CheatsheetBenchmarksBudgetAnalytics

SaaS Ads Cheatsheet: Trial Funnels, CAC Maths and Creative Specs

Trial-funnel maths, observed CAC and payback bands by ACV, creative specs for founder vs product vs customer ads, and the kill/scale gates we use on Meta and LinkedIn for B2B SaaS.

Updated 2026-05-1812 min read

LanguageEnglish

SaaS ads do not sell a checkout — they sell a trial start, then an activation, then a card. This sheet splits PLG and sales-assist funnels, gives observed CAC and payback bands by ACV, and specs the creative that actually moves a trial rather than a vanity CTR.

Optimise the event you can afford to buy, not the one the dashboard offers

Purchase-equivalent optimisation on a $49/month tool with a 14-day no-CC trial will starve the ad set: there are not 50 paid conversions a week at $200/day. The workable chain is impression → qualified click → landing view → trial start → activation (the one aha event) → paid. Bid for trial start only after the landing converts at a known rate; bid for activation only after you have volume; bid for purchase only on sales-assist or CC-required trials where the event is dense enough. If you jump straight to purchase on a long-trial PLG product, Meta will find cheap junk signups or simply never exit learning. Write the funnel as a spreadsheet before you write the hook.

CAC is a payback problem, not a CPM problem

A $16 CPM on US LinkedIn looking expensive next to a $7 DTC CPM is the wrong comparison. What matters is CAC versus gross-margin dollars inside the payback window your cash actually allows. A $39/month PLG product with 12% monthly logo churn needs a very different allowable CAC than a $499/month tool with annual contracts. First-month 'ROAS' on SaaS is a vanity number: the card may not even fire inside the 7-day click window. Report trial CPL, activated-trial CPA, paid CAC, and months-to-payback on one line. If you cannot name those four, you are not ready to scale spend — you are buying clicks that look efficient in Ads Manager and insolvent in the P&L.

Three creative jobs: problem, product, proof

SaaS creative fails when every ad tries to be a 60-second product tour. Cold traffic needs a named job-to-be-done and a pattern interrupt (the messy inbox, the spreadsheet tab, the missed SLA). Warm traffic needs the product UI doing that job in under 12 seconds of true screen capture — not a marketing-site montage. Hot traffic needs proof: a customer saying the metric that moved, with a number you can stand behind, not '10x your workflow'. Founder-to-camera works when the founder is specific; it dies when it is a TED-style origin story. UI-only ads work when the cursor path is the story. Customer ads work when the customer is in the ICP, not a celebrity adjacent to it.

Trial structure is a conversion lever you can change this week

CC-required 7-day trials convert fewer starts and more paid. No-CC 14-day trials do the opposite. Sales-assist 'book a demo' is a lead-gen problem with show-rate and SQL rate, not a SaaS checkout. Pick one primary path per campaign. Mixing 'Start free' and 'Get a demo' on the same cold ad set teaches the algorithm neither. If activation is below your gate, the ad is not the first fix — the empty-state email, the first-session checklist, and time-to-value are. Creative can only deliver a qualified start; product has to finish the conversion.

How to use the tables, including creative supply

Locked: funnel maths with target rates, allowable CAC by ACV and payback, creative specs (ratio, duration, UI-safe captions), trial-structure tradeoffs, and kill/scale gates. B2B also burns through talking-head variants faster than teams can film. Generating founder-style and ICP-avatar scripts from a product URL in Klip Kanvas is a valid way to test problem hooks at volume — as long as you do not fake a named customer logo, a fabricated case-study number, or a 'customer' who never used the product.

5 locked tables: trial-funnel target rates, allowable CAC and payback by ACV, creative specs for founder/UI/customer ads, CC vs no-CC vs demo tradeoffs, and kill/scale gates for Meta plus LinkedIn paid social.

Trial funnel — observed US target rates (PLG, cold paid social)

Outbound CTR (Meta, cold): Good 0.70-0.92% · Average 0.48-0.70% · Kill under 0.35% after 4k impressions
Landing view → trial start: Good 12-18% · Average 7-12% · Kill under 5% (landing/offer problem, not hook)
Trial start → activation (aha in 24-72h): Good 35-50% · Average 22-35% · Kill under 18% — fix product onboarding before scaling ads
Activated trial → paid (no-CC, 14d): Good 12-18% · Average 8-12% · Weak 5-8% · under 5% = trial too long or value too late
Activated trial → paid (CC-required, 7d): Good 28-42% · Average 18-28% · start volume 0.45-0.65x no-CC · paid CAC often similar or better
Demo booked → shown (sales-assist): Show-rate good 55-70% · Average 40-55% · Kill under 35% — calendar/SMS confirm is the lever
Shown demo → SQL / qualified opp: Good 35-50% · Average 25-35% · under 20% = ICP mismatch in the ad
SQL → closed-won (self-serve + assist): Use your CRM baseline · paid social SQL should sit within 0.8-1.1x of inbound SQL close rate; if it is 0.5x, targeting is wrong
3-second hook (Meta/TikTok): Good 26-33% · Average 20-26% · Kill under 16% · SaaS hooks run lower than DTC; do not borrow a 35% DTC target
LinkedIn CTR (cold, thought-leader or doc): Good 0.55-0.85% · Average 0.35-0.55% · CPC $6.50-12.00 US · judge on CPL not CTR

Allowable CAC, payback and observed bands by ACV

PLG $39-79/mo (ACV ~$350-700): Observed paid CAC $210-380 · trial CPL $18-34 · target payback ≤4 months on gross margin · month-1 'ROAS' ignore · US Meta CPM $11.80-17.40 on conversion
PLG $99-149/mo (ACV ~$900-1,400): Observed paid CAC $320-560 · trial CPL $24-44 · payback ≤5 months · US Meta CPM $13.20-19.10
Mid-market $249-599/mo (ACV ~$2.5k-6k): Observed paid CAC $520-880 · demo CPL $55-110 · SQL $180-320 · payback ≤8 months · LinkedIn + Meta mix · LinkedIn CPM $28-48 US
Upper-mid $800-1,500/mo: Do not purchase-optimise on Meta · demo/SQL optimisation only · CAC $1,200-2,400 observed when ICP is tight · 2+ stakeholders, 21-45d cycle
Allowable CAC formula: Allowable CAC = monthly gross-margin $ × payback months × (1 - target logo churn over payback) · example: $45 GM × 4mo × 0.90 = $162 cap
LTV:CAC floor: Scale above 3.0x LTV:CAC on a 12-month LTV · caution 2.0-3.0x · kill under 1.5x even if CPL looks cheap
US Meta CPM (trial-start objective): $9.40-14.80 · ~0.75-0.85x the purchase-optimised CPM on the same ICP
US Meta CPM (purchase / subscribe): $16.50-24.00 · only dense enough on CC-required or high-volume PLG
Blended vs platform: Platform 'subscription' events routinely 1.3-1.8x CRM paid starts on 7d click · use CRM as source of truth for CAC

Creative specs — founder, UI, customer

Duration sweet spot: Problem hook 15-25s · UI demo 20-35s · customer proof 25-40s · skippable YouTube 15-20s to the skip, 30s total
Ratio / export: 9:16 master for Meta Reels + TikTok · 1:1 or 4:5 for LinkedIn feed · 16:9 for in-stream · UI must be re-framed, not letterboxed
UI capture rule: True product screen, cursor path as the story, 1 job in 12s · fake marketing-site pans kill activation quality even if CTR rises
Caption / safe zone: Burned-in captions · UI chrome not covered by 9:16 bottom 20% · numbers in UI enlarged; do not rely on 12px app type
Founder-to-camera: Name the job in sentence one · no origin story before 8s · one metric max, must be substantiated · 3s hook target 28%+
Customer / case: ICP title on-screen · metric with timeframe · logo only with written permission · no fabricated titles
Problem B-roll: Spreadsheet tabs, missed Slack, red SLA clock — filmed or generated · must match the actual job the product does
Text on screen: ≤12 words per card · one number per card · CTA matches landing ('Start 14-day trial' ≠ 'Get a demo')
Social proof density: Cold: 0-1 logos · Warm: 2-4 logos or one deep metric · never a wall of logos in the first 3s
AI avatar caveat: OK for problem/founder-style · never a fake named customer, fake G2 quote, or fake company

Trial structure tradeoffs (same ICP, US paid social)

No-CC, 14-day: Relative trial starts 1.00x · activation 0.85x vs 7-day · paid conversion 0.55-0.70x vs CC-required · use when time-to-value <48h is proven
No-CC, 7-day: Starts 0.80-0.90x vs 14-day · paid 1.15-1.35x vs 14-day no-CC · default for tools with a 20-minute aha
CC-required, 7-day: Starts 0.45-0.65x · paid 2.0-3.2x vs no-CC 14-day · higher quality ICP · more support tickets at day 6
Freemium (no trial clock): Cheap starts, paid 3-8% of activated free users in 60d · only if a hard paywall sits on a must-have feature
Demo-only (no self-serve): CPL $55-110 mid-market · judge on SQL and show-rate · creative must pre-qualify ('for teams of 10+')
Hybrid: trial + 'talk to sales' on warm: Cold = trial · Warm/retarget = demo for accounts that activated but did not pay · do not mix CTAs in one cold ad set
Credit-card friction copy: If CC-required, say so in the ad or you will buy refund-seeking starts · 'no charge today' must be true
Reverse trial (paid default, cancel to free): Highest paid conversion, highest chargeback/refund risk · needs legal/comms review before paid social

Kill / scale gates and budget rules

Learning volume: 50 trial-starts per ad set per week to exit learning on trial-start optimisation · if you cannot hit it, consolidate
Daily budget vs target CAC: Daily budget ≥ 1/10 of target paid CAC while on purchase; ≥ 1/5 of target trial CPL while on trial-start
Creative concurrency: $150-300/day: 3-5 actives · $500-1,000/day: 6-10 · always 1 UI + 1 problem + 1 proof in the mix
Scale increment: +15-25% per 72h on CBO once paid CAC ≤ allowable for 5 consecutive days of CRM data, not platform data
Retarget split: Cold 65-75% · site visitors 7d 10-15% · trial-not-activated 10-15% · activated-not-paid 5-10%
Frequency: Cold 1.6-2.2 / 7d · trial-not-activated 4-6 / 7d is acceptable · over 8 / 7d on cold = fatigue
LinkedIn vs Meta budget: ICP title-sensitive mid-market: 30-50% LinkedIn until Meta lookalikes of SQLs exist · PLG sub-$80: Meta/TikTok 80%+
Kill a creative: Hook under 16% or CTR under 0.35% at 4k imps · or trial-start CVR 30%+ below ad-set median at 20+ starts

Locked

Unlock the full cheatsheet

Enter your email and we'll instantly unlock the complete cheatsheet — plus send you fresh ad playbooks every month.

No spam. Unsubscribe anytime.

Ready to put this into practice?

Create your first AI UGC video ad in minutes — no filming, no actors, no editing.

Try Klip Kanvas free