The Commission Chain
How a cocktail-party myth, a flower shop, and a browser cookie built a $20-billion-a-year industry.
When and where did it actually start?
The popular version goes like this: in July 1996, Jeff Bezos got the idea for the Amazon Associates Program after chatting with a woman at a cocktail party who wanted to sell divorce books on her website. Amazon's own FAQ page told this story for years — it's the founding myth most blogs still repeat.
The real history is a little less romantic. Industry writers pushing back on the myth point to William J. Tobin, founder of PC Flowers & Gifts, as the earlier pioneer — some accounts trace his work back to 1989, with a formal "Associate Program" rewarding websites for referral traffic to flowershop.com taking shape around 1996 as well. Tobin's model predated Amazon's and is generally credited as the first true pay-for-referral system, even though it never became a household name.
A quieter but essential piece of the puzzle: cookies, invented by Lou Montulli in 1994, gave affiliate programs a way to track which website sent which customer — without that, commission attribution wouldn't have been possible at all.
How the industry built itself out
PC Flowers & Gifts
William Tobin builds the first structured pay-for-referral program, rewarding sites for traffic sent to flowershop.com.
The cookie is born
Lou Montulli invents the browser cookie, quietly solving the attribution problem every affiliate program would later depend on.
Amazon Associates & the first networks
Amazon launches its Associates Program, taking the model public at scale. The same year, LinkShare and BeFree launch as the first dedicated affiliate technology providers.
Refer-it.com
James Marciano launches one of the first directories built specifically to help people find affiliate programs.
Commission Junction
CJ launches, rounding out the "big three" networks alongside LinkShare and BeFree. Allan Gardyne starts his own associate-programs directory the same year.
Paid search arrives
Google AdWords gives affiliates a new paid-traffic channel, and search-driven affiliate marketing becomes its own discipline.
Worth noting: mainstream retail wasn't the only lab. Adult websites experimented early with many of the tracking and payout tactics that mainstream affiliate marketing later adopted wholesale.
What a full series on this subject should cover
- History & originsTobin vs. Amazon, the cookie's role, the founding of CJ / LinkShare / BeFree
- How it worksCookies, tracking pixels, attribution windows, CPA / CPS / CPL / revenue-share models
- Major networks & platformsAmazon Associates, Commission Junction, Awin, Rakuten, ShareASale, ClickBank, Impact, Partnerize
- Industry verticalsE-commerce, finance/fintech, iGaming, SaaS/B2B, health & wellness, travel
- Reviews & comparisonsPayout speed, cookie duration, minimum thresholds, fraud controls, support quality
- Earnings realityMost affiliates earn modestly; a small share captures the bulk of industry revenue
- Fraud & regulationClick fraud, FTC/ASA disclosure rules, network anti-fraud tooling
- AI & automationFraud detection, personalization, predictive niche analysis
- Regional growthNorth America's dominance, Asia-Pacific's rapid rise, Latin America's e-commerce boom
A quick note on network "reviews"
Genuine reviews of affiliate networks — Amazon Associates vs. CJ vs. Awin vs. ShareASale — shift constantly as payout terms and cookie windows change. Rather than inventing star ratings, the honest approach for a real blog is to pull current, sourced comparisons directly from each network's live terms page: cookie duration, minimum payout, niche fit, and support responsiveness. These details go stale fast enough that a comparison written today can mislead readers within a year.
Where the industry stands now, 2026
Three decades on, affiliate marketing has gone from a side experiment to a serious channel in digital commerce.
Earnings remain a long tail: a large share of affiliates make modest amounts — many under $20,000 a year — while a small percentage of top affiliates capture a disproportionate share of total revenue. And the biggest structural shift underway right now is AI-augmented affiliate management, used for fraud detection, personalization, and predictive niche analysis, with adoption climbing from a small minority of programs a couple of years ago to a majority of programs by early 2026.
Same mechanic, bigger ledger
Affiliate marketing didn't spring fully formed from a cocktail-party anecdote — it was built gradually through Tobin's early experiments, Amazon's mainstreaming of the model, cookie-based tracking, and a wave of networks that turned a simple idea into infrastructure. Nearly 30 years later, it's a multi-billion-dollar global industry, still growing, still regionally uneven, and now being reshaped by AI — but the core mechanic Tobin and Amazon pioneered, get paid for sending someone a customer, hasn't really changed at all.
Market-size and growth figures above come from multiple 2026 industry reports, which vary by methodology (channel spend vs. platform/software market) — treat exact numbers as estimates rather than a single settled figure.
Read More: This article also appears at ScienceAffiliate.com.
