Affiliate is still worth building before BFCM 2026. If you skip coupon and cashback partners and recruit paid search review sites, publishers, and post-purchase networks, you can measure it. Brands above roughly $5M a year should move now; smaller stores should start with creators and wait.
Coupon and discount sites took 42.4% of US affiliate revenue in the first half of 2025. The partners most likely to bring you a genuinely new customer this Black Friday are the ones who will never apply to your program.
Pull up your affiliate dashboard the Tuesday after Cyber Monday, and it will almost certainly look like a win. Attributed revenue up, conversion rate up, a few partners with numbers that would make any founder smile. The question nobody asks on that Tuesday is how much of it would have happened anyway, because the partner at the top of that list is usually a browser extension that appeared in the last five minutes of a checkout that was already happening.
The channel isn’t small, and it isn’t shrinking. On Cyber Monday 2025, affiliates and partners drove 21.8% of US online revenue according to Adobe, up from 20.3% the year before. More than a fifth of the year’s biggest online shopping day ran through somebody else’s link. The real question is what share of your fifth was new customers and what share was you paying a toll on your own demand.
If you are doing $2M and up, this will help you decide what to launch, cut, and leave for January in the nine weeks before Black Friday. If you are earlier than that, it will tell you honestly why this is not your channel yet.
A modern affiliate program is made up of five partner types that behave like media you only pay for when a sale lands, and coupon sites are the least valuable of them. When most founders hear the word affiliate, they picture coupon codes and cashback portals, and that picture is where the money has been going. eMarketer’s 2026 affiliate breakdown shows cashback and loyalty publishers taking 35% of US affiliate spend in 2024, content publishers just 16%, and discount and coupon publishers capturing 42.4% of affiliate revenue in the first half of 2025, up from 39.7% a year earlier.
Those numbers describe where commission gets paid, not where customers get created. The partner types worth building around look different. Mass media publishers and their commerce newsletters put your product in gift guides, roundups, and reviews with a tracked link. Paid search review sites bid on comparison keywords like “best air purifier” and send shoppers to a top five page before they ever reach your store. Media buyers run paid social, connected TV, or audio campaigns and sell the customers back to you on a cost per acquisition basis. Post-purchase offer networks show your offer on another brand’s thank you page or inside a payment app, reaching a buyer at a moment you could not buy your way into directly. And community partners, from employee benefits platforms to creator groups, put you in front of closed audiences.
Every one of those five reaches someone who was not already on your site. That is the whole test. A coupon extension reaches someone who is already in your checkout, which is why its numbers look spectacular and its incrementality is the weakest in the program.
If you are running $50K to $200K months, do not try to recruit all five. One content partner and one post-purchase network, both measured properly, beats a sprawling program you cannot audit.
Most partners who apply to your affiliate program add nothing or actively cannibalize sales, so the core skill in running one is refusal. Josh Kennedy, founder of the affiliate agency Imagine Marketing, put a number on it that stuck with me: if you simply join a network and accept whoever shows up, he estimates 80% to 90% of the applications will be valueless or value-detracting. The partners who actually create customers mostly never apply. You go find them.
That lines up with a pattern I watched for six years at Shopify, in a different costume. The brands that stalled between $500K and $2M rarely stalled because they were missing a tactic. They stalled because they kept adding apps, channels, and partners faster than they could measure any of them. An affiliate program with 300 approved partners and no audit is the same problem wearing a commission structure. The fix is also the same: fewer things, each one earning its place.
Here is the audit I would run this week, before BFCM traffic makes the numbers impossible to read. Export your top 10 partners by attributed revenue over the last 90 days. For each one, ask a single question: does this partner reach people before they have decided to buy, or after? Loyalty portals, coupon aggregators, and browser extensions sit after. Review sites, publishers, and media buyers sit before. If more than half of your attributed affiliate revenue comes from the after group, you are paying a toll on demand you created elsewhere. The mechanics of how coupon and cashback partners cannibalize checkout and what a pause test reveals have not changed since we first dug into them.
You do not have to remove every coupon partner. Some genuinely save a wavering cart. The only honest way to know which is to pause one for two weeks outside peak season and watch revenue. If nothing moves, you have your answer before Black Friday rather than after.
The most controllable affiliate partners are paid search review sites, and they should be managed like a media plan rather than a list of relationships. A review site in your category is effectively a Google Ads buyer that only gets paid when you do. It bids on research keywords, lands the shopper on a comparison page, and earns commission on the sale. Because you can see which keywords a partner bids on in any keyword research tool, you can also see which searches in your category nobody is covering.
That visibility changes the job. The traditional affiliate manager asks a partner for a better slot on their top five list. The media planner looks at the whole keyword landscape and asks which clusters of demand the brand is invisible for, then funds a partner to go test them. The same agency described one review site partner sending a single client roughly $120,000 a month in affiliate revenue and still scaling. That is one client’s result, not a benchmark, but it shows the ceiling on this partner type is far higher than most founders assume.
The mechanism that makes this work for mid-market brands is the co-funded test. Instead of paying a flat fee and hoping, the brand puts up a small test budget, say $2,500, and the partner matches it. After a few weeks you both know the acquisition cost for that keyword cluster. If it clears your target, the partner keeps it running on commission. If it does not, you have spent a known, capped amount to learn something your own ad account could not tell you.
Subscription brands and anyone with strong repeat economics have the biggest advantage here. The agency’s own client data suggests roughly 20 to 30% of paid search affiliate customers come back to buy again, which means a partner earning a strong commission on a thin first order can still be profitable for you across the lifetime. Treat that as one agency’s directional read rather than a benchmark, and run your own repeat purchase data by partner before you raise a payout. If you do not have repeat data by source yet, that is the thing to fix first.
PR-led affiliate placements acquire customers cheaply, but they arrive in bursts, so they belong inside a program rather than being the program. A mention in a major publication’s gift guide with an affiliate link attached is genuinely valuable. There is no paid media cost baked into the commission, which is why content partners can usually accept lower rates than media buyers who are paying for every click. The problem is shape, not quality. A feature spikes for a few days and fades, and a program built on spikes is a program you cannot forecast.
This matters because a lot of what gets sold as affiliate management today is PR with tracking links added. The placements are real and the relationships are real. What is usually missing is everything else: the paid search partners, the post-purchase inventory, the media buyers, and the measurement discipline that turns a set of wins into a predictable channel. If the agency pitching you talks exclusively about publications they can get you into, ask what they would do with the other 80% of the channel.
The timing reality for this year is blunt. January Digital’s review of Cyber Week 2025 found that premium editorial and gift guide placements got more expensive and needed much earlier lead times than in past years, with partners favoring brands that stayed active all year. Many holiday gift guides are planned well before October. If you are starting from zero today, you are unlikely to land a flagship guide placement this season, and you should not pay a rush premium to try.
What you can do is plant seeds for 2027: build the publisher relationships now, send product in January, and be the brand that shows up all year rather than the one asking for a favor in November.
No single test proves incrementality across every affiliate partner type, so match the proof to how each partner creates the sale. For a newly published article, the evidence is timing: a piece that did not exist last week goes live, and clicks and orders appear from that URL. For a post-purchase network, the evidence is in the sub-IDs or site IDs the network passes back, which tell you exactly which host store showed your offer. For audio and connected TV partners, you are relying on their impression and exposure data, which deserves more skepticism and a longer test window.
For the partners where that evidence is thin, you run a pause or holdout test. Turn the partner off for a defined window, ideally outside peak, and watch whether revenue actually drops. Very few brands do this. Among marketers who say they cannot measure affiliate at the level they want, eMarketer found that 70.6% have not run a geo split or incrementality lift study. That is the single biggest gap between the brands that know what their program is worth and the ones that are guessing.
Pair the platform data with the customer’s own answer. A post-purchase survey that asks where the customer first heard about you will surface review sites, podcasts, and publications that last click attribution never credits, and it costs almost nothing to run. When your survey says a meaningful slice of new customers found you through a review site and your dashboard gives that site zero credit, you have just found money you are underinvesting in.
One rule holds for every partner type: if sales are showing up and you cannot trace where they came from, treat that as a warning sign, not a win. And go one step further than the first order. Most affiliate platforms can track a customer’s second, third, and fourth purchase, which lets you raise payouts for the partners who bring customers who stay and quietly lower them for the ones who bring deal hunters who never return.
Affiliate review content is one of the main sources AI assistants draw on when shoppers ask what to buy, which makes a publisher placement an AI visibility asset as well as a sales channel. The eMarketer analysis above found that almost 70% of the sites ChatGPT cited when discussing one eyewear brand came from affiliate marketing content. When a shopper asks an assistant for the best option in your category, the comparison pages and reviews that affiliates publish are a large part of what the model is reading.
That reframes how you should value your content partners. On a last click dashboard, a review site might look mediocre. In reality, the same article might be shaping the answer a shopper gets from an assistant, sending them to your store a day later through branded search. Adobe measured AI traffic to US retail sites up 670% on Cyber Monday 2025, while noting that the base of users remains modest. Small, fast growing, and high intent is exactly the profile where undercounting a partner’s influence costs you the most.
My take for this quarter: do not chase AI placements as a separate project. The work that earns you a place in AI answers is the same work that earns you a place in a good review, which is a product that holds up to comparison and product data an AI can actually read before you pay for the click. Get that right and the affiliate content about you gets better on its own.
Paid search review partners and post-purchase networks can go live and prove themselves before Black Friday on November 27, while editorial gift guides and a full program build mostly cannot. The stakes are real. Shopify merchants sold a record $14.6 billion over BFCM weekend 2025, up 27% on the year before, and more than 94,900 merchants had their best day ever. But a weekend that big also produces the loudest, least trustworthy attribution data of the year, which is why what you launch now matters more than what you launch in November.
Treat the timelines in that grid as my working estimates from conversations with operators and agencies, not published benchmarks. If you are above $5M a year, this is the window to launch one or two measured tests: a paid search partner on a capped, co-funded budget and a post-purchase network with an offer that makes sense on someone else’s thank you page. Get them live in October so you have clean data before peak distorts it. If you are between $2M and $5M, run the partner audit from earlier in this piece and cut the obvious cannibalizers before Black Friday rather than paying them on your biggest weekend.
Then stop. The mistake I would guard against hardest is signing a new agency or onboarding a batch of partners in the second week of November. Everything I wrote about freezing your stack in the final weeks before BFCM applies here. Any new partner live after early November is a partner you will not be able to evaluate until January.
Brands without a proven conversion rate should skip a full affiliate program this year, because affiliates send traffic where it converts best, and established names convert better than newcomers. This is not a judgment on smaller brands. It is how the partners’ economics work. A review site or media buyer is paying for every click they send you, so they rank brands by how much of that traffic turns into commission. A shopper who recognizes a brand from the shelf converts more readily than one meeting it for the first time, and partners know it.
I interviewed a founder who built a review property on exactly this logic. His site only features products that already rank near the top of Amazon in their category. He is not guessing which brand will win; he is adding fuel to a brand that is already winning, then collecting the commission. That is a smart business, and it also clearly shows how the best partners choose who to work with. The agency I quoted earlier focuses mostly on brands doing roughly $5M and up for the same reason.
If you are under $50K a month, your version of affiliate is creators, not networks. Shopify Collabs lets you find creators and pay them commission from inside your admin, which keeps the setup light and the relationships direct. Start with three to five creators who already talk about your category, pay them fairly, and learn what your conversion rate looks like on warm traffic. That data is what will make serious partners say yes to you later.
Run the eighteen-month test on any affiliate pitch that lands in your inbox between now and Black Friday: will this partner still be sending you new customers in eighteen months, or only this November? Build around the first kind.
It is too late to build a full program, but not too late to launch one or two measured partners. Paid search review sites can go live within days to two weeks, and post-purchase offer networks typically take two to three weeks, so both can produce clean data in October before peak traffic distorts it. Editorial gift guides are the exception: premium placements now book months ahead and cost more than they used to. If you are starting from zero, launch a capped test with one paid search partner, cut obvious coupon cannibalizers, and treat gift guides as a 2027 relationship to build starting in January.
Pause the partner for two weeks outside peak season and watch whether revenue drops; if it does not, the partner was taking credit for sales you already had. Start by exporting your top 10 partners by attributed revenue and sorting them into partners that reach shoppers before they decide (review sites, publishers, media buyers) and partners that appear after the decision (coupon sites, cashback portals, browser extensions). If more than half your affiliate revenue sits in the second group, you likely have a cannibalization problem. A post-purchase survey asking where customers first heard about you adds a second, independent read on the same question.
Commission should follow the partner’s cost to reach the customer, which is why content partners and paid media partners rarely deserve the same rate. Content and PR-led partners have no media cost built in, so they can usually accept lower rates. Paid search and paid social partners are buying every click they send, so they usually need higher rates, a cost per acquisition deal, or a co-funded test budget. There is no universal benchmark here. The right rate depends on your margin and repeat purchase rate, so model lifetime value by partner before you raise any payout.
You can let a few in, but only after you have proven which ones actually save carts rather than intercept them. Coupon and discount publishers captured 42.4% of US affiliate revenue in the first half of 2025, largely because they sit at the end of checkout where last click attribution gives them full credit. Some of them do rescue a hesitant shopper, especially in price sensitive categories. The problem is not their existence; it is paying them full commission without testing. Run a pause test on each one, keep the ones whose removal hurts revenue, and cap or remove the rest before Black Friday.
Affiliate review content is one of the main sources AI assistants draw on when shoppers ask what to buy, so good publisher placements improve your AI visibility as well as your sales. eMarketer found that almost 70% of the sites ChatGPT cited for one eyewear brand were affiliate marketing content. That means a review site that looks average on your last click dashboard may be shaping the answers shoppers get from AI assistants. Check whether your best placements are being cited, value those partners accordingly, and make sure your own product data is clean enough that the reviews written about you are accurate.
