmarks are $80-300, so validate against LTV in 10-reverse-kpi-global before committing more than a test budget.
5. Funnel allocation
Stage
% Budget
Audience
Creative
Primary KPI
TOFU — awareness/traffic
30-40%
Cold: broad, interest, lookalike
Strong hook, educate, entertain
CPM, CTR (Meta median 1.57-2.19%)
MOFU — lead gen
35-45%
Warm engagers 7-30 days plus filtered cold
Educate plus offer, solution proof
CPA, lead quality
BOFU — conversion/retargeting
15-25%
Hot: page visits, carts, open conversations 1-7 days
The funnel view and the campaign-type view must reconcile to the same total. Use this table as the source for cutting campaigns in 52-account-structure-global.
7. Ramp-up timeline
Express ramp in percentage of the period budget and in multiples of the learning-phase floor, not in fixed amounts — the same plan must work in a $15 CPM market and a $3 CPM market.
Week
Stage
% of period budget
Focus
Checkpoint
1-2
Testing
~30%
3-5 ad sets, 2-3 creatives each
Pause any ad set above 2x target CPA after 3 days
3-4
Scaling
~50%
Scale winners +20-30% per step, never double
CPA stable, volume rising
5+
Maintaining
~70% steady state
Refresh creative, continuous small tests
Frequency below 2.5, ROAS at or above target
Every ad set must clear the learning floor derived in 52-account-structure-global: (50 events x target CPA) / 7. Ramp only when the previous step held CPA.
8. KPI targets per channel and decision rules
Channel
CPM expectation
CTR target
CPA target
ROAS target
Meta
$13-20 global median; adjust to market tier
>= 1.6% (median band 1.57-2.19%); below 0.9% is bottom quartile
[from step 1]
[from step 1]; global median 1.93-2.79
TikTok
$5-15
Use account history — no global median in the repo benchmark file
[from step 1]
[from step 1]
Google Search
CPC $1-2 broad, $5-50+ on commercial intent
Use account history
[from step 1]
[from step 1]
YouTube
$9-30
Use view rate instead
[from step 1]
[from step 1]
Pinterest
$5-15
Use account history
[from step 1]
[from step 1]
LinkedIn
$30-100+
Use account history
CPL $80-300 typical for B2B/SaaS
Measure on pipeline, not on ROAS
Weekly decision rules, expressed relative to the target CPA so they hold in any currency:
Actual CPA
Frequency
Action
Below target
< 2.0
WIN — scale +20-30%
100-125% of target
< 2.0
Acceptable — monitor
125-150% of target
2.0-2.5
Optimize — change creative
Above 150% of target
> 2.5
PAUSE — replace or stop
9. Contingency plan
Situation
Signal
Action
CPA above 2x target
After 3 days
Pause the ad set, launch new creative
CTR far below the median band
Day 1-2
Change creative or hook immediately
Frequency above 3
Week 2+
Refresh creative, widen the audience
Underspending
Pace below 80% of plan
Widen audience, raise bid cap, check for rejected ads
CPM spikes
Seasonal peak or auction pressure
Reduce budget temporarily, shift to a cheaper market or channel
Q4 or election-period cost jump
CPM up 25%+ vs Q1
Trigger the pessimistic scenario budget from step 1
