The Marketing Brain

What people actually buy (it's never the product)

People hire products to make progress in a specific situation. McDonald's morning milkshake didn't compete with other shakes — it competed with bagels, bananas, and boredom on a 30-minute commute. The "job" was making the drive less dull and staying full until lunch. Once you see purchases as hires, you stop competing on features and start competing on the progress the buyer is trying to make.

So the template I write for every product before I touch a headline: "When [situation], I want to [motivation], so I can [expected outcome]." If I can't fill that in from real buyer conversations, I'm guessing — and guessing is what makes landing pages convert at 0.4% instead of 4%.

The five awareness levels — and why most copy talks to the wrong one

The biggest single reason solo founders burn ad budgets is writing copy for an audience that doesn't exist yet. Your prospect sits at one of five levels, and the wrong message at the wrong level kills the sale instantly.

• Unaware — doesn't know they have a problem.
• Problem-aware — feels the pain, doesn't know a solution exists.
• Solution-aware — knows solutions exist, doesn't know yours.
• Product-aware — knows your product, hasn't bought yet.
• Most aware — knows you, trusts you, just needs the right offer or trigger.

Most founders write "Buy now, 20% off" copy (level 5) and point it at cold traffic (level 2). The cold audience has no idea who you are. That's not a conversion problem — that's a stage mismatch. Cold ads and content should always speak to problem-aware and solution-aware. Save the offer-driven copy for retargeting and email sequences where the buyer has already let you in.

Why loss aversion changes everything (including pricing)

Losses hurt roughly 2.5x more than equivalent gains feel good. It's the most replicated finding in behavioral economics, and it should be rewriting every headline you have. "Stop losing $500/month on bad trades" beats "Make $500/month more on trades." "Don't miss the AI wealth transfer" beats "Join the AI wealth transfer." The pain of missing out is a stronger motivator than the pleasure of gaining — every time.

The same mechanic explains why annual subscribers stay 2.8x longer than monthly (40 months vs 14, per Buffer's data). Annual stacks three quiet psychological mechanisms: status quo bias (cancellation feels like effort), sunk cost ("I paid for the year, I should use it"), and the endowment effect (people overvalue what they already own). None of this is manipulation — it's how the brain is wired. The job of pricing architecture is to align your offer with how decisions actually get made.

The persuasion stack I put on every landing page

Six mechanisms drive almost every yes a human gives: reciprocity (give first, people feel obligated to give back), commitment (small yeses lead to big yeses), social proof, authority (credentials, data, media), liking (we buy from people we identify with), and scarcity (limited availability raises perceived value). A seventh — unity, the shared-identity move ("we ADHD folks") — is the strongest of all when it fits.

Most solo-founder landing pages run zero of these. Just a feature list and a buy button. Minimum viable persuasion is three: one specific social proof element near the CTA, one authority signal (a real number, a credential, a data point), and one honest scarcity element (a deadline or a cap). Stacking three beats a perfect single hook every time.

One last frame that changes how I think about audiences over 45. Younger markets buy accumulation — more stuff, more status. Markets over 45 buy preservation — vitality, legacy, relevance. Selling accumulation to someone who's already accumulated is why most marketers fail with older demographics. Match the message to the life stage and the same product converts at a wildly different rate.

The first 8 seconds (and the next 15 minutes)

The textbook funnel — Awareness, Interest, Decision, Action — is a lie that costs solo founders real money. Real SaaS buying looks like this: a person ignores a problem for weeks, the pain spikes, they Google it, they find 3-5 options in 10 minutes, and they eliminate 2-3 of those options in roughly 5 seconds based on website first impression. Then they try 1-2 trials simultaneously, use one for about 15 minutes, and quit if it doesn't click. The actual buying decision is made twice — at the website impression and inside the first 15 minutes of use. Everything else is theater.

This means your landing page is doing 30-50% of the conversion work through headline clarity alone. Eye-tracking shows visitors read 20-28% of text on a page, scanning in an F-pattern: first line, down the left, occasional full line, leave. The blur test catches it — squint at your page; if you can't identify what it offers and where to click, your visual hierarchy is broken. Three questions must be answered without scrolling: Can I tell what this is? Is it for me? What should I do? Miss any one and the visitor is gone before your features ever load.

Why annual subscribers churn at renewal — and what that actually means

Annual subscribers stay an average of 40 months versus 14 months for monthly — but the renewal moment is brutal. Monthly churns at 8.5-12%; annual-equivalent churns at 2.4%. The mechanism is the sunk cost effect: paying upfront forces usage, and usage builds habit. But the year-one renewal cliff exists because 40-60% of subscribers stop reading the emails by month 8, forget the product exists, and cancel when they see the charge. The #1 cancellation reason isn't "the product is bad" — it's disengagement plus the feeling they never acted on anything.

• Optimal annual discount: 2 months free (16.7%) — less doesn't justify the commitment, more devalues the product
• Display annual as monthly-equivalent ("$23/mo billed annually") — the framing alone increases annual uptake
• Don't push annual at signup; let new users go monthly first, then upsell at day 30-60 once they've experienced value
• Build the re-engagement email at month 4 and the value-summary email 60 days before renewal — both prevent the silent-cancel pattern
• 20-40% of SaaS churn is involuntary (expired cards, failed payments) — automated dunning recovers 50-70% of that with zero creative effort

The 2026 AI-native retention crisis

AI-native B2B SaaS is posting a median gross retention of just 40% versus 92-95% for traditional SaaS. At $100M ARR, a 40% gross retention company churns $60M annually and needs $70M in new ARR just to grow 10%. The math doesn't work. Three mechanisms are doing the damage: prompt portability (the prompt is the product, and prompts move freely), novelty decay (the wow wears off in weeks), and capability mismatch (users showed up expecting the demo and got the median session). The Glass Slipper Effect makes it worse — first-cohort adopters lock in deeply, but later cohorts churn 3-5x faster.

The cognition bottleneck explains the rest. AI capability is becoming abundant; human integration into actual workflows is the scarce resource. Roughly 85% of AI use generates no measurable business value because the product showed up before the workflow did. This is why traditional SaaS that adds AI features sees retention improve (one case study cut churn 42%) while AI-native products bleed. The cyborg model wins; full replacement loses.

The fix isn't more model capability — it's architecture. Design accumulated user data as the switching cost from day one. Tier pricing so heavy users pay for heavy use (Anthropic's $17/$100/$200 tiers target the top 5% of consumption). Get users through onboarding inside 7 days — that single threshold lifts 12-month retention by 72%. Use AI to predict churn 60-90 days out, then intervene with a human, not more automation. AI can prevent up to 71% of churn when paired with human follow-up; alone, it accelerates the problem. Retention is now a habit-design discipline, not a feature-shipping one — and the products treating it that way are the only ones that will still be here in 2027.

Why most products miss the market

The fatal pattern is almost always the same: a technical founder builds the thing that excites the builder, not the thing that converts the buyer. You are not your customer. Tech stack, model choice, data pipeline — buyers don't care. They care about how their life changes Tuesday morning after they pay. "Powered by GPT-4" is a feature; "be the person who saw the trend before CNBC reported it" is an outcome — and outcomes are the only thing wallets respond to.

This is why 42% of startups die from "no market need" and another ~30% die from marketing and sales problems. Building is maybe 20% of success. Distribution is 80%. Thiel's line is uncomfortable but accurate: superior distribution alone can create a monopoly with no product differentiation; the reverse is not true.

The trap is real because building feels productive and distribution feels icky. So solo founders spend 95% of their time in the IDE and 5% in front of buyers, then act surprised when launch day generates crickets. Invert it the moment the MVP works — and run the idea through the Durable Work Filter: could a new ChatGPT user replicate this next month? If yes, the engine isn't the moat. Pick a topic where the moat is in the right place: domain expertise, proprietary data, distribution relationships, or compliance knowledge in a niche too unsexy for the next wave to touch.

The pricing mistake everybody makes

Solo founders consistently underprice by 2-5x. ProfitWell studied 5,000+ SaaS companies and the ones who raised prices grew roughly twice as fast as the ones who competed on price. The reasons founders underprice are predictable: projection ("I wouldn't pay that"), fear of rejection (lower price = lower stakes if nobody buys), and imposter syndrome dressed up as humility. None of those are pricing signals. They're emotional bandages.

Worse, low pricing trains the market against you. Launching cheap "to get traction" anchors expectations, attracts price-sensitive customers who churn the second you raise, and makes it psychologically brutal to ever charge what the work is worth. The fix is unglamorous: double your planned price and test it. If buyers don't object to the number but instead ask "what do I get for this?" — the price is fine and the offer needs sharpening, not discounting.

There's a counterintuitive layer on top: for considered purchases, higher prices often increase BOTH perceived value and conversion. When nobody's buying at $279, dropping to $99 doesn't fix it — the problem is messaging, targeting, or trust. Cutting price just hides the real issue under a smaller number.

Features vs outcomes — the rewrite test

Walk every headline on every page through one filter: does this describe what the product DOES, or how the customer's LIFE CHANGES? If it's the first one, rewrite it. Features tell you nothing about whether to pull out a credit card. Outcomes do. And remember the scanning reality — Nielsen Norman's eye-tracking shows users read roughly 20-28% of text on a page. The first headline is the only thing most visitors actually consume. Body copy is for the small slice who've already decided you might be worth their time.

Two final mechanisms that quietly kill solo-founder launches. First: confused buyers don't buy. Hick's Law is brutal — every extra choice on a pricing page, every extra CTA, every extra nav item is cognitive load that taxes conversion. The jam study went from 30% conversion at 6 choices to 3% at 24. Strip ruthlessly. Second: treating the launch as an event instead of a process. The Lifetime Deal → community → MRR pattern, the demand-harvesting habit of answering people who are RIGHT NOW posting "does anyone know a tool that does X," the weekly content cascade — these are systems, not moments. Founders who keep shipping into a thin distribution channel beat founders who polish in private and stage one big reveal. Less product, more distribution. Constraints over optionality. Ugly-but-shipped over polished-but-waiting. Every time.

The guru-trust formula that built a $1B industry

Subscribers are not buying stock picks. They are buying certainty in an uncertain world, belonging to an "informed insider" group, hope of getting ahead, the convenience of having research done for them, and — most underrated — permission to act. Information-overloaded investors are paralyzed; they need an authority figure to say "buy this now." Strip those five jobs out and you are left with a PDF nobody renews.

That is why the most durable finpub businesses are built around a persona, not a publication. Balanced analysis does not sell subscriptions; CNBC gives that away for free. Subscribers pay for someone willing to say BUY or SELL with conviction and reasoning. The guru supplies five ingredients no spreadsheet can:

• Authority — credentials, track record, insider connections that justify the seat at the front of the room
• Personality — a distinctive voice the reader feels they "know" after three emails
• Conviction — clear calls with reasoning, not hedged "on one hand, on the other hand" analysis
• Accessibility — first-person, direct address, the parasocial feel of a private letter
• Narrative — a worldview (a "big idea") that re-explains the market in one provocative sentence

The economics also explain the long sales letters. The $49–$99/year front-end is a loss leader; the real money lives in the $500–$5,000 back-end services sold to proven buyers. Always ship at least two tiers from day one — an ultra-premium tier exists to anchor the middle tier as reasonable. Even 15–20% migration from entry to mid-tier lifts customer lifetime value by ~72%, and cross-selling an existing subscriber costs $5–$15 versus $100+ for a new one. Retention is 5–7x cheaper than acquisition; the math always favors it, but founders find acquisition more exciting.

Why email beats every other channel — when it's sequenced right

Email delivers roughly $36–$45 in return per $1 spent, a 72:1 ratio nothing else in the stack touches. The catch: that number assumes you are sequencing, not broadcasting. Three sequence types do almost all the work: a welcome series (typically 3–7 emails, ~44% open rate on email #1), a nurture sequence (10–13 emails over 30 days, converting 12–18% of leads), and a launch sequence (5–12 emails, 40–50% open on launch day). The common mistake is treating the inbox as a megaphone instead of a relationship.

Frequency is where most operators leave the most money on the table. A single send produces ~9% response; one follow-up bumps it to ~13%; multi-step sequences hit ~27% with warm audiences. Fear of over-emailing costs far more revenue than unsubscribes ever will — and unsubscribes are a feature, not a bug. A smaller engaged list beats a large dead one every time, so resist softening the voice to reduce churn; that waters down whatever attracted the best subscribers in the first place.

The emails that actually convert

Beyond the sequences, a handful of email shapes carry most of the revenue. The ultra-short curiosity email (3–5 sentences, one hook, one link) is the 80/20 of the inbox. The belief-buster opening leads with the reader's #1 limiting belief instead of the product — people are more motivated by removing internal blocks than gaining external benefits. The myth-busting frame names the lies the industry tells, then positions you above the fight. The extended-metaphor email runs one image start to finish so a 500-word email feels short. And the P.S. remains the second-most-read line in any sales letter — never waste it on a signature.

The newest variable is the inbox itself. Gmail Gemini and Apple Intelligence now generate AI summaries before the email is opened, and ~90% of inbox usage runs through Gmail + Apple Mail. The subscriber may never read the words as written. Three adjustments: lead with the value in the first 100 words (the AI summarizer treats your opening like a headline), keep one idea per section, and use descriptive headers over clever ones. Plain-text formats are quietly winning again — they render consistently in AI-summarized views and feel unmistakably human against the rising tide of AI sludge. Use AI to brainstorm and restructure; never let it write the whole email. When everything sounds the same, nothing stands out, and in email, voice is the moat.

The affiliate math that makes ClickBank work

Digital products have near-zero marginal cost, which is why 50-75% commissions are sustainable and why anything less than 50% gets ignored by serious promoters. The first 10-20 active affiliates are disproportionately valuable because gravity is a self-feeding signal — gravity 20+ legitimizes you to mid-tier affiliates who drive real volume. The fastest path is direct recruiting from the affiliate pool of adjacent products plus an exclusive 90% rate for first movers and a cash bonus to the first affiliate who sends 50 sales.

The economics only hold if refunds stay under 8%. Most refunds happen between days 3-10 — not because the product failed but because buyers got overwhelmed and stopped engaging. A single early "quick win" email guaranteeing a visible outcome within 20 minutes can cut refund rates by 30-40%.

Front-end economics get built through AOV stacking. A $5 core that 100% of buyers take, an upsell #1 at $197 (20% take), a downsell at $97 (10%), a second upsell at $67 (10%), with its own downsell — pushes a $5 product to roughly $62 AOV. Two rules matter: the biggest upsell must come immediately after the initial purchase (never after a low-value decision that drains willpower), and the order bump should be a done-for-you tool, never more training. Buyers at checkout are in action mode — they want shortcuts, not more lessons.

Native ads vs display: why the gap is 4-8x

Native ads (Taboola, Outbrain, Adbytes, MGid) succeed by being mistaken for editorial content. CTR is driven by curiosity headlines; conversion is driven by what comes next. The advertorial bridge page is not optional — sending native traffic directly to a sales page doubles CPA from roughly $23 to $50 on the same clicks. Native users expect content; a direct sales page triggers immediate bounce. The advertorial IS the conversion mechanism on native.

Realistic cold-traffic CPA bands on a $35 info product for the 45-65 demographic: Meta $40-80, Google Search $50-120, Native $30-80, Reddit $60-120. Native wins on cost per acquisition when paired with a real bridge page; it loses badly without one. CPMs are $5-15, CPCs $0.25-0.60 — the cheapest legitimate cold traffic for education, finance, and health angles. Worst for anything requiring UI demonstration.

Creative discipline matters more than placement. Plan 10-20 headline/image combos knowing 80% will underperform and 2-3 will carry the campaign. Faces beat objects; surprising juxtapositions (a 60-year-old using new tech) beat stock. Run $30-50/day for 7 days minimum before judging — and assume any policy slip on income or AI claims triggers a weeks-long account review.

Direct response for the 50+ buyer

The 50+ buyer isn't trying to get rich — they're trying to feel relevant and capable. "Making money with AI" is the proximate motivation; the underlying one is "proving I can still learn new things." Marketing that touches that deeper motivation outperforms pure income claims by 3-4x. They're not more skeptical — they're differently skeptical. Burned by decades of make-money-online schemes, they respond to authority and specific social proof, not hype. Avoid "passive income," "side hustle," "hustle," "grind." Use "consistent income," "reliable system," "works on your schedule." Testimonials should show modest but believable results ($200-$800/month), never six-figure fantasies.

Every 50+ prospect is asking three questions in this order: Is this a scam? Can someone like me actually do this? Will I be left alone to fail? Address them explicitly, in that order, on the page. The sales-page structure that converts:

• Credibility anchor first — name, number, association with something real, before any promise
• Identity mirror — a specific person, age, prior occupation, specific dollar result
• Mechanism in plain English — never use "AI," "algorithm," or "automation" in the first screen
• Proof stack — your claim is weakest; testimonials are mid; demonstration is strongest. Layer all three
• Risk reversal repeated twice, with an objection-killer line above the button: "You don't need to know anything about computers beyond email"

The persuasion equation under all of it: Urgent Problem + Unique Promise + Unquestionable Proof = Irresistible Offer. Remove any one element and conversion collapses. Specificity is the credibility multiplier — "$4,327 in your first 30 days" outperforms "$5,000" because the odd number implies it was measured, not invented. (And specificity here means the product's capability, never a fabricated buyer outcome — invented case studies are the fastest path to a legal letter.)

One last discipline: the Power of One. One big idea. One core emotion. One reader. One promise. One call to action. Every "additional benefit" that gets bolted on as insurance dilutes the central idea by 30%. The 50+ audience punishes complexity hardest — one big promise, one clear action, maximum three steps to purchase. That's the entire game.

The 7-month organic search reality

Organic search still posts the best long-run economics of any channel — 702% ROI on the SEO playbook — but the breakeven point is seven months, and that hasn't moved. What HAS moved is the click-through rate: organic CTR dropped 54.6% in a single year as Google AI Overviews ate the top of the SERP. Traditional ranking still pays, but only if you also rank inside the AI summary itself. If you don't, you're ranking for an audience that no longer scrolls.

The small-site path that works in 2026 is unchanged in shape: pick a niche narrow enough that authority sites haven't bothered, build a pillar page plus 8-12 cluster pages, and grind for keyword specificity arbitrage — queries with KD under 10, 50-1000 monthly searches, where a Wikipedia result at #1 signals a gap. "AI skills for people over 50" is winnable. "AI skills" never will be. Technical SEO is one-day work. After that, 10 high-quality links beat 100 directory links every time.

GEO/AEO — optimizing for AI answer engines

The mechanism is structural, not stylistic. AI engines ignore flowery prose and love extractable facts in tables, bulleted lists, and FAQ blocks. They cite original research, named case studies, and verifiable data they can't get from a thousand other re-summarized blog posts. They cannot cite a paragraph that hedges. Schema markup (FAQ, HowTo, Product, Breadcrumb), an llms.txt file at the site root, and a clean machine-readable pricing block now do work that meta descriptions used to do.

The four GEO/AEO actions to take this month:

• Add Schema markup (FAQ, HowTo, Product) to every existing money page — twenty minutes per page, permanent payoff
• Drop an llms.txt file at the root of each site so AI crawlers know what to index and how to summarize you
• Restructure your top three pages into extractable fact blocks: short paragraphs, real numbers, named sources, tables for comparisons
• Publish one piece of original research or first-party data per quarter — something AI engines literally cannot get anywhere else, so they have no alternative but to cite you

The deeper shift is that the web is splitting into two layers. The Human Web is hero sections, salesy copy, emotional triggers. The Agent Web is APIs, structured data, MCP endpoints, bot-to-bot transactions where discovery and purchase happen without a human loading the page. Build only for human eyes and you're invisible to the agent layer. Build only for agents and you lose the trust humans still need before delegating to their AI. The answer is both.

Why YouTube Shorts beat TrueView on CPV

YouTube Shorts ads run $0.10-$0.30 CPV — about a 56% cost saving versus TrueView in-stream. The catch is what they convert. YouTube as a direct-purchase channel converts at 0.05-0.5%; as a lead-gen channel feeding email it converts at 40-60%. The whole platform is top-of-funnel. The right architecture is sequential: Shorts for awareness, Discovery and Demand Gen for consideration (Demand Gen is showing 40% lower CPC for early adopters), email for the close. Sending YouTube traffic straight to a checkout page is the most common waste of budget on the platform.

For Shorts creative, the iron rule is the first five seconds. Roughly 70-80% of viewers skip in-stream ads at the 5-second mark, so the first five seconds aren't your hook — they ARE your ad for most of the people you paid to reach. Talking-head outperforms animation for trust, vertical native production beats horizontal adaptation by 10-20% on conversion, and 30-60 second creatives hit the sweet spot before the 60-second drop-off cliff.

Reddit, TikTok, LinkedIn — pick by audience, not hype

The real Reddit playbook is mechanical: manual subreddit targeting (disable automated, it wastes budget), 1-3 months of organic karma first, free-form ads that look organic, modest test budgets. TikTok Spark Ads — boosting your already-proven organic posts — drive 30% higher completion and 142% higher engagement than standard in-feed at CPMs of $2.60-$6.60, meaningfully cheaper than Meta. LinkedIn stays expensive ($5-$15 CPC) but drives 75-85% of all B2B social leads and only works when each lead is worth $500+. Thought Leader Ads from a personal profile beat company-page content by 2-3x. Channel choice in 2026 is downstream of who you're selling to, not whichever platform has the cheapest CPM that week.

Why Substack and Medium beat zero-volume social

If you don't already have an audience, posting on Twitter or Facebook into the void is one of the worst time-investments in marketing. Zero-volume social posts get zero distribution — the algorithm rewards what's already getting engaged with, so brand-new accounts without paid promotion are invisible by design.

Substack and Medium work the opposite way. Both publish your content on their own high-authority domains, which means your articles get indexed by Google immediately and ride the platform's domain authority. AI engines (ChatGPT, Perplexity, Gemini, Claude) cite Substack and Medium posts heavily — they treat these platforms as trustworthy primary sources. A single thoughtful Substack post can pull more qualified traffic in six months than 300 tweets ever will.

The deeper logic: backlinks from high-authority domains are still the #1 organic ranking signal, and Substack + Medium hand them to you for free. Posting on your own low-traffic domain alone is fighting the algorithm with no leverage; using these platforms as an indexing layer plus referral funnel is using their authority against itself.

Why the ascension model actually works

The ascension model works because it splits your business into two machines with completely different jobs. The front end is a Customer Factory — its job is to manufacture paying customers at break-even. It is a cost center disguised as a product. The back end is a Profit Factory — that is where 100% of the money lives. Once you accept that the $5-$50 front-end product exists to acquire customers (not to make you rich), everything else clicks. You stop worrying about margin on the tripwire. You start over-delivering wildly — $1,000 of real value for $37 — because the back end pays for the generosity.

The 70/20/10 split nobody talks about

Most of your audience will never buy past the front end. That is not a failure — it is the design. The natural distribution: 70% DIY (they want the $5-$50 product and a PDF and they are done), 20% Done-With-You ($2K-$10K group coaching, one-to-many delivery), 10% Done-For-You ($25K-$100K+ implementation). Inside that 10%, roughly a tenth of total customers ends up generating something like 80% of the revenue. This is the math that makes a $5 book funnel a multi-million-dollar business. You are not "converting everyone" — you are building a wide enough top of funnel that the small slice of high-intent buyers reveals itself, then you have a real offer ready for them. The implication: every front-end product needs a credible next rung. If your $37 book has no path to a $497 cohort and a $5K done-with-you tier, the 10% who would have paid you $5K never get the chance. You are not leaving money on the table — you are leaving the entire table.

The 7-11 day buying window and the AOV stack

After someone buys, you have a measurable burst of momentum: roughly 80% of follow-on purchases happen within 7-11 days. They are in buying mode, card warm, identity shifted ("I am someone who buys this kind of thing"). After that window, attention decays fast — by day 30, most have forgotten you exist. Compress the whole post-purchase sequence into the window: instant delivery, first upsell within seconds, daily value content days 1-5, a belief-shifting "awareness bridge" piece day 5-7, call invitation day 7-10. A 30-day nurture is a 30-day forgetting curve.

The AOV math is what funds it. A clean stack on a $5 core: core $5 (100%) + upsell #1 at $197 (20% take = $39) + downsell at $97 (10% of refusers = $10) + upsell #2 at $67 (10% = $7) = roughly $60+ AOV from a $5 product. Target rule: AOV should hit 10x your core price. A $5 book wants $50+. A $37 front end wants $370+. Two non-obvious rules: put the biggest-dollar upsell FIRST (every decision drains willpower — spend it on the high-value choice), and cap at 2-3 upsells before refund rates spike across the entire order.

Diagnose before you scale spend

When revenue dips, the reflex is "the funnel is broken — kill it." It almost never is. 90% of the funnel is fine; one component broke. Before throwing more at ads, audit these four separately:

• Offer — Is the value-to-price ratio still 5-10x? Did the guarantee weaken? High refunds = product problem, not marketing problem.
• Creative — CPM crept up or CTR dropped? Creative fatigues; the offer didn't get worse, the hook got stale.
• Audience — Are you mining the same lookalike to the bone? Saturation looks identical to a "broken" funnel on the surface.
• Economics — Watch for "the wobble": at 50-100+ sales/day, CPA drifts up while AOV drifts down. If the lines cross, the offer tanks. Refresh creatives and re-fire upsells BEFORE they cross.

One more lever almost nobody pulls: the anxiety audit. Print every page of the funnel. Mark each element red (adds buyer anxiety) or green (removes it). Refund policy on every page, visible phone number, one-step checkout, "here's exactly what happens next" messaging — these move conversion more than any new persuasion gimmick. Persuasion and anxiety cancel each other out. Make people feel safe first, then sell.