The most revealing number in Infillion’s history is not a click-through rate. It is $22 million - the cash price the company paid in 2023 for the assets of MediaMath, a demand-side platform that had once handled more than $500 million in annual ad spend. MediaMath had filed for bankruptcy after acquisition talks collapsed. Its pipes went dark. Customers scattered. A bidder called Genius Sports offered $20.5 million. Infillion offered $22 million and took home the machinery.
Eight months later, MediaMath was trading again. More than 20 supply platforms had been reconnected, and spend had doubled month over month during the quarter before the April 2024 relaunch. The rescue explains Infillion better than any corporate adjective can. This is a company that shops the ad-tech salvage yard, retrieves durable components and tries to make the rebuilt machine run as one.
Nine companies in a trench coat
Infillion did not begin with MediaMath. Its family tree starts in 2007 with SocialVibe, built by Joe Marchese, Brandon Mills and David Levy to help people direct support toward charities. Advertising became the engine: brands paid for interactions, and some of that value flowed to causes. In 2010, a Bing promotion inside Zynga’s FarmVille collected more than 425,000 Facebook fans in less than a day. The click was not decoration; it was the billable event.
SocialVibe became TrueX in 2013 and sharpened that bargain into “value exchange” advertising. A viewer chooses to watch or interact, often in return for fewer interruptions or access to content. Fox bought TrueX in 2014 in a deal reported around $200 million. It passed to Disney with the Fox acquisition, then Disney sold it in 2020 to Gimbal, a location-intelligence company led by Rob Emrich. Gimbal and TrueX became Infillion in 2022.
That lineage left useful parts all over the garage. TrueX knew whether somebody actively chose an ad. Gimbal knew, with permission and aggregation, where devices went in the physical world. Drawbridge contributed identity resolution. MediaMath supplied enterprise programmatic buying. InStadium brought media inside sports venues. Infillion says it has acquired and rebuilt nine category-specific technologies, representing $750 million in accumulated investment.
What it actually sells
For an advertiser, Infillion can be a media team, a self-service dashboard or infrastructure accessed through APIs. Its MediaMath demand layer buys display, connected TV, online video, audio and digital out-of-home inventory. Its “Brain” optimization system evaluates opportunities and adjusts bids toward a campaign goal. Private marketplaces package proprietary audiences and higher-impact formats for buyers who prefer another DSP.
TrueX is the conspicuous product. Instead of counting an ad merely because pixels appeared on a screen, it invites the viewer to opt in. Infillion says those active views produce five times the brand recall of standard video. Its interactive video products extend the idea to shoppable and explorable units across TV, desktop and mobile. A Porsche execution let viewers configure a car inside the ad. A WHOOP campaign produced a click-through rate the company reported at nine times its industry benchmark.
Gimbal handles the walk after the watch. Marketers can create audiences from opted-in location signals, draw geofences around places and estimate whether exposed devices later visited a shop, resort or stadium. That makes Infillion useful to businesses whose sale does not happen inside a browser. A Casper pop-up campaign, for example, was reported to lift visits by 15.7 percent, with an average visit lasting 47 minutes.
The receipt joins the impression
In February 2026, Infillion bought Catalina’s U.S. business. Financial terms were not disclosed. Catalina grew from the checkout-coupon business into a purchase-intelligence network spanning about 70 retail banners and 130 million U.S. households. Infillion describes the data as covering hundreds of billions of dollars in annual consumer spending. Exact totals differ across current company pages, but the strategic point does not: this is observed shopping behavior, not inferred intent.
Connect that data to MediaMath and an advertiser can build an audience from what people bought, run media, then measure sales at the product level. A bidding model can optimize toward verified purchases rather than clicks. For a cereal brand, the closed loop is clean: find likely buyers, show the ad, look for the box at checkout and adjust. Catalina’s U.S. purchase data is expected to become exclusive to Infillion over time, giving its otherwise reproducible software stack a harder-to-copy input.
A third way between suite and sprawl
The market gives media buyers two familiar bad choices. A closed suite from Google or Amazon is convenient but constrains data, inventory and logic. A collection of independent tools offers choice but produces duplicated fees, transfers and reporting disputes. The Trade Desk, Amazon DSP, Google DV360, Yahoo, Adobe, StackAdapt, Basis, Criteo, Nexxen and Viant all compete for some portion of Infillion’s budget.
Infillion’s proposed third way is “composable”: customers select demand, data, creative or supply components, connect their own systems and leave out the rest. A retailer can white-label the platform. An agency can operate it through a UI. A technical customer can use REST APIs. Its Agent Connector, launched in late 2025 and currently documented as a closed beta, exposes campaign planning, creative, reporting, audiences and inventory through the Model Context Protocol. The AI agent can operate the controls; the customer is supposed to own the data and set the constraints.
The failure came before the bargain
MediaMath is the caution label on Infillion’s plan. A celebrated platform can still fail when its capital structure, commercial momentum and ownership process fail first. Infillion did not buy an intact business; it bought assets after shutdown. The low price came with customer loss, disconnected partners and the unglamorous labor of restoring trust.
What changed Infillion’s mind about that risk was the residue: MediaMath’s 2022 business had been EBITDA-positive, with more than $100 million in net revenue and loyal customers, according to Infillion at the time of purchase. The software had strategic distribution. Rebuilding it was expensive in time and credibility, but still faster than inventing a global DSP from an empty repository.
The part worth copying
Buy capability, not a trophy. Look for a failed capital structure wrapped around durable technology, customer habits and expensive integrations. Restore the pipes fast. Then add a proprietary input - in Infillion’s case, attention, location and purchase data - that makes the recovered product more useful than it was alone.
The recipe is not universal. It depends on acquiring assets cleanly, retaining engineers who understand old systems, winning back counterparties and carrying enough capital through a revenue gap. It also works best when the acquired pieces share customers and data. A pile of unrelated products with incompatible contracts is still a pile. Composability can become a friendly word for integration work pushed onto the buyer.
Who should care
The best fit is an advertiser or agency that wants enterprise programmatic buying plus unusual creative and real-world measurement without surrendering everything to one closed ecosystem. Retail media networks are an especially sharp target. Walmart Connect Mexico uses a white-labeled platform powered by MediaMath, combining Walmart audiences with omnichannel activation and closed-loop reporting. In one public case, its campaign reached return on ad spend as high as 21.3.
That variety also explains how Infillion makes money. It can charge for managed media, platform access, data, measurement and custom enterprise work; private marketplaces provide another route to its inventory and audiences. There is no public menu of rates. That is normal in enterprise ad tech, but it makes an outside cost comparison frustratingly slippery. A buyer has to examine the working-media share, data charges, service layer and any integration expense together. The cheapest bid fee can become the expensive option after three vendors take a toll. Infillion’s economic promise is fewer hops and fewer overlapping contracts, not a universally lower sticker price.
Smaller advertisers wanting transparent swipe-a-card pricing will find little to copy. Infillion does not publish a standard fee sheet; platform, media, data and service agreements are negotiated. Customers also need sufficient volume and clean outcome data for optimization to mean much. An AI connector cannot rescue a vague brief, weak creative or a conversion signal that arrives six weeks late.
Still, the direction is easy to see. Ad buying is moving away from one operator clicking through one dashboard. Retailers want infrastructure wearing their own name. Agencies want their logic to travel between vendors. AI systems need callable tools, permissions and machine-readable feedback. Infillion is positioning itself underneath all three. Its bet is that the winner will not own every strategy. It will own the cleanest way to execute them.