Roku Pitches Data Quality Over Quantity to Win CTV Performance Ad Dollars

Full analysis

Roku showed up at Cannes Lions 2026 with a pitch that sounds almost humble: connected TV advertising needs fewer, better data signals. Quality over quantity. Roku VP Sal Candela's argument is that sitting at the TV operating system layer, the software that runs the whole television and watches behavior across every app on the device, gives Roku a cleaner view than any single streaming app can have. The proof points are real products: a June 2026 deal with Smartly to buy CTV at social speed, an April 2026 bundle called Roku Curate that stitches Roku data to retail purchase signals from Best Buy Ads, Criteo, Instacart, and Kroger, and a SimpliSafe case from an expanded iSpot partnership showing a 23% lift in leads and 31% in website visits.

What's actually being decided: not "is Roku's data good." It's whether the TV OS layer becomes a data toll booth that everyone licenses, or just another commodity signal source in a crowded pool. Hard to undo? For Roku, medium. These are packaging and partnership moves, easy to rejigger. For agencies deciding whether to route CTV buys through Roku's product or through a demand-side platform (the software agencies use to buy ads), the sunk-cost pull builds over a couple of upfront cycles. What sets the deadline: the Q3 2026 upfront conversations, where this either becomes a CPM story or doesn't.

The Council

The Market Analyst. Roku is playing defense here. Squeezed from two sides: The Trade Desk pushing agencies toward buying CTV direct through its platform, and Amazon building a first-party signal advantage on Fire TV that Roku can't match on retail data. Plugging into iSpot and Criteo, tools agencies already trust, is smart because it avoids asking anyone to adopt a new currency. But it's also the problem. Integrations that work because they're familiar become table stakes fast, and table stakes don't hold a premium price. Roku's "number one TV OS" share number anchors the whole valuation story. OS share and monetizable signal quality are not the same number, and Roku wants you to treat them as one. For the non-specialist: Roku is trying to sell its position as the software running the TV, before that position turns into a commodity everyone undercuts.

The Skeptic. One home-security brand with an easy-to-measure "click buy" action is not proof that CTV works for brand advertisers or long purchase cycles. SimpliSafe is the friendliest possible test case. "Quality over quantity" is also exactly what every vendor says when their data footprint is smaller than the competition's. Roku only sees viewing on Roku devices, and that audience skews older and budget-conscious. The Smartly deal moves inventory faster, it doesn't make the data smarter. And here's the packaging tell: Best Buy, Criteo, Instacart, and Kroger all keep their own direct data relationships. Roku Curate bundles signals Roku doesn't own exclusively. Anyone can assemble a similar basket. For the non-specialist: Roku is reselling a mix of data that its partners are free to sell to everyone else too.

The Operator. Trading desks have a real decision inside 90 days. Roku Curate collapses three separate vendor conversations, CTV inventory, retail signal, and closed-loop measurement, into one line item. That changes how you structure an upfront commitment, and one line item is harder to unbundle and negotiate down than three. The more immediate pressure is Smartly. If performance buyers trained on social can spin up CTV at social speed, the programmatic specialists inside agencies lose their gatekeeping leverage. Watch Criteo's managed-service team feel this first. Roku just turned Criteo's retail signal into a feature inside Roku's product instead of a standalone thing Criteo sells you directly. For the non-specialist: Roku is turning three shopping trips into one, which is convenient for buyers and awkward for the vendors being bundled.

The Customer / End User (the agency buyer). The buyer wants two things and Roku is offering one clearly. Lower-funnel proof, yes, they've begged for that from TV for a decade. But buyers also want leverage, and a bundle removes leverage. The 23% lead lift is attractive right up until you ask whether you could get it cheaper by buying Roku inventory, Criteo signal, and iSpot measurement separately and squeezing each. The question every buyer should ask: am I paying a convenience premium for a basket I could assemble myself? If Roku Curate carries a CPM floor that doesn't move, the answer is yes, and sophisticated desks will test the unbundled route within a cycle or two.

The Tensions

Is the OS-layer position a moat or a feature? The Strategist read (not voiced above, but live in the room) says Roku can be the identity spine of the living room the way Apple is for mobile. The Skeptic says streaming apps route around the OS layer whenever they want, and the data isn't exclusive anyway. This is the whole ballgame. Apple's mobile control was enforced by owning the hardware and the store rules. Roku's control over app-level signal is much softer.

Does bundling raise price or invite arbitrage? The Operator says one line item is stickier and protects the CPM. The Customer says one line item is exactly what a smart buyer disassembles to find the markup. Both are right for different buyers. The question is which kind of buyer dominates the next two upfronts.

Is closed-loop retail measurement Roku's to win? Everyone agrees closed-loop, tying an ad to an actual purchase, is where CTV dollars are heading. Nobody in this council thinks Roku sets that benchmark. Amazon does, because Amazon owns the purchase data outright instead of licensing it from four partners.

Synthesis

This hinges on one belief: whether Roku's view from the OS layer is genuinely hard to replicate, or just a convenient bundle of signals its own partners sell elsewhere. The evidence in this story leans toward bundle. Criteo, Best Buy, Instacart, and Kroger all keep their data relationships. Smartly is distribution. iSpot is measurement Roku doesn't own. The one thing that is genuinely Roku's, the cross-app viewing data on Roku devices, is real but capped to a skewed audience.

The council leans skeptical on the moat and positive on the near-term commercial move. Roku Curate is good execution that lifts the Q3 upfront CPM conversation. It is not a structural lock. Before committing budget, an agency should test the unbundled route on one campaign: buy the pieces separately, compare the closed-loop result and the effective price against Roku Curate. That test settles the convenience-premium question fast.

What Roku needs is for the retail-signal bundle to become standardized enough that buyers stop pricing the components separately. That's a race against Amazon, whose Fire TV plus owned purchase data does the same closed loop without licensing anyone.

The Prediction

Prediction: By the close of the 2027 upfront negotiations (roughly June 2027), Amazon Ads, not Roku, will be the CTV player agencies cite most often as the benchmark for closed-loop retail measurement, and Roku Curate's per-partner retail signals (Criteo, Best Buy, Instacart, Kroger) will remain available to buy directly from those same partners outside Roku's bundle.

Confidence: Medium — Amazon owns the purchase data outright; Roku licenses it from partners who keep their own direct sales channels open.

Why: The closed loop that ties a TV ad to a real purchase is only as strong as your grip on the purchase data, and Roku's grip is a license from four partners who each keep their own direct sales channels open. Amazon owns its shopper data outright and pairs it with Fire TV viewing, so it can run the same measurement without asking anyone's permission or paying a data toll. When agencies decide whose closed-loop number to trust as the standard, they follow whoever controls both ends of the loop, which is Amazon. The opposite outcome, Roku becoming the cited benchmark, would require its licensed basket to out-credential a rival that owns the actual receipts, and licensed data rarely beats owned data on trust.

Revisit by 2027-06-30: We're right if trade coverage and agency commentary through the 2027 upfronts treat Amazon as the closed-loop CTV measurement standard while Roku Curate's partner signals stay independently purchasable from Criteo, Best Buy, Instacart, or Kroger. We're wrong if Roku Curate becomes an exclusive or clearly preferred path to those retail signals, or if agencies broadly name Roku as the closed-loop benchmark over Amazon.

The interesting wrinkle: Roku's best defense isn't out-measuring Amazon, it's being the neutral, interoperable option agencies reach for precisely because they don't want to feed Amazon more leverage. Candela's "interoperable platform" language is doing exactly that work. Neutral and licensed can still win share. It just won't set the benchmark.