A cookware company has an obvious place to advertise: a cooking show. The audience is already thinking about dinner. The product belongs in the picture. Yet the interesting part of Made In’s television experiment began when it wandered away from that comfortable logic, into sports and news. A good customer, it turned out, did not have to be watching someone sauté.
Made In started with a $75,000 test, according to Tatari’s customer account. Its digital growth was flattening. Television offered another audience, but also an awkward question: how would a company accustomed to watching online performance know whether a commercial worked? Tatari’s answer was to make the purchase, the response, and the next buying decision part of the same process.
- Plan, buy, and measure streaming and linear TV together.
- Test creative and audiences before committing more budget.
- Judge new business separately from conversions that would happen anyway.
02A familiar screen. A different question.
Tatari sells software and services for planning, buying, measuring, and improving advertising across streaming television, conventional broadcast and cable, and online video. Brands can operate the platform themselves or buy help. Agencies can license it for their clients. It joins media buying with analytics.
Philip Inghelbrecht, a Shazam co-founder and former TrueCar president, started Tatari with Mike Swinson in 2016. Swinson had led data science at TrueCar. Their backgrounds joined consumer technology with statistical modeling.
The company describes an engineering and data science-heavy team. A buyer must choose networks, schedules, prices, creative, and audiences, then work out which combination deserves another dollar. A spreadsheet can record those choices. It cannot make the uncertainty disappear.
03First, change the commercial
Made In’s experience contains a small rebuke to the idea that better targeting fixes everything. Tatari reports that replacing a basic direct-to-consumer-style video with a cinematic, chef-led commercial more than halved cost per visitor. The original creative was the thing to improve. Buying more exposure for it would have multiplied the expense.
Then came the less obvious audience. Discounted sports and news placements offered reach outside food programming. In its 2023 presentation with Made In, Tatari described direct Hulu deals with CPMs 66 percent below the programmatic alternative. Those opportunities depended on timing, inventory, and a brand willing to test.
The transferable move is to separate an intuitive story about the customer from evidence about response. Test the commercial. Test the context. Keep the acquisition target visible. A cookware buyer may also watch football; a media plan built entirely around cooking never gets to discover that.
“TV used to feel like a black box.”
Chip Malt / Made In co-founder
04Buying a result is harder than counting a view
Television lacks the trail of a search ad click. Someone sees a commercial, makes tea, searches on a phone, and orders tomorrow. Tatari uses different methods for different screens. For linear television, it models changes in traffic against an expected baseline. Streaming supplies impression-level signals that can help connect an exposed household with later activity.
The distinction that matters is attribution versus incrementality. Attribution connects exposure with a subsequent action. Incrementality asks how much activity happened because of the advertising. Tatari offers modeled incremental lift alongside view-through approaches. A modeled baseline is not a randomized holdout experiment; both are ways to investigate the same difficult counterfactual.
The conditions matter. Measurement needs correctly implemented event collection and enough historical or response data to distinguish a campaign from ordinary business. Shared IP addresses can produce false matches. Retail sales require different inputs from website purchases. For a marketer, the practical discipline is to choose the success measure before spending, then compare methods when their answers diverge.
05The expensive bit between buyer and screen
Tatari’s other wager concerns the route the money takes. Programmatic advertising can involve buying platforms, selling platforms, exchanges, and their fees. Tatari combines programmatic access with direct publisher execution. Its argument is that premium television does not always need the same chain of intermediaries used to trade a sprawling internet of display inventory.
The company’s own infrastructure took a detour. TheViewPoint, acquired in 2022, provided supply-side technology. In explaining its evolution, Tatari said the commoditized exchange business offered limited room for innovation or monetization. Meanwhile, direct publisher relationships were creating a different bottleneck: insertion orders, trafficking, and billing still required too much manual work.
The response became Upstream, now a sister company under Infra.tv. Its February 2026 rebrand marked the shift toward automating direct sales through publisher ad servers. Integrations include NBCUniversal, Paramount, Tubi, and Warner Bros. Discovery, with Disney reached through its technology stack. The lesson is refreshingly operational: inspect the paperwork around the transaction before inventing another marketplace.
06What the experiment actually costs
Tatari earns platform and service fees around media buying; the commercial arrangement depends on the offering. One published comparison lists self-serve streaming retargeting from $350 per week and prospecting from $5,000 per week. Those are entry points for those campaign types, not an all-in production budget. Its pages describe other arrangements, so a buyer needs a campaign-specific proposal.
The alternatives include MNTN for performance television, broader programmatic platforms such as The Trade Desk, and buying from publishers directly. Tatari’s case rests on combining linear and streaming, measurement, and direct execution. A team wanting one small streaming campaign may value simplicity differently. A brand without measurable response data cannot purchase certainty by adding a dashboard.
07A media buyer with a longer memory
The AI Planning Engine, introduced in July 2024, turns budgets, performance targets, and historical results into suggested buys. Humans retain control. In August 2026, Tatari reported a four-week randomized comparison in which unmodified AI plans matched or exceeded human-adjusted plans in 17 of 18 valid cases. That company-run test has a defined sample; it cannot establish universal performance.
Customers now extend beyond early direct-to-consumer adopters: Chime, Calm, Tecovas, and agency partners appear among its published examples. In its October 2026 anniversary announcement, Tatari said more than 400 brands and agencies used the platform, and that revenue doubled between 2023 and 2025 without increased headcount. It also reported annual profitability since founding and no venture-capital rounds.
For the advertiser, the useful idea remains smaller than that corporate story. Give television a question it can answer. Buy enough exposure to learn something. Change the creative when it disappoints. Expand when evidence supports it. The cooking channel will still be there. The next customer might be watching the game.