AI Creative HTML · Free

AI Creative Brief Template

The single document that turns 'we need ads' into 30-50 production-ready variants per month.

For whom: Performance marketers, creative leads, and founders running 5-figure paid budgets who keep getting bottlenecked on creative.
Time to value: 2 hours to fill, weeks of clarity after
Last updated: April 30, 2026

Why this exists

AI Creative production fails when there is no shared definition of what the campaign is supposed to do. Most teams brief in Slack messages and meet about it three times. The brief here is one document that captures audience, hook angle, format mix, success metrics, and brand guardrails. Fill it once, hand to the production team, get 30 to 50 ad variants in 14 days.

The reason most creative engagements stall is bad briefs. The reason performance plateaus is variant velocity. This brief solves both. Fill it once for the campaign, then use it as the source of truth for every variant generated by an in-house team or AI creative shop.

Section 1: Campaign context

✓
Campaign goal. what specific KPI moves (CAC, ROAS, signups, MQLs)
✓
Total spend / month and per-platform allocation
✓
Funnel stage. TOF (cold), MOF (consideration), BOF (retargeting)
✓
Geo / language scope
✓
Timeline. campaign duration, decision cadence (weekly review, monthly pivot)

Section 2: ICP and persona

✓
Primary ICP. pulled from your ICP Builder Worksheet
✓
Persona pain. top 3 specific pains, in their language not yours
✓
Persona desire. top 3 outcomes they want, ranked
✓
Persona objections. top 3 reasons they say no, plus how creative can preempt

Section 3: Hook variants (the most important section)

Generate 8-15 hook variants for the campaign. A hook is the first 3 seconds of a video, the headline of a static, or the subject line of an email. Most teams test 1-2 hooks per quarter. Xpand teams test 8-15 per month.

Hook angle	Example	When it works
Pain	Stop losing 30% of leads to broken tracking	ICPs that are problem-aware
Outcome	Cut CPL in half in 21 days	ICPs that are solution-aware
Curiosity	The one Meta CAPI mistake costing you 30K/month	Cold ICPs, problem-unaware
Social proof	How 14 SaaS companies fixed attribution in Q1	Mid-funnel, comparison-mode
Contrarian	Stop A/B testing creative. Test ICPs.	Industry experts, BOF retargeting
Trigger event	Just raised? Read this before deploying paid budget	Funded founders, ABM
Mechanism	The 9-point audit that explains your CPL	Methodology-driven buyers
Numbered list	5 attribution leaks to fix this week	Blog-style retargeting

Section 4: Format mix

✓
UGC video. 15-30s vertical, talking head, real customer or actor
✓
Static image. square + vertical, hook copy + visual
✓
Carousel. 3-5 slides for B2B, 5-7 slides for ecommerce
✓
Animated text. kinetic typography, no actor, fast variant velocity
✓
Voice-over screen-record. product demo with hook narration
✓
Founder talking head. high trust, lower variant velocity

Section 5: Production cadence

Cadence	Variants / month	Production model
Light	8-12	1 hero asset, 8 cuts / hooks variations
Standard (Xpand baseline)	30-50	8-15 hooks × 3-4 visual variants
Heavy	80-150	Daily generation, AI-assisted production, weekly review

Section 6: Performance benchmarks

✓
Define winning thresholds before launch. CTR, CPL, ROAS, retention rate
✓
Define losing thresholds. when to kill a variant
✓
Decision cadence. weekly review, monthly hook refresh, quarterly pivot
✓
Iteration loop. winners get 3-5 new variants weekly, losers get killed within 7 days

Insight

Variant velocity is the single biggest predictor of creative performance. Teams shipping 30+ variants/month consistently outperform teams shipping 5-10. the volume creates statistical signal.

How to use it well

The brief is a contract, not a wishlist. Specify hook angles in the buyer's language (not yours), name 2 to 3 reference creators by handle, and define the success metric BEFORE production starts (CTR > 2%? Hook rate > 30%? Meta CPM under account average?). Reject any production work that does not tie back to the brief. Scope creep always starts there.

What good looks like

Real creative brief for a B2B SaaS demand-gen campaign on Meta + LinkedIn.

Audience: VP Demand Gen at Series B+ B2B SaaS, 50-500 FTE, US/UK/CAN
Awareness stage: solution-aware (knows about MQL→SQL leaks, evaluating fixes)

Hook angles (3):
1. Pain-led: 'Your MQL→SQL conversion is below 30%. Here is the routing fix.'
2. Contrarian: 'SDR teams are not the bottleneck. Your lead-routing logic is.'
3. Proof-led: 'How [client] cut handoff time from 4 days to 90 minutes.'

Format mix:
- 5 UGC videos (15s, 30s, 60s)
- 8 static carousels (LinkedIn-first)
- 3 spokesperson cuts (founder, CMO, head of demand gen)
- 4 animated text/data viz (Meta-first)

Reference creators: @demandcurve, @growthletter, @kieranflanagan
Success metric: Meta CPL under $185, LinkedIn CTR > 1.2%
Brand guardrails: no comparison ads with named competitors, no testimonials without written approval, 1-second logo at end of every video

FAQ

Why so many hook variants?

Algorithm fatigue. Meta and TikTok require fresh creative every 7-14 days to maintain reach on cold audiences. With 8-15 hooks rotating across formats, you generate 30+ daily variant impressions and the platform's optimization has signal to find winners. Two hooks per quarter dies in week three.

Do I need to film all this?

No. AI-assisted production (HeyGen, Synthesia, Runway, ElevenLabs) covers 60-70% of variants. Use real footage for hero assets and one founder testimonial. AI-generated variants for hook and visual rotation. The cost difference is 10-20x in favor of AI for variant production.

What's the right CPL for cold creative?

Industry-dependent. B2B SaaS cold UGC: 50-150 USD/lead. DTC product launches: 15-40 USD/lead. Compare against your blended CPL. winning creative should beat blended by 20-30%. If your best creative is at blended CPL, the offer or ICP is the constraint, not the creative.

How long should a variant run before it's killed?

7 days at meaningful spend (200 USD/day minimum on Meta, 100 USD/day on TikTok). Less than that, you don't have signal. Beyond 14 days without hitting performance thresholds, kill it. Letting losing variants run drains spend that should go to winning ones.