Ecommerce / Retail · Deep dive

Rokt

The AI company that turned the ecommerce 'thank you' page into a multi-billion-dollar ad network — profitable, cash-generative, and circling a 2026 IPO.

emerging

The question that decides it: Can Rokt keep taking a ~50% cut of the confirmation-page whitespace once Amazon, Walmart, and Shopify can serve their own post-transaction offers with first-party data they already own — or does the Transaction Moment get insourced the way onsite search ads were?

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

Snapshot

Rokt is an Australian-born, New York-headquartered AI company that monetizes the most valuable and most wasted slice of ecommerce: the Transaction Moment — the checkout-to-confirmation window when a shopper has their wallet out. Instead of a static “Thank You” page, Rokt’s ML engine serves a relevant offer from another advertiser in that whitespace, splits the revenue with the retailer, and keeps roughly half the ad spend. It is a genuine two-sided network — retailers wanting incremental revenue, advertisers wanting high-intent customers — and, unusually, profitable and cash-generative. Reported revenue for the year ended August 2025 was ~US$743M (+48%) on ~US$100M EBITDA, with the business circling a 2026 US IPO at secondary marks well past its US$3.5B January 2025 mark.

Founding story

Bruce Buchanan did not come from adtech. He spent six years at Boston Consulting Group working with airlines, then ran Jetstar as Group CEO for five years, expanding the low-cost carrier across 16 countries. The formative lesson was commercial: airlines make much of their margin on ancillaries — seats, bags, insurance, hotels — in the booking flow, at the moment a customer has committed and is entering payment. Buchanan looked at ecommerce and saw the opposite: the highest-intent moment in the funnel, the confirmation page, was being thrown away on a receipt and a “Thank You.”

He left Jetstar in 2012 and founded Rokt (initially “Rocketer”) in Sydney with co-founders Justin Viles and Ben Voltz. The thesis was narrow and durable: the Transaction Moment is the most under-monetized real estate in commerce, and the way to unlock it is more relevance, not more ads — the right offer to the right buyer, or nothing at all. That “or nothing at all” discipline became the product’s spine and its defense against the “spammy confirmation page” critique. Rokt proved the model in Australia, then moved its center of gravity to New York, where the US checkout volume and advertiser budgets live.

How it works

The mechanics are concrete. A shopper buys something on Retailer A’s site — a concert ticket, a meal kit, a flight. At confirmation, Rokt’s SDK renders a placement in whitespace that would otherwise be blank. In milliseconds, the Rokt Brain — its proprietary ML engine, said to be trained on billions of transactions a year — evaluates who the buyer is, what they just did, and which offer in the advertiser pool is most likely to convert them: a HelloFresh trial, a streaming subscription, a cashback card. If no offer clears the relevance bar, Rokt shows nothing. The retailer keeps brand safety, the shopper isn’t spammed, and Rokt wins only when it matches intent.

Two sides make it a network. On the demand side are transaction-heavy platforms — Live Nation, Ticketmaster, AMC, Macy’s, PayPal, Uber, Lyft, Wayfair, Domino’s — who plug Rokt into checkout and earn incremental high-margin revenue. On the other side are advertisers: Rokt runs a closed marketplace of hundreds of vetted brands across retail, travel, media, financial services and food delivery, paying to reach a buyer at the moment of proven intent. Rokt owns the relevance layer and shares economics with the retailer.

Product and business overview

Rokt sells a suite organized around the checkout journey. Rokt Ads / Rokt Thanks is the flagship — the confirmation-page offer engine that made the company. Rokt Ecommerce extends the logic earlier in the funnel: cart optimization (Upcart), payment monetization (Pay+), product discovery (Catalog), and post-purchase upsell (via the 2024 AfterSell acquisition). mParticle, acquired for US$300M (January 2025), adds a real-time customer-data platform to act on richer first-party signals — a hedge as third-party cookies erode.

Strategically, Rokt wants to be the default infrastructure for the Transaction Moment across two customer types: enterprise retailers wanting a turnkey checkout media network, and the long tail of Shopify merchants reached via AfterSell. The mParticle deal and late-2025 board build-out read as pre-IPO scaffolding.

Business model and pricing

Rokt is a performance/CPA network, not an impression business. It gets paid when a shopper takes an action — a converting click, a sign-up, a purchase — on confirmed transaction data. The advertiser pays for the acquired customer; Rokt shares that revenue with the retailer and keeps the spread. Third-party estimates peg Rokt’s take at ~50% of network ad spend, the retailer receiving the other half as near-pure-margin revenue on a transaction it had already completed.

That structure is the appeal and the vulnerability. For the retailer it is free money on whitespace that cost nothing; for the advertiser, efficient acquisition at the highest-intent moment. For Rokt, a ~50% take on a fast-growing performance network is why it throws off cash. But a 50% cut is exactly the number a large retailer eyes once its volume justifies staffing the function in-house. Rokt’s counter is the Brain — ML trained across the whole network should beat any single retailer’s attempt, the aggregator’s argument right up until the largest customers leave.

Traction over time

The scale numbers are the story. Rokt reported it was on track to power 7.5B+ transactions in 2025, with 33,000+ active clients against ~3,000 ecommerce partners. Annualized revenue reportedly crossed US$900M by October 2025, up from a US$600M ARR exit in 2024. Crucially it reached profitability — ~US$100M EBITDA for the year ended August 2025 — making it an IPO candidate rather than a cash-burning growth story. Headcount clusters around 700, and Rokt has granted staff roughly a quarter of the company over its life, which is why its secondaries are framed as employee-liquidity events ahead of a listing.

Market analysis

Rokt rides the fastest-growing line in the digital ad budget. eMarketer put US retail-media ad spend at ~US$58.8B in 2025 and projected ~US$69B in 2026; globally the category was ~US$140B in 2024 and near US$165B by 2026. Retail media now approaches ~18% of US digital ad spend, a top-three channel with search and social. The driver: retailers have first-party data and attention, advertisers have lost third-party cookies, and the confirmation page is a rare place where intent is confirmed, not inferred.

The catch is concentration. eMarketer estimates Amazon and Walmart will capture ~89% of net-new US retail-media spend in 2026 — they own the on-site search and sponsored-listing budgets. Rokt’s wedge is the part they ignored: the post-transaction, non-endemic offer (a HelloFresh ad on a ticketing site) rather than the endemic sponsored product. A defensible niche — but downstream of a market whose economics are set by two players with every incentive to eventually run the Transaction Moment themselves.

Competitive intel

Competition comes from three directions. The retail-media giants — Amazon Ads ($50B+ ad revenue), Walmart Connect, Instacart’s Carrot Ads — define the category and command the budgets but mostly monetize on-site search and listings; the wave Rokt surfs and the platforms most able to insource its niche. Commerce-media incumbents — Criteo (~$2B revenue), Fluent, Wunderkind, Attentive — chase the same enterprise budgets with broad reach but without Rokt’s checkout-specific ML and closed marketplace. The direct niche — RevLifter, Wildfire, Button — attacks the same post-purchase whitespace with intelligent offers and cashback at a fraction of Rokt’s scale.

Where Rokt wins: 13 years building a relevance engine and vetted advertiser network purpose-built for the confirmation page, on a model aligned with retailer revenue. Where it is exposed: no moat stops a Walmart or Shopify from serving its own post-transaction offers with data it already owns — exactly how onsite search advertising migrated from adtech into Amazon’s and Walmart’s walls.

History and evolution

• 2012 — Buchanan leaves Jetstar; founds Rokt (as “Rocketer”) in Sydney with Justin Viles and Ben Voltz.
• 2013-2019 — Builds the product in Australia; raises Series A ($8M, 2013), B ($34.5M, 2017), C ($70M, 2019); expands into the US and Europe.
• Oct 2020 — $112M Series D at a $630M valuation; New York becomes the operational center of gravity.
• Dec 2021 — $325M Series E led by Tiger Global at a $1.95B valuation; joins the unicorn class with Wellington, Whale Rock, and Square Peg aboard.
• Dec 2022 — Secondary led by Square Peg and Wellington lifts the mark to $2.4B less than a year after the Series E — an early sign the story was liquidity, not fresh capital.
• 2024 — Acquires AfterSell (Shopify post-purchase upsell app); exits the year at ~$600M ARR, growing ~43%.
• Jan 2025 — $335M secondary tender at a US$3.5B valuation (led by Tiger Global and Square Peg); ~$122M went to employees and early seed investors. Simultaneously announces the ~US$300M acquisition of CDP mParticle.
• 2025 — Reaches profitability (~$100M EBITDA, year ended August), ~$743M revenue, ~$900M annualized by October; adds Datadog and ex-Wise finance leaders to the board for IPO readiness.
• Late 2025 / 2026 — Reported secondary trades (via MA Financial) mark the company well above $3.5B, with figures from ~$7-8B up to a reported ~A$12B circulating and unconfirmed; a dual ASX/Nasdaq listing is floated for 2026.

What people say

The case for. The financials are the strongest testimony: Rokt is the rare late-stage company that grows ~45-48% and makes money, which is why secondary demand has repeatedly outrun its willingness to sell and it can pick its IPO window rather than need one. Merchant sentiment in trade coverage is positive — post-purchase monetization framed as “win-win” incremental revenue on a page that cost nothing — and Rokt appears in Gartner’s commerce-media coverage and on best-workplace lists in New York. The AfterSell and mParticle deals read as moves to build the data and distribution needed to defend the niche.

The complaints. Two clusters, both real. First, the culture. Glassdoor reviews (357+, ~3.4/5 on work-life balance) describe an intense, high-pressure, high-comp environment reviewers call explicitly “not for everyone” — Buchanan has reportedly likened employees to Navy SEALs, and reviews cite shifting and retroactive sales quotas, leadership-through-fear in the go-to-market org, and heavy senior attrition. Comp rates highly (~90th percentile) but burnout is described as normalized; Capital Brief’s own headline quoted the “culture that’s not for everyone” line. Second, the consumer critique: post-transaction offers can feel like spam or dark patterns, and retail media as a whole draws scrutiny (FTC dark-pattern enforcement, the EU DSA’s ban on manipulative interfaces). Rokt says it prohibits dark patterns, requires a clear decline option, and shows nothing when relevance is low — but its economics depend on shoppers who just paid engaging with one more offer, and skeptics ask whether “relevance” serves the buyer or the network.

Outlook: the open question

For the bull case, Rokt must prove the Transaction Moment is a durable specialist market — that its network-wide ML converts better than any retailer’s in-house attempt, and that retailers would rather share ~50% than staff the function themselves. For the bear case, nothing new has to happen: the same insourcing that pulled onsite search ads inside Amazon and Walmart reaches the confirmation page. Everything else about Rokt is unusually clean: fast growth, real profit, an experienced operator, a defensible wedge in advertising’s fastest-growing line.

The tension sits in the take rate. A ~50% cut is spectacular precisely because it is a large number — and large numbers on someone else’s checkout attract the customers with the volume to replace you. mParticle is the tell: Rokt is buying the first-party data layer that makes its relevance hard to replicate and ties clients in. If that works, Rokt IPOs into a category tailwind as the profitable specialist and the secondary marks look cheap. If Shopify, Walmart, and the largest retailers decide the Transaction Moment is theirs, Rokt keeps the long tail while its best logos build in-house, and the multiple compresses. The 2026 listing will price which future the market believes.

How a challenger would attack it

Undercut the 50% take. Rokt’s economics are its exposed flank: it keeps roughly half of network ad spend on real estate the retailer owns. A challenger would attack as the retailer’s agent, not another network — sell the confirmation-page engine as white-label software at a 10-20% take or a flat SaaS fee, let the retailer keep the advertiser relationships and 80%+ of the revenue, and pitch it explicitly as “your Transaction Moment, not Rokt’s.” Instacart’s Carrot Ads has already proven retailers will buy someone else’s ad stack to run as their own; the same template applied to post-purchase whitespace turns Rokt’s aggregator argument inside out. The demand side is attackable too: Rokt runs a closed marketplace of hundreds of vetted brands, so an open, self-serve advertiser platform — programmatic access to confirmation-page inventory the way trade desks buy retail media — offers advertisers price discovery Rokt’s opacity denies them. And a challenger would weaponize the consumer critique: with FTC dark-pattern enforcement and the DSA circling, a “consumer-safe checkout offers” positioning — fewer, clearly labeled, genuinely declinable — gives risk-averse enterprise retailers cover to switch. Rokt can’t match the low take without dismantling the ~$100M EBITDA story it is about to IPO on.

Same playbook, new buyer

Sell the Transaction Moment where Rokt’s marketplace doesn’t reach. Rokt’s network is built on US consumer checkout — ticketing, delivery, retail — with a closed pool of consumer brands. The cleanest shift is B2B and vertical checkouts: SaaS billing pages, business-travel bookings, wholesale marketplaces, insurance and utility payment flows — high-value confirmed transactions where the relevant cross-sell is a business product Rokt’s consumer advertiser pool can’t serve, and where offer density is zero today. Rokt won’t chase it pre-IPO; its Brain is trained on consumer transactions and its growth story depends on deepening the existing network, not rebuilding the demand side. Second shift: regulated-market geographies. The EU’s DSA-constrained environment punishes Rokt’s engagement-maximizing model but rewards a consent-first, utility-framed version (warranty registration, loyalty enrollment, relevant services) built for European compliance from day one — a brand position the “one more offer” incumbent can’t credibly occupy. Third: Buchanan’s own origin market, airlines and travel ancillaries outside the US, where booking-flow monetization is understood but the confirmation page still runs on legacy ancillary vendors rather than an ML relevance layer.

Sources and further reading

• Sacra — Rokt revenue, funding & business breakdown (Sacra, 2025)
• Rokt — Series E led by Tiger Global ($325M, $1.95B) (Rokt, December 2021)
• Rokt — Secondary transaction to $3.5B valuation, Anita Sands to board (Rokt, January 2025)
• Bloomberg — Rokt’s valuation hits $3.5 billion after secondary share sale (Bloomberg, January 2025)
• AdExchanger — Rokt acquires mParticle for $300 million (AdExchanger, January 2025)
• Capital Brief — ‘It’s a culture that’s not for everyone’: Rokt pulls off $510m secondary (Capital Brief, 2025)
• The Aussie Corporate — Rokt nears $12.1B valuation ahead of IPO (Pick & Scroll News, November 2025)
• eMarketer — Retail media ad spending forecast, H1 2026 (eMarketer, 2026)
• Glassdoor — Rokt employee reviews (Glassdoor, accessed July 2026)
• Forbes Australia — Rokt-star: Bruce Buchanan has the Midas touch (Forbes Australia, 2023)