Paid Advertising (Google Ads / Facebook Ads / PPC)

No single 'best' channel; depends on product, audience, and margins. Google Ads (search) = highest intent (user searching for solution), 3-5x ROAS, $1-5 CPC, best for high-margin products/services. Meta/Instagram (social) = mid intent (retargeting + lookalike), 2-4x ROAS, lower CPM in many verticals, excellent for e-commerce + DTC brands. TikTok (social) = discovery/awareness, 1-2x ROAS initially, improving, best for viral/brand-building products targeting Gen Z. Rule of thumb: search for conversion, social for brand+retargeting. Test all 3 with $500/month each before committing.

iOS 14.5+ users (~85%) opt out of tracking; pixel data is incomplete. Solutions: (1) Implement Conversions API / server-side tracking (Meta, Google) — sends conversions server-to-server, bypassing iOS restrictions; (2) Use first-party data (email list + web pixel) for remarketing; (3) Run incrementality tests (hold-out groups) to estimate true ROAS independent of attribution; (4) Adopt cohort analysis (track LTV by weekly cohort over 12 weeks, not daily ROAS); (5) Use MMM (Marketing Mix Modeling) for budget planning. Modern best practice: combine cohort LTV analysis + server-side Conversions API + monthly incrementality testing. Accept 20-30% data loss visibility; everyone is blind, competition is too.

Test sequentially: (1) Core message (discount vs urgency vs benefit), (2) Creative format (video vs image vs carousel), (3) Audience (lookalike size, interest overlap), (4) Placement (feed, stories, search network). Never test more than 2 variables at once. Run for at least 500-1000 conversions per variation (minimum 7-14 days for high-volume campaigns, 2-4 weeks for low-volume). Statistical significance: use Bayesian rules (not frequentist p-values) — stop when 95% confidence one variant is better. Triple Whale's built-in test advisor helps; use it. Document every test (winner, learnings) in a shared doc — compound knowledge across campaigns.

Start with 60/30/10: 60% to proven winners (usually Google), 30% to mid-performing (Meta/LinkedIn), 10% to experimental (TikTok/YouTube). Measure by LTV:CAC ratio, not ROAS alone (ROAS = short-term, LTV captures long-term value). Every 2 weeks, shift 5-10% from lowest LTV:CAC to highest. Scale winners by 20-30% if healthy unit economics (CAC payback <3 months, LTV:CAC >3:1). Use a spreadsheet to track by channel: [Date, Channel, Spend, Conversions, CAC, Est. LTV, LTV:CAC, Action]. Most teams underfund the channel doing best because they chase the latest shiny thing; resist that.

In-house: hire at L2+ seniority (3+ years), slow ramp (3-6 months productive), $120k+ salary, but develops institutional knowledge and can iterate daily. Agency: $2-5k/month, fast execution, lacks context, hard to iterate on creatives, high turnover. Best practice: hire in-house L2 strategist to own strategy/budgets/testing, use agency for one-off creative production or paid media buying (Superscript, P&L, Tinuiti). For startups <$2M ARR with <$20k/month ad spend, agencies are faster. For >$50k/month spend, hire in-house. Hybrid model (in-house strategist + agency ops) = sweet spot for growth-stage SaaS.

Unsustainable growth signals: (1) ROAS declining >10% week-over-week despite increased spend (audience fatigue or quality drop), (2) CAC rising faster than revenue (margins compressing), (3) CTR/conversion rate declining (creative fatigue), (4) New cohort LTV lower than older cohorts (lower-quality customers acquired). Scale sustainably: increase spend 20-30% if LTV:CAC >3:1, hold spend if 2-3:1, decrease if <2:1. Refresh creatives every 2 weeks (Meta frequency score >8 = stop). Use cohort-based LTV not short-term ROAS — a campaign with 1.5x ROAS but 10x LTV should scale, a campaign with 3x ROAS but declining LTV should pause.

MTA = digital-only, clickstream-based (user saw ad → clicked → converted), works for e-commerce and high-conversion-velocity businesses, breaks down post-iOS 14.5. MMM = aggregate-level, uses econometric modeling to estimate each channel's contribution to revenue, works for all channels (TV, radio, paid search, organic), but can't attribute individual conversions. For paid ads budgeting: use MTA for tactical daily/weekly optimization (which campaign drove this conversion?), use MMM for monthly strategic planning (should we shift $100k from Google to Meta?). Combine both: MTA feeds daily campaign decisions, MMM validates long-term budget allocation every quarter.