General Marketing
The Ratings Are Getting Synthetic. Your Sales Aren’t.
Nielsen plans to roll out new streaming measurement in early 2027 that fuses platforms' own data with its panel. MediaPost says the method behind it is literally called synthetic. Great for counting a huge CTV audience. Not a substitute for counting your sales.
Nielsen just announced what might be the biggest change to how it measures TV in decades, and almost nobody noticed.
There was no big launch. No splashy keynote. Just a quiet update on its press site saying new "content measurement solutions" are coming in early 2027.
According to Nielsen, the new products pair streaming platforms' own first-party data with Nielsen's national panels and third-party big data. The goal, in its words, is to measure "the fragmented content ecosystem that sample-based estimates often miss." And eventually, these products will feed key Nielsen reports, including future versions of The Gauge.
On Saturday, MediaPost's Joe Mandese, who has covered Nielsen for nearly half a century, put a name on what he thinks is underneath it.
The name is STAR. Synthetic Total Audience Ratings.
Yes. Synthetic.
What "Synthetic" Actually Means Here
Let's be fair about this, because the word sounds scarier than it is.
It doesn't mean AI-generated fake viewers. Per Mandese's reporting, Nielsen's own executives described it to researchers in March as closer to "fusion": matching people in one database with similar people in another so you can combine what you know about both.
Here's how it works, as Mandese describes it.
Nielsen has a panel of about 40,000 households. Streaming platforms have server logs covering millions of people and devices. STAR splits that panel upwards of 30 times to match it against the platforms' data. Nielsen's executives put the adjustment weight at a magnitude of about 1,000x.
"We are creating a synthetic panel," a Nielsen data science executive told the room.
The method has already been used outside the U.S., in Denmark. Nielsen had planned to have the Media Rating Council audit it this past summer, then pulled it at the last minute. Mandese expects it back in the accreditation process in 2027.
Here's the Part That's Actually Good News
Why would Nielsen do this at all?
Because CTV got too big for the old way.
Mandese writes that viewing now happens across a "virtually atomized" marketplace of streaming services and connected TV devices that a panel-based system "cannot possibly keep up with."
Read that again. The single most famous TV measurement company in the world is rebuilding its method because the streaming audience outgrew the panel.
That's not a problem for CTV. That's a scoreboard for CTV.
The audience is there. It's huge, it's spread out, and it's watching on the biggest screen in the house. The measurement industry is scrambling to catch up with it, not the other way around.
Here's the Asterisk
Every rating is a model. That was true before STAR, and it will be true after.
What's changing is how much of the number is modeled, and who supplies the raw data.
Mandese calls these "bespoke ratings," because they're built partly on each streaming provider's own data. He also points out a side effect: the industry may soon be comparing "more apples to oranges" than ever. One source told him advertisers are already comparing broadcast NFL audiences directly against Amazon's integrated numbers, and that traditional NFL networks may be giving up roughly a 15% audience deficit by not doing the same kind of data integration.
That's one source's estimate, not a settled fact. But the point stands. When two networks count their viewers two different ways, the bigger number doesn't automatically mean the bigger audience.
We wrote last week about how the networks keep fighting over ratings currency. This is the next round of that fight. And it's a fight about how media gets bought and sold.
It isn't a fight about whether your ads worked.
Ratings Are for Buying. Sales Are for Proof.
This is the distinction that gets lost every time the measurement headlines heat up.
An audience rating answers one question: how many people probably saw this content? That's useful. It helps set prices. It helps you plan reach.
But it was never built to answer the question your CFO actually asks: did this sell anything?
IAB's senior director of measurement, Matt Shapo, said it plainly in a Beet.TV interview this week: "First and foremost, know that you need to choose an outcome before you design a measurement."
He also said marketers are increasingly combining attribution, mix models and incrementality testing, to figure out whether marketing caused a business outcome or "merely happened to be standing nearby when it occurred."
That's exactly right. And notice what's missing from that list. Nobody serious is saying the audience rating is the outcome.
If you judge your CTV campaign on a modeled audience estimate, you're grading your business on somebody else's math. If you judge it on a last-click report, you're grading it on the wrong moment. Either way, the CFO is going to have questions.
What to Do About It
You don't need to become a ratings methodologist. You need a measurement plan that doesn't care which method wins.
Pick the outcome first. Sales, new customers, store visits, booked appointments. Write it down before the campaign launches, like Shapo says.
Use your own sales data as the anchor. Your order system doesn't fuse, weight or model anything. It knows what sold, when, and where it shipped.
Read results by market. Compare sales where your ads run against each area's own baseline. If demand moves where the TV ad runs, you'll see it in your own numbers.
Keep holdouts. A few areas where you don't advertise is the cleanest way to prove what the ad actually caused, and to measure the alpha CTV creates instead of the sales that would have happened anyway.
Triangulate. One instrument always lies a little. Line up several: your sales by market, branded search, direct traffic, a "how did you hear about us?" question at checkout.
Treat audience numbers as buying tools. Use them to plan reach and judge price. Don't use them to decide whether the channel works.
The Final Take
Nielsen is going synthetic because the streaming audience got too big and too spread out for a panel of 40,000 homes to keep up.
That's a compliment to CTV.
The ratings will keep evolving. They'll get more modeled, more bespoke, and probably more argued over. Let them. Those fights are about how media gets priced.
Your proof lives somewhere else. In your orders. In your markets. In the customers who walked in after your ad ran on their TV.
Ratings can be synthetic. Your sales can't.
That's how we measure every campaign at CS & Co. It's built into the CS Method™. Want to know whether CTV can move your actual sales? Find out if CTV is a fit.
Cory Poccia
CEO, CS & Co. Marketing Studio™