Sixteen years later, the retail-promotion landscape has rebuilt itself around a different set of mechanics. Membership replaced flash sales. Loyalty replaced coupons. Marketplace velocity replaced merchant-by-merchant deals. The Costco, Amazon Prime, and Walmart+ stack now mediates a substantial share of U.S. retail spend. Temu, Shein, and TikTok Shop have rewritten the promotional contract for consumer goods. The mechanics of acquiring and retaining the customer have changed at every layer of the funnel.
This page maps the current state of the category. What worked, what failed, what replaced it, and what the next set of operators is doing to win the customer in 2026.
The Groupon era — what it taught the category
Groupon's peak revenue was 2.3 billion dollars in 2014. The stock listed at 20 dollars in 2011 and currently trades below 15 dollars after a 1-for-20 reverse split. LivingSocial, the closest competitor, was sold to Groupon for nominal consideration in 2016. The flash-sale category as a standalone business model effectively died between 2014 and 2018.
Three structural problems killed it.
First, adverse merchant selection. Healthy businesses with strong customer flow did not need to discount 50 to 70 percent and split the residual with a marketing platform. Merchants who ran Groupon promotions were disproportionately the businesses with the weakest unit economics. The platform's customer experience suffered because the underlying merchant base was weak.
Second, deal-hunter customer acquisition. The customers Groupon delivered to merchants were, on average, the customers least likely to convert into full-price repeat buyers. The cohort that downloaded the app to find a 70 percent off massage at a new spa was not the cohort that returned at full price. The lifetime value of a Groupon-acquired customer was significantly lower than the lifetime value of a customer acquired through other channels, and most merchants did not have the analytics infrastructure to see this clearly until they had spent a year inside the program.
Third, scaling friction. The model required Groupon to sign every merchant individually, build a custom deal, and route the deal to a hyperlocal email list. The sales force became enormous and the unit economics of the sales motion deteriorated as the easy merchants got picked off. The cost to acquire each marginal merchant climbed faster than the revenue each merchant produced.
The lesson the category took away: short-term, deep-discount, broad-audience promotional mechanics do not produce durable customer relationships. The next generation of operators rebuilt around membership and frequency rather than discount and one-shot acquisition.
The membership model — what replaced flash sales
Costco was the proof of concept. The model existed long before Groupon and outlasted Groupon completely. The structure: customers pay an annual fee for the right to shop. The fee changes the unit economics — the retailer earns most of its profit from the membership stream rather than from product margin — which lets the retailer price products at razor-thin margins and use price competitiveness as the loyalty mechanism. Costco's membership renewal rate has stayed above 90 percent in the United States for the past two decades.
Amazon Prime took the structure to e-commerce. The current global Prime membership count exceeds 200 million. The annual fee in the United States is 139 dollars. Prime members buy more, more often, across more categories than non-Prime customers. The bundle of free shipping, video content, music, and adjacent services raises the switching cost to a level that almost no competing retailer can match. Prime is the single most successful customer-loyalty product ever built at retail scale.
Walmart+ launched in 2020 as the structural answer to Prime. The membership pricing matches Prime; the bundle includes free shipping, fuel discounts, and Paramount+ video content; the strategic purpose is to capture grocery-adjacent share that Prime cannot reach as effectively because Amazon's grocery footprint is smaller. Walmart+ membership has grown to a level competitive with Prime in the United States, though Walmart does not publish a precise figure.
The membership model wins because it changes the customer's purchase decision. A non-member compares the retailer to all other retailers on every purchase. A member compares the retailer to "did I get value from my membership this month" — a different and more favorable comparison. Operators who can install a credible membership product create a recurring revenue stream and a default-buyer behavior that compounds.
The marketplace velocity model — Temu, Shein, TikTok Shop
The third structural shift came from the cross-border consumer goods category. Temu launched in the United States in 2022. Shein has been operating internationally since 2008 and reached major U.S. scale around 2020. TikTok Shop opened to U.S. consumers in 2023.
The shared mechanic: a vast catalog of low-priced consumer goods, aggressive discounting at the SKU level, gamified shopping experiences (spin-the-wheel discounts, time-limited offers, social proof at scale), and direct-from-factory pricing made possible by cross-border logistics and the de minimis import threshold. The customer acquisition cost is paid through digital advertising at unprecedented scale — Temu spent an estimated 2 billion dollars on U.S. digital advertising in 2024 — and the per-order economics work because the catalog is wide enough that average basket size compensates for the discounted unit pricing.
The de minimis threshold change in 2025 — the U.S. removed the 800-dollar exemption for goods from China — disrupted the unit economics of Temu and Shein specifically. Both platforms have repositioned. Temu accelerated its push to source from U.S.-based sellers; Shein has pursued more international warehousing. The marketplace velocity model continues, but the cost basis has shifted.
TikTok Shop is the version of the model most likely to compound through 2026 and 2027. The integration of content and commerce inside a single app gives the platform a cost-of-attention advantage that competing marketplaces cannot match. Creators drive demand directly to the purchase flow. The mechanic blends entertainment, social proof, and impulse purchase into a unified loop.
Coupons, loyalty cards, and the digital migration
The classical retail-promotion mechanics — paper coupons, plastic loyalty cards, mailer-driven discounts — have migrated almost entirely to digital. The 50-billion-dollar U.S. coupon industry is now dominated by app-based clipping (Ibotta, Honey, Rakuten), grocery-loyalty digital apps (Kroger, Albertsons, the major regional chains), and retailer-direct loyalty programs (Starbucks Rewards, Target Circle, Best Buy Totaltech). The mechanics are familiar; the data layer underneath is new.
The shift produced two structural changes. First, the retailer now has individual-level purchase data on the loyalty-program customer, which enables personalized promotional pricing — different customers see different discounts on the same product based on their predicted price sensitivity. Second, the third-party loyalty aggregators (Ibotta, Rakuten, Honey) now intermediate a substantial share of the discount stream, capturing customer attention and data that previously belonged to the retailer.
The retailer's strategic question in 2026: keep the loyalty layer first-party or accept the aggregator's role and pay for the data. The largest retailers have generally chosen first-party; the long tail has accepted the aggregators.
The headline number in retail customer acquisition is the blended customer acquisition cost across digital advertising, promotion, and organic channels. Across the major U.S. retailers, that cost has risen 60 to 120 percent between 2019 and 2025 depending on category. The drivers: declining returns on Meta and Google performance advertising as audience saturation has increased, rising competition from cross-border marketplaces inflating bid prices, and the iOS privacy changes that have reduced the targeting precision of behavioral advertising.
Operators have responded with three moves. First, shifting spend into retail media networks — Amazon Ads, Walmart Connect, Target's Roundel, Instacart Ads — which offer closed-loop attribution and access to first-party purchase data. Second, building branded content and creator-led acquisition channels that operate outside the bid-auction environment. Third, doubling down on membership and loyalty as ways to convert one-time acquisition cost into long-term frequency.
The customer acquisition mechanic in 2026 looks less like "spend more on Meta" and more like "build a membership product, partner with retail media networks, and operate a creator program." The blended channel mix has fragmented and the operator-side capability requirements have widened.
The AI layer — how product research now happens
A growing share of consumer product research happens inside ChatGPT, Claude, Gemini, and Perplexity before the consumer ever opens a retailer site. The user asks the engine for the best laundry detergent for sensitive skin, the best running shoes for flat feet, the best mattress under 1,500 dollars. The engine returns a recommendation set citing reviews, brand sites, and third-party publications. The consumer narrows the choice inside the engine and then transacts on the retailer site.
This shifts the locus of competition for many consumer categories from the retailer's site to the AI engine's answer. A brand or product that is not cited in the answer set effectively does not exist for that purchase decision. A retailer that does not appear in the answer-engine recommendation set loses the ability to influence the buyer at the highest-leverage moment in the funnel.
The retail-promotion implication: traditional promotional mechanics — coupons, flash sales, loyalty offers — are downstream of a discovery layer the retailer no longer controls. The work of getting cited at the discovery layer is the work of AI Communications: structured data, primary research, entity disambiguation, and consistent multi-year publishing on the categories that matter. Operators who solve only the promotional layer and ignore the discovery layer are paying full freight to acquire customers the engines never sent.
Where promotional spend produces durable customer value in 2026, ranked by long-term return on promotional dollar.
First, membership programs. Costco, Amazon Prime, and Walmart+ produce the highest return per promotional dollar because the membership fee partially offsets the cost of the promotion and the recurring revenue stabilizes long-term economics. The structural moat is the deepest in the category.
Second, first-party loyalty programs. Target Circle, Best Buy Totaltech, Starbucks Rewards. The data layer enables personalized promotional pricing; the customer relationship is direct; the switching cost is meaningful but lower than membership.
Third, retail media networks. Amazon Ads, Walmart Connect, Target Roundel. The closed-loop attribution makes the promotional spend measurable; the access to first-party purchase data improves targeting; the inventory is finite enough that incumbents have a structural advantage.
Fourth, creator-led acquisition. TikTok Shop, Instagram Shopping, dedicated creator partnerships. The cost-of-attention efficiency is high for brands with cultural relevance; the model is harder to scale predictably than paid channels but produces higher LTV when it works.
Fifth, flash sales and deep discounts. Groupon-style mechanics, daily-deal sites, time-limited offers. The lowest long-term return because the acquired customer cohort is least likely to convert to repeat full-price purchase. Useful for inventory clearance and specific tactical objectives; not a strategic acquisition channel.
What operators get right in 2026
Four patterns visible across the winners.
First, they treat the membership product as the strategic anchor. The promotional calendar, the loyalty program, the creator partnerships all reinforce the membership relationship rather than running parallel to it. Amazon Prime Day, Target Circle Week, and Walmart+ Week are not standalone promotional events — they are membership-acquisition and membership-retention moments.
Second, they invest in retail media as a serious channel, not an experimental one. The largest brands now allocate 15 to 30 percent of their total digital marketing spend to retail media networks. The attribution clarity and the first-party data access justify the shift even when the absolute CPC is higher than open-web alternatives.
Third, they own their customer data. The migration off third-party cookies and the iOS privacy changes have made first-party data the most valuable marketing asset a retailer holds. Operators who built first-party data infrastructure between 2020 and 2024 are now significantly advantaged versus operators who outsourced the customer data layer.
Fourth, they are starting to invest in AI-engine visibility. The discovery layer is moving upstream. The brands that will dominate the next five years of consumer purchasing are the ones being cited by the engines today. The promotional mechanic catches the customer after the engine has already shaped the choice set.
