The least glamorous verb in advertising may be “serve.” Someone makes the commercial. Someone buys a spot. Someone else, quietly, makes sure the right file appears on a Roku in Queens, a phone in Phoenix, and a browser in Birmingham - then records what happened without counting the same household three times. That quiet someone is the business Innovid built.
Innovid is enterprise software for creating, delivering, measuring, and optimizing ads across connected TV, linear television, digital, and social channels. Its users are not viewers. They are global brands, agencies, streaming publishers, and the people who traffic campaigns while the rest of us hunt for the mute button. Before it went private, the company named Anheuser-Busch InBev, CVS, Kellogg’s, Mercedes-Benz, Target, and Sanofi among its customers. It also worked with Disney, Hulu, ESPN, NBCUniversal, Paramount, Roku, and the major agency holding groups.
That customer list explains the appeal. A large advertiser may run one campaign through dozens of publishers, formats, screens, identity systems, and reporting conventions. Native platform dashboards can each report their own version of success. Innovid offers another view: deliver the ad from an independent system, collect comparable logs, measure reach and frequency across the mess, and adjust creative or spending while the campaign is still alive.
The product is a loop, not a dashboard
The first pivotThe clickable idea that became plumbing
Zvika Netter, Tal Chalozin, and Zack Zigdon founded Innovid in Israel in 2007, around the GarageGeeks hackerspace. The name was a mash-up of innovation and video. Their first sharp idea was interactive video: put objects, buttons, and other behavior inside an ad instead of treating video like a tiny television spot. Innovid filed an early patent in 2008 and collaborated as web-video standards such as VAST and VPAID took shape.
The first thing to fail was not the technology. It was the assumption that interactivity alone would be a large, durable category. A clever ad unit is a feature; every campaign still needs trafficking, device compatibility, reporting, and evidence that it worked. Meanwhile, television was leaking into internet-connected devices, each with its own rules. Innovid followed the recurring headache. In 2012 it built one of the first video ad servers designed for connected TV. By 2015 its technology was included in Roku’s software kit. The company moved from decorating the ad to operating the road it traveled on.
“An irrelevant impression is just as bad as an unviewable one.”Grant Parker, Innovid president
That distinction became its moat. Innovid did not buy media. Because it was not selling the inventory it measured, it could argue that it had fewer reasons to grade generously. It built direct publisher integrations, pursued Media Rating Council accreditations, and kept logs at the impression level. Google was the obvious ad-server rival; Jivox and Celtra attacked creative; iSpot, VideoAmp, Nielsen, and Comscore measured television; publisher tools remained convenient. Innovid’s pitch was the combination - and the neutrality.
How the meter runsA usage business riding a secular shift
Legacy Innovid made most of its money by charging a fixed price per volume of ads served. More impressions meant more revenue. Minimum commitments made some of that demand predictable. Measurement software and creative services added adjacent dollars. It is an elegant model when television budgets migrate into streaming: Innovid could grow inside an existing account without winning the account again.
The same model exposes the weak spots. Advertising is seasonal and cyclical. A major customer represented 16 percent of 2023 revenue. Cloud and data costs rise as measurement expands. Closed platforms can restrict the signals an independent company needs. And if an advertiser runs a simple campaign inside one ecosystem, the native tool may be good enough. Innovid works best when complexity is genuine: many publishers, many creative versions, significant impression volume, and a team able to act on fresh data.
Measurement made the platform stickier. In 2022, Innovid paid about $160 million for TVSquared, using roughly $100 million in cash and the balance in stock. TVSquared brought linear-TV attribution into Innovid’s streaming-heavy world. It was expensive, and integration added hosting costs and operational risk. But it changed the product from “we know where the digital ad went” to “we can compare what happened across television.” Panera later used those insights to shift publisher budgets in flight, reporting a 5 percent gain in incremental reach and a 4 percent reduction in cost per unique reach. Audi used creative automation to expand from two or three models per market to 12 across 67 regions.
The valuation resetFrom $1.3 billion to a certain $500 million
Innovid entered public markets through a SPAC merger in late 2021 at an implied $1.3 billion valuation. The timing was merciless. Growth stocks repriced, the SPAC glow vanished, and Innovid’s shares closed below $1 for the first time in April 2023. The business itself kept moving: 2023 revenue rose 10 percent to $139.9 million, though the company recorded a $31.9 million net loss. By the third quarter of 2024 it had reached $38.3 million in quarterly revenue and $4.7 million in net income. The stock still traded as if patience had expired.
The public-market haircut
Different valuation measures and capital structures, same blunt lesson: the strategic exit landed far below the SPAC-era headline.
Mediaocean first approached with an indicated $1.70 per share in 2023. Negotiations stretched, financing wobbled, and Innovid briefly cut off talks in October 2024 after the buyer lowered its price and lacked a full equity backstop. What changed the board’s mind was not romance. It was math and certainty. The final $3.15 offer carried a 94 percent premium to the last unaffected close of $1.62. The board judged that price to be the best reasonably available and chose cash over hoping the public market would rediscover the company.
Mediaocean closed the acquisition in February 2025 at roughly $500 million in enterprise value and $525 million in equity value. It merged Innovid with Flashtalking, then kept the Innovid name for the combined ad-tech business. The fit was straightforward: Innovid was strong in CTV delivery and measurement; Flashtalking brought broader creative personalization, social, and omnichannel execution; Mediaocean connected both to planning, buying, billing, and verification.
The new wagerCan AI operate the control room?
In June 2026, Mediaocean launched NIVO AI inside Innovid. The product language is theatrical - NIVO “thinks,” Orchestrator “connects,” and specialized agents “act” - but the jobs are concrete. Agents can generate and score creative, translate spreadsheets into campaign taxonomies, traffic placements, run quality checks, answer reporting questions, and recommend or execute optimizations. Company pilots with brands and agencies reported workflow-efficiency gains of up to 90 percent compared with manual campaign setup.
This is a more credible place for agentic software than a blank chat box. Innovid already sits next to the campaign, the logs, the creative, and the measurement. It can potentially close the loop between seeing a problem and fixing it. But the conditions matter. An agent cannot repair missing data rights, contradictory naming conventions, a publisher that will not expose signals, or an approval process that still ends in a weekly meeting. Enterprise customers also need auditability, privacy controls, and humans governing expensive decisions. Automation magnifies clean operations; it also accelerates dirty ones.
Lead with a visible, specific capability. Use it to enter the recurring workflow underneath.
Delivery creates data. Data creates measurement. Measurement creates the right to optimize.
Usage pricing rides customer growth, while minimum commitments soften the seasonality.
Do not sell infrastructure complexity to a customer whose native dashboard already solves the job.
What to copyOwn the unphotogenic layer
The copyable lesson is not “start an ad-tech company.” It is to notice when the bright feature sits on top of an ugly, repeated operation. Innovid’s interactive ads opened doors. Ad serving made the relationship recurring. Impression logs enabled measurement. Measurement made optimization possible. Each step reused the same integrations and data rather than wandering into an unrelated market.
The playbook also requires patience and scale. Direct publisher integrations take years. Accreditation is slow. Global advertisers demand service, security, and reliability. A usage meter becomes attractive only after meaningful volume arrives. The approach will not work when customers are small, switching costs are low, inputs are inaccessible, or the supposedly neutral platform depends on one gatekeeper. It also fails if measurement arrives after the team can still change the outcome.
Innovid’s exit did not vindicate its SPAC price. It did something more useful: it admitted that CTV infrastructure alone needed broader creative reach, media workflow, and capital behind it. The company that began by making video ads clickable now wants machines to launch and tune them. Viewers may never know. For infrastructure, obscurity is often the point.