When Affiliate Programs Cut Commissions: Documented Cases and How to Protect Your Income

Every affiliate income stream runs on terms the other side can change. Here is what the documented record shows, and the playbook that survives it.

Published July 17, 2026

Affiliate programs can and do cut commission rates, usually with 30 days' notice or less: the reference case is Amazon Associates on April 21, 2020, when most category rates were cut, several by more than half, with about two weeks' warning. Our changelog, which diffs all 160 tracked programs daily, has logged 0 field-level changes so far (0 material). The defense is structural, not predictive: diversify across programs, know your terms' change clauses, and monitor the record at /changes.

The uncomfortable premise of affiliate income

An affiliate commission is not a contract you negotiated. It is a standing offer the merchant publishes and reserves the right to modify, and nearly every program agreement contains a clause saying exactly that: rates, cookie windows, qualifying actions, and the program's very existence can change at the merchant's discretion, usually effective on short notice, sometimes effective immediately.

Most affiliates read that clause once, if ever, and then build income projections as if the current rate were a law of nature. The documented record says otherwise, and the pattern behind cuts is consistent enough to be worth learning: rates get set generously when a merchant is fighting for distribution, and get cut when the merchant's brand is strong enough that the traffic arrives anyway. Commission generosity is a bidding war, and bidding wars end.

The reference case: Amazon, April 2020

On April 21, 2020, Amazon Associates, the largest affiliate program in the world and the income backbone of a huge share of content sites, cut commission rates across most product categories with roughly two weeks' notice. The cuts were reported at the time by CNBC and across the industry press, from Amazon's own notice to associates:

CategoryBeforeAfter
Furniture, home, home improvement, lawn & garden8%3%
Grocery5%1%
Headphones, beauty, musical instruments, business supplies6%3%
Health & personal care4.5%1%
Sports, outdoors, tools5.5%3%
Amazon Fresh3%1%

Sites earning most of their revenue from home and garden content lost more than 60% of that revenue in a fortnight, with no recourse and nothing mislabeled: the terms always allowed it. Three durable lessons came out of that fortnight.

Since 2020, repricing has stayed routine across the industry: network migrations that reset terms, recurring deals gaining 12-month caps, first-payment percentages quietly trimmed. Most cuts are smaller than Amazon's and completely unannounced, which is why documentation matters more than anecdote.

What a daily diff of 160 programs actually records

Because program pages change silently, we snapshot every tracked listing daily and publish the diff. The dataset is young, and honesty about that is the point of citing it: as of 2026-07-17, the changelog holds 0 field-level changes, of which 0 are material (commission, cookie, network, recurring status, or grade), and 0 of those moved against the affiliate.

Two properties make this record worth more than its size. It cannot be backfilled: a change nobody logged on the day it happened is gone, which is why running the log daily from now beats any retrospective. And it is public: when a program's terms move, the before and after sit on the record whether or not the program announced anything. Every material change also goes out in the weekly digest, which you can join from the homepage.

Reading a program's terms for change risk

Before committing serious traffic to any program, spend ten minutes on four questions. The answers are in the affiliate agreement, and where they are missing, that absence is itself the answer.

The protection playbook

Cap any single program's share of income. The Amazon lesson in one rule: when one program passes roughly a third of your affiliate revenue, expanding elsewhere earns more per hour than optimizing the leader. The point of the cap is not pessimism about any specific program; it is that you cannot price a risk the counterparty controls.

Diversify by counterparty, not by link count. Ten programs on one network still share a failure mode (the network), and ten categories in one program share everything. Real diversification is across programs, networks, and commission models, one reason to hold both recurring streams and flat-bounty deals; the trade-offs between those two are worked through in recurring vs one-time commissions.

Prefer terms with structural commitment. A program that states a notice period, answers the grandfathering question in writing, and has a stable rate history is materially safer than one with identical headline numbers and silence on all three. Rate history is exactly what a changelog gives you; category norms to compare against are in commission rates by category.

Monitor mechanically, not by vibe. Nobody rereads terms pages for fun. Subscribe to something that does it for you (our changelog and weekly digest exist for precisely this) and calendar a monthly skim of the terms for your top three programs. Ten minutes a month is the entire cost of never being surprised by a cut that was published quietly.

Keep switching costs low. The affiliates who absorbed April 2020 best were the ones whose content recommended the best tool in a category rather than being structurally welded to one merchant. Write so that swapping the recommended program in a category costs you an afternoon, not a rewrite. Comparison-shaped content (like our compare view) has this property built in.

Why programs cut rates: the four triggers

Cuts look arbitrary from the receiving end, but they cluster around four causes, and knowing which one is in play tells you whether the cut is a one-off correction or the start of a slide.

Customer-acquisition rebalancing. Affiliate commissions compete with paid ads for the same budget. When a merchant's other channels get cheaper, or when finance decides affiliates are being paid for customers who would have arrived anyway (the Amazon logic), the rate drops. These cuts are usually deep, permanent, and announced in bulk across categories.

Ownership or margin pressure. Acquisitions, new investors, and profitability pushes reliably reach the affiliate line within a few quarters, because it is one of the easiest costs to trim without touching the product. A program whose parent company just changed hands deserves a closer watch for the next year.

Category maturity. Rates get set high when a market is a land grab and drift down once the winners are established. VPNs and AI writing tools both show the pattern: early entrants paid whatever it took for distribution, and rates compress as brands consolidate. If a category's generosity is a bidding war, plan for the armistice.

Fraud and quality waves. After a program absorbs a wave of incentivized or fraudulent traffic, the response is often stricter qualification terms rather than a headline cut: longer retention requirements, tighter geo rules, slower payouts. Your rate survives but your effective earnings drop. Watch qualification terms with the same attention as the percentage.

Warning signs that precede the email

None of these signs is proof, and that is not the standard. You are not trying to predict cuts; you are deciding which programs deserve the concentration you are giving them, with the change clauses priced in.

What a cut should trigger, step by step

Methodology

Changelog figures (0 changes, 0 material, 0 against the affiliate) are computed at build time from our daily snapshot diff of all 160 tracked programs, public at /changes; the dataset began recently and grows continuously, and we state its youth rather than dress it up. The Amazon April 2020 figures come from Amazon's notice to associates as reported contemporaneously by CNBC and the industry press. No other specific historical cut is quoted here because we hold no primary documentation for one; when our own log accumulates enough history, this page will cite it instead of anyone's memory. Program Scores referenced across the site rate terms, not payout behavior; see the methodology.

FAQ

How much notice do affiliate programs give before cutting commissions?
Anywhere from about 30 days to effectively none. Amazon's April 2020 cuts came with roughly two weeks' notice. Most program agreements reserve the right to change terms at any time, and notice periods, where they exist at all, live in the terms of service rather than the marketing pages.
Do existing referrals keep the old commission rate after a cut?
It varies by program, and most terms pages do not answer it in advance. Some programs grandfather active referrals at the old rate, others apply the new rate to all future payouts including from customers you referred years ago. The only reliable way to know is to ask the affiliate manager in writing before you build a business on the stream.
What was the biggest documented affiliate commission cut?
Amazon Associates on April 21, 2020: most category rates were reduced, several by more than half in one step, including furniture and home improvement from 8% to 3% and grocery from 5% to 1%, with about two weeks' notice. It was widely reported at the time, including by CNBC, and reshaped many content businesses' revenue overnight.
How can I find out when a program changes its terms?
Programs rarely announce cuts loudly. Practical monitoring: our public changelog diffs every tracked program daily and lists each change; the weekly digest emails the material ones; and for programs you depend on heavily, reading the terms page monthly is worth the ten minutes.
What share of program changes are cuts rather than improvements?
Too early for a stable answer from our young dataset: 0 material changes logged so far, 0 of them in the affiliate's disfavor. The number that matters more is that changes happen at all, on terms you do not control, which is an argument for diversification rather than prediction.