Paid ads

Paid advertising is the scale lever in direct response, when the offer has been validated through the other three quadrants of the core four. Run paid before validation and you burn money. Run paid after, and you can scale a profitable unit into a real business in months.

When paid ads are the right move

• Your offer has converted with warm and cold outreach, you have data on what works
• Your unit economics are known (CAC, LTV, payback period)
• You can afford to run the test (typically $5K–$25K before conclusive data)
• You have bandwidth to iterate creative and landing pages weekly
• Your sales/fulfillment can handle volume if it works

When paid ads are a mistake

• Pre-product-market-fit, you're paying to find out what you should have found free
• Your LTV is under ~3x your likely CAC in the channel
• You don't have conversion tracking set up
• You can't dedicate an operator to creative testing

The core metrics

• CPM, cost per thousand impressions. Platform / audience priced.
• CTR, click-through rate. Creative quality proxy.
• CPC, cost per click. Function of CPM and CTR.
• CVR, conversion rate. Landing page / offer quality.
• CAC, cost per acquisition. CPC ÷ CVR.
• ROAS, return on ad spend. Revenue ÷ spend.
• Payback period, time to recoup CAC in gross profit.

Track all of them. Optimize the one that moves others when it moves.

The platforms, by use case

Meta (Facebook + Instagram)

Best for: consumer products, info products, services with clear visual hooks. Interruption channel, prospects aren't searching; they're scrolling. Creative-driven; creative fatigue is the main failure mode.

Google Search

Best for: high-intent queries, B2B SaaS, local services. Prospect has already named the problem. Higher CPCs but much higher intent.

YouTube

Best for: info products, high-ticket services needing storytelling, brand-adjacent video. Long-form selling works. Best for creative-heavy teams.

TikTok

Best for: consumer products with visual appeal, creator-led brands. Creative cadence is brutal (weekly at minimum).

LinkedIn

Best for: B2B, high-ticket, professional services. High CPMs offset by buyer quality.

Reddit, Quora, podcast ads, niche publications

Specialized but powerful when the ICP lives there. Lower volume, higher quality.

The creative stack

One ad can't carry the channel. You need:

• 3–5 hook variants, different opening promises, different angles
• 3–5 creative formats, video, static, carousel, UGC-style
• Weekly testing cadence, new creatives every week, winners scaled, losers killed

The creative fatigue cycle is real. A winning ad on Meta usually has a 4–12 week shelf life before frequency and CTR decay force new creative.

The hook-promise-proof-close structure

Short-form paid ads compress the full stack into 15–60 seconds / 3 paragraphs:

1. Hook, pattern interrupt, first 2 seconds. A surprising claim, a specific moment, a face looking at the camera.
2. Promise, what they'll get if they engage. 5–10 seconds.
3. Proof, one specific case or statistic. 10–20 seconds.
4. Close, the CTA. Specific action. 5 seconds.

The landing page

An ad alone doesn't convert. Ad → landing page → offer. The landing page is where most paid campaigns fail:

• Message match, the ad's headline should echo the landing page's headline
• Speed, every second of load time costs conversion
• Mobile-first, most paid traffic is mobile
• Single goal, one CTA, one offer
• Proof density, testimonials, logos, case studies, specific numbers
• Form friction, only the fields you actually need

The testing discipline

Paid is the ultimate territory, everything can and should be tested:

1. Test one variable at a time: headline OR creative OR landing page, not all three
2. Give each test enough budget to reach statistical significance (generally 3,000–5,000 clicks minimum)
3. Name a clear winner; kill the loser
4. Roll the winner into the control; test a new challenger against it
5. Keep a running document of what's been tested and what won

The scaling playbook

Once you've found a winning ad + landing page + offer:

Horizontal scale

Run the same creative on new audiences, new platforms, new geographies. Same creative, new pools of people.

Vertical scale

Increase budget on the winning setup. Algorithms punish sudden budget jumps, scale by 20–30% every 2–3 days, not 10x overnight.

Creative scale

Produce 10 variations of the winning creative, different hooks, different first frames, different end cards. Rotate them to fight fatigue.

Funnel optimization

Once spend is high, small improvements in conversion compound. 5% better landing page conversion = 5% more customers at same CAC. Ruthlessly optimize.

The attribution problem (2026)

iOS 14, third-party cookie deprecation, and multi-touch customer journeys mean pure platform attribution under-reports. What to do:

• Rely on platform attribution for directional signals, not absolute truth
• Track a ground-truth metric, actual customers, actual revenue, weekly
• Use UTMs consistently for your own reporting
• Run "marketing lift tests", pause a channel for 2 weeks, see what revenue drops
• Consider MMM (marketing mix modeling) for multi-channel attribution once scale is large enough

The discipline, weekly cadence

A healthy paid program runs a consistent weekly cycle:

• Monday, review last week's numbers (spend, conversions, CAC, LTV signals)
• Tuesday, identify winning / losing creatives; prep new tests
• Wednesday–Thursday, launch new tests, kill underperformers
• Friday, report out; document what's worked

Without the cadence, paid drifts. Creatives go stale, CAC drifts up, and by the time you notice, you've burned $50K.

Related: The core four · Scientific testing · Headlines