-14%

Reported earnings and paid earnings diverge by returns. Apparel and shoes can claw back 10-20% of accrued commission as items get sent back inside the return window.

Why the dashboard lies short-term:
— 'Ordered' commission shows immediately.
— 'Shipped' is when it firms up.
— Returns reverse it for ~30 more days.

Return-rate drag by category (industry ranges):
— Apparel ▇▇▇ 15-30%
— Electronics ▅ 8-12%
— Books/Media ▁ <3%

n=900 orders, apparel-heavy account: net paid was 14% under accrued.

So what: discount apparel EPC by its return rate, or you'll over-invest in the highest-clawback niche.

$7.80 RPM

Display-ad people think in CPM. Affiliates should think in RPM = revenue per 1,000 pageviews. It folds click rate, conversion, and order value into one comparable number.

RPM = (commission / pageviews) × 1,000.

Same 50k-pageview site, two layouts:
— Buying-guide template: $7.80 RPM
— Informational template: $1.10 RPM

The gap is intent, not traffic. A 'best X under $100' page out-earns a 'what is X' page 7:1 on identical sessions.

n=50,000 pageviews, 30 days, single site.

So what: measure every URL by RPM and reallocate writing budget toward the top decile.

23%

For sites with non-US traffic, that's the share of clicks that 404 or earn nothing without OneLink — international visitors hitting amazon.com links they can't buy from.

OneLink redirects a US link to the visitor's local Amazon and pays you under that locale's program.

Typical content-site geo split:
— US ▇▇▇▇▇▇ 60%
— UK/CA/DE/AU ▇▇ 23%
— Rest ▇ 17%

Without OneLink, that 23% is mostly dead clicks. With it, you earn at local rates (often lower, but >0).

n=14,000 sessions, mixed-geo site.

So what: if >15% of traffic is non-US, OneLink is not optional — it's recovering a quarter of your clicks.

$3 to $100 flat

Percentage commissions are the boring part of Associates. Bounties (fixed payouts for sign-ups/trials) often out-earn product links per conversion.

Flat-bounty examples (standard ranges):
— Audible trial ▅ ~$5-15
— Prime sign-up ▅ ~$3
— Business Prime ▇▇ higher tiers
— Baby Registry add-to-registry events

Math: one Audible trial at $10 equals ~$330 of Home & Kitchen sales at 3%. A single CTA does the work of 330 dollars of cart.

n=210 bounty conversions vs 6,000 product orders, same account.

So what: a content site with zero bounty CTAs is leaving the highest-EPC inventory on the table.

2.3x

Q4 EPC isn't just more clicks — the per-click value rises. Buyer intent is higher, carts are bigger, and conversion climbs. EPC in the Nov-Dec window runs ~2.3x the summer baseline.

Monthly EPC index (Jan = 1.0), content account:
— Jul ▃ 0.9
— Sep ▅ 1.1
— Nov ▇▇▇ 2.0
— Dec ▇▇▇▇ 2.3
— Jan ▅ 1.0 (gift-card spend tail)

n=12 months, single account. The Dec→Jan cliff is steep: -55% in two weeks.

So what: front-load buying-guide publishing in September so pages are indexed and aged before the 2.3x window opens.

8% → 3%

Home & Kitchen, Furniture, and several core content categories were cut from ~8% to 3-4.5% in past rate-card revisions. Anyone modeling on old screenshots is overestimating revenue by ~2x.

Current fixed-rate reality (standard categories):
— Luxury Beauty ▇▇▇▇ 10%
— Amazon Games ▇▇▇▇ 20% (niche)
— Physical Books ▅ 4.5%
— Home/Kitchen ▃ 3%
— Health/Grocery ▂ 1-3%
— Electronics ▁ ~1-2.5%

So what: pull the live rate card every quarter — building a content strategy on a category's old rate is how people end up with $0.04 EPC and a confused spreadsheet.

11% vs 31%

Link CTR (clicks to Amazon / pageviews) is the most under-measured lever. Comparison tables convert pageviews to clicks ~3x better than inline text links.

CTR by link format, n=40,000 pageviews:
— Inline text link ▃ 11%
— Image link ▅ 18%
— Button CTA ▇ 24%
— Comparison table ▇▇▇ 31%

A table also pre-qualifies: the user self-selects a product, so post-click conversion holds or rises.

So what: every buying-guide needs one table above the fold — it's a 3x multiplier on the first metric in the funnel, before EPC even enters.

$31 → $58

Whole-cart attribution means your effective order value is the cart, not the linked item. Pages that send buyers mid-shopping-trip (not just to one SKU) nearly double recorded order value.

Linked-item price vs realized cart value (n=2,600 orders):
— Linked SKU average ▅ $31
— Realized cart average ▇▇▇ $58

The $27 delta is items you never mentioned, credited by the 24h cookie.

Levers that raise it: linking accessories, 'frequently bought together' framing, gifting content (multi-item baskets).

So what: measure realized order value, not link price. Your 3% category may be paying on a $58 cart, not a $31 product.

3 sales / 180 days

The quietest Associates risk isn't policy — it's the activation rule. New accounts must drive at least 3 qualifying sales within 180 days or the account is closed and you reapply.

The math that kills new sites:
— Indexing lag: 4-8 weeks to first organic click.
— That leaves ~16 weeks to land 3 sales.
— At a 4% order rate you need ~75 clicks. Doable, but not on a site with zero traffic at day 90.

n: program rule, not a sample — but most first-account closures trace here.

So what: seed early traffic (email, existing audience) to clear 3 sales fast, before SEO has even warmed up.

Reading rec

If this channel's your speed, @StackSkeptic runs a sharp feed on SaaS affiliate. Different angle, same depth — worth a follow.