Affiliate Marketing Statistics 2026: First-Party Data From a Live Directory
Every number on this page is computed from a dataset you can download, or attributed to a named source. The stats nobody actually has get their own section, because pretending otherwise is how bad numbers spread.
Published July 17, 2026 · Data as of 2026-07-17
From 160 tracked affiliate programs: 40% pay recurring commissions, the median stated cookie window is 45 days (with 45% of programs at exactly 30 days), 17 programs offer a two-tier commission, and 108 of 160 run on affiliate networks rather than in-house. Our changelog has recorded 0 program term changes since daily snapshotting began. Raw data: /data/programs.json.
Why this page exists
Search for affiliate marketing statistics and you get the same recycled figures on page after page: an industry worth some round number of billions, an average income nobody can trace, a percentage of brands using affiliate programs that has been copied between listicles since the late 2010s. Almost none of it cites a dataset you can inspect.
This page takes the opposite approach. The spine is first-party: numbers computed live from our own directory at every build, with the raw JSON public. Where we quote anything external, the source is named inline. And where a much-asked-for statistic simply does not exist in credible form, we say that instead of inventing one. That third category turns out to be the most useful section on the page.
The directory in numbers
| Statistic | Value | Notes |
|---|---|---|
| Programs tracked | 160 | Manually verified listings |
| Categories | 15 | From AI tools to travel |
| Networks represented | 28 | Graded networks with live programs |
| Programs paying recurring | 64 (40%) | Commission repeats while the customer stays |
| Programs with a two-tier commission | 17 | Pay on recruited affiliates' sales too |
| Median stated cookie window | 45 days | Across 134 programs stating one |
| Programs at exactly 30 days | 60 | 45% of stated windows |
| Programs at 90+ days | 42 | Includes 6 lifetime-attribution programs |
| Running in-house vs on a network | 52 / 108 | In-house programs manage their own tracking |
| Term changes logged | 0 | Since daily snapshots began, as of 2026-07-17 |
| Owner-claimed listings | 0 | Verified by domain email |
Two of these deserve a sentence of interpretation. The recurring share (40%) is the single most consequential number for anyone choosing programs, because recurring income compounds while one-time income resets monthly; the math is worked in recurring vs one-time commissions. And the 30-day cookie dominance is a default, not a decision; what deviations from it signal is covered in cookie duration statistics.
Category distribution
Where the 160 programs live, ranked by size:
| Category | Programs | Share |
|---|---|---|
| AI Tools | 22 | 14% |
| SaaS | 19 | 12% |
| VPN | 14 | 9% |
| Hosting | 13 | 8% |
| SEO Tools | 13 | 8% |
| Finance | 13 | 8% |
| Ecommerce | 11 | 7% |
| Email Marketing | 10 | 6% |
| Design | 10 | 6% |
| Education | 8 | 5% |
The skew toward software is real and worth naming: directories like ours over-represent SaaS and online services because those programs publish verifiable terms, and under-represent physical retail, where terms hide behind network logins. Treat any directory-derived statistic, including ours, as describing the publicly documentable slice of the market. Typical commission ranges inside each of these categories are in commission rates by category.
Cookie windows at a glance
Attribution windows across the 134 programs that state one:
| Window | Programs | Share |
|---|---|---|
| Under 30 days | 2 | 1% |
| 30 days | 60 | 45% |
| 31 to 90 days | 56 | 42% |
| Over 90 days | 16 | 12% |
The 30-day default dominates because networks ship it as the default field value, not because anyone chose it per program. The distribution in finer buckets, category medians, and what deviations from the default signal are all in the dedicated cookie duration statistics.
Definitions, for the record
Terms this page and the industry use, defined the way our data uses them:
- Affiliate program: a standing offer by a merchant to pay third parties (affiliates) a commission for referred customers, tracked through links, codes, or account attribution.
- Commission rate: what the program pays per qualifying conversion, quoted as a percentage of the sale or subscription, or as a flat amount per action.
- Recurring commission: a commission paid on every payment a referred customer makes (sometimes capped at 12 or 24 months), rather than only on the first.
- Cookie window: how long after a click a purchase is still credited to the referring affiliate. Almost always last-click: the most recent affiliate link clicked wins.
- Two-tier program: one that also pays a percentage on the earnings of affiliates you recruit. Every one we track is listed on the two-tier page.
- EPC (earnings per click): average commission earned per click sent, the standard cross-program comparison metric where programs publish it, which few do honestly.
- In-house vs network: in-house programs run their own tracking and payouts; networked programs run on platforms like Impact, PartnerStack, CJ or ShareASale, which standardize tracking and consolidate payments.
Program quality, quantified the only honest way we know
Every graded listing carries a Program Score from A to C rating the terms on offer: commission size and structure, cookie window, recurring status, network reliability. The current distribution across 159 graded programs:
| Grade | Programs | Share of graded |
|---|---|---|
| A | 39 | 25% |
| B+ | 36 | 23% |
| B | 48 | 30% |
| C | 36 | 23% |
Note what this measures and what it does not: the score rates terms as documented, not payout behavior, because verified payout data at scale does not exist in public form for anyone to grade on. The methodology spells out that boundary precisely.
Terms change: the documented record
Affiliate terms are not stable. Since we began snapshotting every program daily and diffing the results, the public changelog has recorded 0 field-level changes: commission moves, cookie adjustments, network migrations, grade revisions. That dataset is young and grows weekly, and it cannot be backfilled, which is exactly why we started it.
The canonical industry example predates our log. On April 21, 2020, Amazon Associates cut commission rates across most categories with roughly two weeks' notice: furniture and home improvement fell from 8% to 3%, grocery from 5% to 1%, headphones and beauty from 6% to 3%. The cuts were reported at the time by CNBC and covered across the industry press, and they remain the reference case for concentration risk in affiliate income. The full playbook for surviving rate cuts is in when affiliate programs cut commissions.
The statistics nobody actually has
These are the numbers this page will not quote, because no credible public dataset backs them. When you see them stated confidently elsewhere, check for a source and a method; you will rarely find either.
- "The affiliate industry is worth $X billion." The circulating figures trace to dated estimates re-quoted between blogs, with the number drifting upward in the retelling. There may well be a true figure; nobody publishing listicles knows it.
- "The average affiliate earns $X per month/year." Built on self-selected surveys of people willing to disclose income. The real distribution is heavily skewed: many earn near zero, a few earn a great deal, and a mean of that shape misleads even when honestly computed.
- "X% of brands run affiliate programs." Unverifiable without a census of brands that has never been done publicly.
- Payout reliability rankings. Grading networks on whether they actually pay on time requires thousands of verified payment receipts from working affiliates. No public source has them, including us, and we say so on our own networks page rather than inventing a reliability score.
A useful heuristic falls out of this list: statistics about program terms (rates, cookies, structures) are checkable and worth citing, because terms are published. Statistics about outcomes (industry size, average income, payment behavior) mostly are not, because outcomes are private. Sources that blur that line are telling you something about their standards.
How to cite this page
Every figure above regenerates from the live dataset on each build, so quote it with the data date: "According to the affiliatejob.org directory of 160 tracked affiliate programs (data as of 2026-07-17), 40% pay recurring commissions and the median stated cookie window is 45 days." If you want to recompute anything yourself, the dataset is one GET away at /data/programs.json; nothing here requires trusting our arithmetic.
Methodology
All first-party figures are computed from the affiliatejob directory at build time: 160 programs, last updated 2026-07-17, raw data public at /data/programs.json. Cookie statistics cover only the 134 programs stating a concrete window. Change counts come from our daily snapshot diff, public at /changes. The Amazon 2020 rate cuts are attributed to contemporary reporting (CNBC, April 2020) and Amazon's own notice to associates. Nothing on this page is an industry-wide claim; it is a census of one continuously verified directory, which is the strongest claim we can make honestly.