Ecommerce / Retail · Deep dive

Mirakl

The enterprise marketplace operating system — SaaS that lets Best Buy, Macy's, Kroger and Airbus run their own third-party marketplaces, dropship programs and retail media, now repositioning as the neutral infrastructure for AI-agent commerce.

emerging

The question that decides it: Does agent-led shopping route through the branded enterprise marketplaces Mirakl powers — the bet behind Mirakl Nexus and the December 2025 Stripe partnership — or does AI-mediated discovery collapse purchasing into Amazon, Google and ChatGPT rails, leaving a take rate already compressed from 4.5% (2015) to ~2% (2024) shrinking faster than new marketplace launches and Ads/Payout attach can offset, ahead of a rumored 2026-27 Euronext IPO priced off the 2021 $3.5B mark?

My take ★★ Very interesting ★ Interesting − Maybe ✗ Not interesting

Snapshot

Mirakl sells the software that lets large retailers, distributors and manufacturers run what Amazon runs: a third-party marketplace, a dropship program, a retail-media business — without building any of it. Founded in Paris in 2012 and co-headquartered in Boston, it powers 450+ enterprise platforms across 40 countries (company, March 2025) — Best Buy, Macy’s, Kroger, Lowe’s, Tesco, Airbus Helicopters, HPE — that moved roughly $14.6-15B of GMV in 2025, up 31% year over year (Sacra). It raised ~$948M of equity, last priced at $3.5B in Silver Lake’s September 2021 Series E, hit $218M ARR with its first full year of group-wide profitability in 2025, and is the subject of persistent Euronext Paris IPO chatter for 2026-27. It matters now because it has repositioned itself, via Mirakl Nexus and a December 2025 Stripe partnership, as the neutral infrastructure for AI-agent commerce — a bet that will either re-rate the company or expose it.

Founding story

Philippe Corrot and Adrien Nussenbaum met in 2004 and in 2006 founded SplitGames, a French marketplace for video games. Fnac, the French electronics and media retailer, bought it in November 2008 — then handed the founders the keys to Fnac’s own third-party marketplace, which they built and ran for three years. That sequence is the company: they are among the few people who had operated both an independent marketplace and one grafted onto a large incumbent retailer, and they watched firsthand how much custom machinery — seller onboarding, catalog mapping, order routing, payout compliance — a retailer needs to sell other people’s products. In 2012 they left and productized it as Mirakl, betting that Amazon’s 3P model would become table stakes for every large retailer and B2B distributor. The structure has stayed unusual: two co-CEOs, one in Paris (Corrot, product and Europe), one in Boston (Nussenbaum, US growth), for over a decade — a genuine dual-HQ company rather than a French firm with a US sales office.

How it works

Mirakl is orchestration middleware that sits behind a client’s existing storefront; shoppers never see it. When Best Buy’s marketplace (relaunched on Mirakl in August 2025) sells a third-party espresso machine, the buyer checks out on bestbuy.com as normal. Mirakl’s platform routes the order to the correct seller, tracks acceptance, shipment and delivery SLAs, splits the payment — seller’s share versus operator’s commission — and handles the regulated money movement to the seller via Mirakl Payout. Upstream, the operator recruits sellers through Mirakl Connect, a network of 13,000+ pre-vetted sellers with pre-built connectors, or invites its own; sellers self-onboard through wizards that map their product feeds (API, CSV, or connectors from Shopify and others) into the operator’s taxonomy, with AI-assisted catalog matching and enrichment cleaning the data before listing. The same rails run in two modes — marketplace (seller of record is the third party) or dropship (retailer is seller of record, supplier ships) — switchable per seller. On top sits Mirakl Ads, sponsored-product retail media sold against marketplace traffic. Since 4-6 months of integration into ERP, OMS and storefront systems is required (Sacra), the platform is genuinely sticky once live: ripping it out means re-onboarding thousands of sellers.

Product and business overview

The suite has five named layers. Marketplace Platform, the core: seller lifecycle, offers, orders, quality scores, commissions. Dropship, the same machinery with retailer-as-merchant-of-record — the wedge into conservative retailers that want assortment without third-party branding. Mirakl Connect, the cross-operator seller network — Mirakl’s closest thing to a network effect, since each new operator makes the network more valuable to sellers and vice versa. Mirakl Ads, AI-driven retail media launched with the December 2024 acquisition of Adspert’s ad-bidding technology, plus Payout for seller payments and Catalog for product data (Target2Sell, a personalization vendor, was acquired in April 2022). The newest layer, Mirakl Nexus (launched with 2025 results), packages agentic-commerce infrastructure: Agentic Activation (April 2026) restructures product data so AI agents can find and transact it — Mirakl claims fewer than 1% of product pages are LLM-ready — and Business Agents (June 2026, GA in H2 2026) put conversational AI advisors inside Mirakl Ads. The Stripe partnership (December 2025) supplies the payment rails for agent-initiated checkout.

Business model and pricing

Revenue is booked as SaaS subscription plus usage. There is no public rate card; every deal is quoted. Reconstructed figures: a minimum platform license around $90K/year, typical enterprise licenses of roughly €180K-325K/year, $500K+ for custom-SLA deployments, plus a GMV-linked component estimated near 2%, and $100K-500K+ of implementation on top (Mercur, Spree and VirtoCommerce analyses, 2025-26). Sacra puts typical first-year cost around $493K. The structurally important number: Mirakl’s effective take rate has fallen from ~4.5% in 2015 to ~2.0% in 2024 (Sacra) as the company deliberately shifted weight from transaction fees toward enterprise subscription — good for revenue predictability, but it means GMV growth no longer converts to revenue at the old rate, which is exactly why 2024 GMV grew 30% while ARR grew only 15%. Ads, Payout and professional services are the margin recovery plan: monetize the same GMV twice.

Traction over time

Headcount tells the other half: a hiring sprint toward ~1,000 people in 2021-22 (including a plan for 350 engineers), then cuts — Glassdoor accounts describe layoffs two weeks before Christmas 2022 after two sales-org overhauls that year — down to ~891 (ZoomInfo, June 2026). Growth reaccelerated in 2025 after the 2024 dip, which is the single most IPO-relevant fact in the table.

Market analysis

Marketintelo sizes marketplace-platform software at $12.4B in 2025, growing ~11.8% annually to $28.7B by 2034 — but the real argument is penetration: marketplaces took 67% of global ecommerce in 2024 (Mirakl-cited data), while only ~3% of enterprises operate one (Sacra), and B2B commerce — Mirakl’s second act, with clients like Airbus Helicopters — dwarfs B2C in transaction volume. Adjacent expansion multiplies the wedge: retail media (a $100B+ global opportunity Sacra flagged), marketplace payments, and now agentic commerce. The structural headwinds are equally real: retail-media dollars concentrate with the biggest traffic owners, mid-market marketplace software is commoditizing, and if AI shopping agents centralize discovery, the case for operating your own marketplace — the premise underneath every Mirakl contract — weakens.

Competitive intel

See the competitor table for detail; the shape of the field: Rithum (~$106M revenue est. 2025; 18,000+ customers; $50B GMV touched) is the biggest direct threat in US dropship, with a supplier network far larger than Mirakl’s Connect but a post-merger platform reviewers call expensive; Mirakl wins on full marketplace operation and international reach. VTEX (NYSE, ~$1B market cap in 2025) bundles marketplace into the commerce platform itself and wins where the client wants one vendor. Marketplacer undercuts on price in mid-market and B2B. Shopify/Adobe/Salesforce bundling caps Mirakl’s down-market expansion, even as Adobe and Salesforce partner with it at the top. Systems integrators still build $3-12M custom marketplaces for the very largest operators. And Amazon is the structural competitor: every prospective operator can instead just become an Amazon seller — cheaper, faster, and increasingly where AI agents will default. Mirakl’s moat is enterprise depth, switching costs, and the operator-side network in Connect; its exposure is everything below the enterprise tier and everything after the transaction moves to an agent.

History and evolution

• 2006-08 — SplitGames founded; acquired by Fnac (Nov 2008); founders run Fnac’s marketplace ~3 years.
• 2012 — Mirakl founded in Paris; early adopters Darty, Fnac, Carrefour.
• Jul 2015 — $20M Series B (83North, Felix); 55 customers.
• Feb 2019 — $70M Series C (Bain Capital Ventures); US push accelerates.
• Sep 2020 — $300M Series D (Permira) at $1.5B — unicorn during the COVID ecommerce surge.
• Sep 2021 — $555M Series E (Silver Lake) at $3.5B, ~33x ARR.
• 2022 — Target2Sell acquired (Apr); Payout launched; two sales reorgs and pre-Christmas layoffs as the post-COVID correction bites.
• Aug 2023 — ~$110M non-dilutive credit line (BNP Paribas, HSBC, J.P. Morgan) — runway without a down round.
• Q4 2023 — Platform reaches profitability; ARR growth slows to 20% (2023) then 15% (2024).
• Dec 2024 — Adspert acquired; Mirakl Ads rebuilt as AI-native retail media.
• Aug 2025 — Best Buy relaunches its marketplace on Mirakl — nine years after shutting its first attempt in 2016.
• Dec 2025 — Stripe partnership for agentic commerce; “Santa Quits,” a fully AI-generated first-ever ad campaign.
• 2025-Jun 2026 — Mirakl Nexus launched; Agentic Activation (Apr 2026); Business Agents for Ads (Jun 2026); Euronext IPO rumors for 2026-27.

What people say

The case for. G2 reviewers consistently praise fast seller onboarding, platform reliability at peak volume (Mirakl claims 100% uptime through peak periods), and a support team that behaves like a partner (G2, accessed 2026). Gartner Peer Insights ratings are strong, and the customer list is its own testimonial: Best Buy re-entering marketplaces chose Mirakl over building; Macy’s, Kroger, Lowe’s and Tesco run on it. Glassdoor sits at a solid 4.2/5 across 205 reviews, with culture and career scores above 4.0.

The complaints. Customers: implementation is long and expensive — G2 reviewers cite a learning-curve over a month, “unbundled” APIs, documentation lacking clarity, and complexity that grows in B2B deployments; third-party analyses put three-year total cost above $1M for a mid-size operator. Trade commentary (McFadyen Digital and others) has long noted that most enterprise marketplaces fail on seller recruitment and unit economics, not software — a failure mode Mirakl gets paid through but cannot fix. Employees: the negative Glassdoor tail is sharp — accounts of a toxic US sales culture, “revolving door” leadership, zero reps hitting quota in 2022 after two reorgs, layoffs delivered by HR rather than managers, and a lavish company trip to southern France weeks before December 2022 cuts. Structurally, skeptics note ARR grew 15% in 2024 against 30% GMV growth — take-rate compression in plain sight — and that the 2021 33x-ARR mark still overhangs any IPO.

Outlook: the open question

The bull case requires two things to be true by the IPO window: monetization per dollar of GMV stops falling, and agentic commerce strengthens rather than bypasses operator-owned marketplaces. Concretely: Ads, Payout and Nexus attach must lift effective revenue per GMV dollar back above the ~2% floor (watch whether ARR growth ≥ GMV growth in 2026, reversing 2024); the 2025 reacceleration to 23% with group-wide profitability must hold for a 2026-27 Euronext listing to clear the 2021 $3.5B mark rather than reprice below it; and early agent-driven transactions must demonstrably route through Mirakl-powered storefronts via the Stripe rails, proving Nexus is infrastructure and not marketing. If all three land, Mirakl is the neutral commerce layer for 450+ enterprises at software-margin profitability — a scarce, IPO-able asset in European tech.

The bear case is that both engines stall. Take-rate compression continues as operators demand subscription-heavy deals; mid-market bundling (Shopify, VTEX, Marketplacer) caps expansion beyond the ~500 logos; and AI shopping agents — trained on the deepest catalogs and cleanest data — default to Amazon, Google and ChatGPT surfaces, making the branded enterprise marketplace a legacy channel and Nexus a defensive rebrand. The tells: whether 2026 new-customer signings hold near 45-52 per year or shrink; whether any named customer attributes material GMV to agent-initiated purchases by mid-2027; whether the IPO happens, at what mark, and on which exchange; and whether Rithum’s post-PE-turmoil pricing pushes dropship deals Mirakl’s way or into a price war. The company survived the 2022-23 squeeze and bought its way to profitability; the question is whether it owns the next channel or just renamed itself after it.

How a challenger would attack it

Sell the outcome, not the software. Mirakl gets paid through the failure mode it cannot fix: trade commentary has long noted most enterprise marketplaces die on seller recruitment and unit economics, not platform capability — yet Mirakl charges ~$493K in year one and $1M+ over three years before a single seller lists. A challenger flips the contract: success-fee pricing tied to live GMV, seller supply bundled in from day one (Rithum’s 18,000-supplier network shows the asset Mirakl’s 13,000-seller Connect only partially matches), and AI-driven implementation that collapses the 4-6 month, $100K-500K integration Mirakl still requires. G2 complaints hand over the product roadmap — “unbundled” APIs, unclear documentation, month-long learning curves — so an agent-native platform where onboarding, catalog mapping and seller support are done by AI rather than configured by consultants attacks both the price point and the timeline simultaneously. The economics also cooperate: Mirakl’s take rate has already compressed from 4.5% to ~2%, meaning it must defend enterprise subscription revenue precisely when a usage-priced entrant makes subscriptions look like shelfware. And the mid-market Mirakl abandoned as overkill — where Marketplacer and Shopify bundling already win — is the beachhead it structurally cannot reprice to defend without cannibalizing its ~500-logo enterprise book ahead of an IPO priced off a $3.5B 2021 mark.

Same playbook, new buyer

Marketplace rails for buyers Mirakl’s price excludes and geographies it deprioritizes. The most promising shift is deep B2B verticalization: Mirakl serves Airbus Helicopters and HPE with the same horizontal platform it sells Macy’s, but industrial distribution — parts, chemicals, MRO — needs vertical-specific machinery (contract pricing, punch-out catalogs, compliance docs) that a horizontal operator bolts on rather than builds around. A vertical B2B marketplace OS with domain data models wins deals Mirakl quotes at €180K-325K plus heavy customization, and Mirakl won’t follow because verticalizing fragments the single codebase its software margins depend on. Second: sell the same orchestration to marketplace sellers rather than operators — the 13,000+ Connect sellers each juggling multiple Mirakl-powered storefronts have no neutral layer of their own, and Mirakl can’t serve them aggressively without alienating its paying operators. Third, geography: Mirakl’s dual HQ anchors it to the US and Western Europe while VTEX shows Latin America rewards a regional-first platform; Southeast Asia and the Gulf have enterprise retailers with no local equivalent, and a pre-IPO Mirakl optimizing for profitability will not fund a loss-making regional land grab.

Sources and further reading

• Mirakl revenue, funding & growth rate (Sacra, accessed Jul 2026)
• Mirakl at $218M ARR (Sacra, 2026)
• Mirakl reached $218M ARR (+23%) and launched Mirakl Nexus (Mirakl, Feb 2026)
• Mirakl notches $11.2 billion in 2024 annual GMV (Digital Commerce 360, Mar 13, 2025)
• Mirakl Announces $555 Million Series E Funding (PR Newswire, Sep 21, 2021)
• Mirakl Closes $20M Series B From VCs That Love Online Marketplaces (TechCrunch, Jul 22, 2015)
• Mirakl Reaches Profitability on Its Platform Business, 20% ARR Growth (Mirakl, Feb 2024)
• Mirakl Pricing in 2026: What It Actually Costs (Mercur, 2026)
• Best Buy’s New Third-Party Marketplace is Now Live (Retail TouchPoints, Aug 2025)
• Mirakl debuts AI-generated ad campaign as it stakes out agentic commerce (Digital Commerce 360, Dec 12, 2025)
• Mirakl Reviews (Glassdoor, accessed Jul 2026)
• Marketplace Platform Software Market Research Report (Marketintelo, 2025)