Curious Streaming, Smarter TV

Lisa Ciancarelli
Jul 28
7 min read

Quark Insights and Infinitum — Quark Insights: Better Results from Curiosity

How Precision Metrics & Quality Turn Campaigns Into Living Systems

If you work anywhere in TV advertising, the bar has changed. Reach alone doesn't impress clients anymore. They want proof that a campaign did something. This article lays out a simple way to think about leveraging streaming and linear TV together, so you can defend your results with more confidence, drive trust and increase future budgets.

Whether you are a vet in the trenches planning campaigns or looking for new inspiration, the framework shared in this article can be easily mastered and replicated: use streaming to learn, use quality to know what to scale, and use plain language to make sure people actually understand what you found.

The Big Idea: Start Small & Scale

Streaming and connected TV are where you learn; linear TV is where you repeat what you've learned at scale.

In earlier Quark Insights writing, streaming is described as the brain of TV advertising. That brain tests different messages, channels, regions and time windows, and then watches what people do afterward. Linear TV is the body: the part of the system that carries those proven patterns across broader audiences.

For everyday work, this model leads to a simple approach:

Use streaming to ask questions and run experiments.

Use linear to extend and amplify your strategy.

It's a small mental shift, but it changes how you plan and how you defend your strategies.

Why This Matters Right Now

Streaming and CTV are not the side project anymore. Industry forecasts show CTV ad spending climbing steadily through 2029, with CTV expected to pull in more ad dollars than linear TV by 2027. Combined, streaming and linear are projected to make up close to a fifth of total media ad spending in 2025. Advertisers increasingly plan the two together as one converged strategy rather than treating them like separate worlds.

Measurement hasn't fully kept pace with that shift, though. A lot of teams still lean on old habits: reach, cost per thousand views, basic demographics. Those numbers tell you who technically saw an ad. They don't tell you what happened next.

More teams now want to know whether someone visited a website after seeing a streamed ad. Did they start an application? Request a quote? Actually buy something? This is where measurement companies like Infinitum have built their approach, tying streamed ad exposure directly to outcomes like site visits and sales under a promise they call "Absolute. Streaming. Measurement." It's a shift from counting eyeballs to counting actions, and once you start counting actions, curiosity becomes a lot more useful than it used to be.

Make "Curious" a Habit.

People love to say a good analyst is "curious," like it's some rare gift. It's really just a set of habits you can build. When you treat streaming as the brain, you naturally start asking better questions:

Which version of a creative drove more visits or signups?
Which streaming apps or channels performed better than others?
Which regions or times of day gave you more value for the same money?

The advantage of streaming is that deterministic measurement, meaning data tied directly and reliably to specific outcomes, lets you answer these questions while the campaign is still running. You're not just writing a report after the fact. You're shaping what happens for the rest of the flight.

Make Quality a Priority

Here's where a lot of analyses stop short. Most curiosity in streaming gets pointed at audiences and creative alone, which is useful but incomplete. A CIMM (Coalition for Innovative Media Measurement) paper by Erez Levin and Gabriel Dorosz introduces a concept called Media Quality that fills the gap.

Media Quality isn't about who saw the ad. It's about the conditions surrounding the ad itself. Was it large on screen or barely visible? Was the sound on? Did it run during a calm viewing moment or get buried in a stack of a dozen other spots? What kind of show was playing when it appeared?

The paper (see link to CIMM post at the end, a must read) defines quality as an observable driver of effectiveness in two terms:

Placement Prominence: how much attention the ad placement can reasonably earn in terms of time on screen, player size, audibility affecting opportunity for the ad to be viewed & processed.
Contextual Receptiveness: how open the viewer is likely to be in that specific moment, based on time of day, device, geography, and content environment.

In simple terms: was the ad set up to succeed, or just present? That's the question Media Quality is really asking, and it ties directly to campaign performance.

These aren't just opinions or vibes. According to the authors, prominence and receptiveness can be measured, and they help predict how effective an impression actually turns out to be. Streaming and CTV make these conditions easier to observe than older TV formats do, because you can slice data by time of day, region, channel, and creative. Quality stops being an abstract idea and becomes something you can actually track.

Put the brain and body model together with Media Quality, and your planning logic looks like this:

Use streaming to measure outcomes and watch the quality conditions around each impression.
Use curiosity to spot where certain conditions clearly outperform others.
Use linear to extend those high-quality patterns to a much bigger audience.

Simple on paper. But it's a very different approach than treating every impression as interchangeable, which is what a lot of legacy TV planning still quietly assumes.

Putting It Into Practice

Theory is easy to nod along to. The real test is whether a framework holds up when you're working on an actual campaign, with actual numbers attached to actual outcomes. That's where Infinitum comes in, not as a case study you read about once and forget, but as a working example of what happens when you actually build measurement around this brain and body idea instead of just talking about it.

Infinitum's approach is useful here precisely because it doesn't stop at delivery. It ties streamed impressions to what viewers did next, and it does that across the same conditions we've been discussing: region, daypart, creative, channel. So instead of treating Media Quality as a concept you file away after reading a CIMM paper, you get to watch prominence and receptiveness show up in real campaign data, right alongside the outcomes advertisers actually care about.

That's the shift worth soaking in before we go any further. Quality isn't a side conversation you have after the campaign wraps. It's visible while the campaign is still in progress, in the same numbers you're already looking at. You just have to know to look for it with intent. Having all the signals in front of you is what lets you see your data in context. If you've read this blog before, you know I'll always push you toward that. It's how you get ahead of the inevitable "So what?"

So let's walk through what that actually looks like, not as an abstract framework, but as a set of patterns you can expect to see, and language you can use to explain them to the people who control next quarter's budget.

Tactics for Your Own Campaigns

You don't need a big budget or fancy tools to borrow this framework. Here's a simple way to apply it to whatever data you already have access to. Frame streaming as the brain in your story. Tell your team or client, plainly, that you're using streaming to learn which combination of message, channel, and timing works best. Make the goal about learning, not just about reach.

Pick a small set of quality conditions to track. Don't try to measure everything at once. Choose two or three: maybe region, time of day, channel, and creative length. Track outcomes like visits or signups against those conditions. Don't boil the ocean!
Look for real differences, not noise. You're hunting for clear signals, not statistical hair-splitting. Something like this is a signal worth acting on: "evening streaming in these regions beats late night by a wide margin." Keep the story short when you present it, and use a visual if you can.
Suggest one or two ways to scale what worked. Propose shifting more budget toward the streaming patterns that performed well, and mirror those patterns in linear plans where it makes sense. Linear becomes a way to carry proven results to a bigger audience, not a separate guess made from scratch.

Explain it in plain, human terms. Don't just hand someone a table of numbers. Say something like, "when the ad ran in this calmer evening slot, people were more likely to visit the site," or "cutting the number of ads stacked around ours improved response rates." That kind of plain explanation is what actually gets non-technical colleagues on board with your recommendation.

Wrapping It Up

A few ideas are worth carrying forward from all this. Streaming works best when you treat it as the brain of TV advertising, testing and learning, while linear plays the body that scales what the brain figures out. Curiosity isn't some special talent; it's a habit of asking specific questions about creative, channel, region, and time of day, then using streaming measurement to find the answers. Media Quality adds another useful lens, reminding you to pay attention to how and where an ad appears, not just who happened to see it. And real campaign data backs this up, showing that some impressions are genuinely worth more than others depending on region, daypart, creative, and channel.

Put those ideas together and you get something valuable: clearer insights, smarter planning, and a much easier time convincing the people around you that your strategy deserves a shot.

So here's a question worth sitting with. If you picked one upcoming streaming campaign and treated it as a small experiment, testing just one quality condition like time of day or channel, how would you design that test? And once you had an answer, how would you use it to shape a bigger decision down the line?

Ready to put this to work in your own campaigns? If you're staring down a messy dataset or a planning meeting where nobody agrees on what "quality" even means, that's exactly where I come in. I help teams build simple, repeatable structure around exactly this kind of question, from the first hypothesis to the final recommendation.