A ship's navigator never trusts one instrument.

The compass drifts. Magnetic north is not true north, and the error grows the farther you sail.

The speed log accumulates mistakes. A half-knot off, multiplied by twelve hours, and you are miles from where the math says you are.

GPS drops signal. A cloud, a canyon, a solar flare, and the screen goes blank at the worst moment.

Every instrument on the bridge, trusted alone, will eventually put the ship on the rocks.

So the navigator does something the rest of us forgot how to do. They take a bearing on three fixed points. Three landmarks. Three stars. Three satellites. They draw three lines.

Where the three lines cross is the truth.

Not one line. Not two. Three.

That is what triangulation means. And it is the only honest way to measure CTV.

The Single Instrument Everyone Still Trusts

Most brands measure CTV with one instrument.

Last-click attribution.

It is the compass that always points to the last thing the customer touched before they bought. Usually Google. Usually direct traffic. Usually the channel that caught the customer on the way in, not the channel that sent them.

It drifts. It drifts badly. It systematically undercounts every channel that does its work before the click — which is every channel that builds demand instead of harvesting it.

CTV does its work three days before the click. The viewer watches the ad on Tuesday. They search your name on Friday. Last-click hands the credit to Friday and pretends Tuesday never happened.

A navigator who trusted a compass that wrong would be relieved of command.

But in marketing, we built entire budgets on it. We pulled CTV dollars because the one instrument we trusted could not see what CTV did.

The instrument was not broken. It was just one instrument. And one instrument lies.

Three Bearings on the Same Truth

There are three ways to measure whether CTV is working. Each one takes a bearing from a different position. Each one sees something the other two cannot. And each one is blind to something the other two see clearly.

The first bearing is closed-loop IP matching. It tells you what happened.

The second bearing is incrementality. It tells you whether it would have happened anyway.

The third bearing is media mix modeling. It tells you how it all fits together.

Three different questions. Three different instruments. One position where they cross.

Let me walk the bridge.

The First Bearing: Closed-Loop IP Matching

This is the instrument up close. The depth sounder. The thing that tells you exactly what is under the hull right now.

Closed-loop IP matching connects the impression on the living room television to the conversion on the household's devices. The smart TV has an IP address. The laptop that placed the order shares that same IP address. The match is deterministic. It is not modeled. It is not estimated. It is observed.

The ad ran on this household's screen. This household converted. Same roof. Same network. Same IP.

It is fast. It is granular. It works at the household level, in near real time, while the campaign is still running. It can tell you which audience segment converted, which creative drove the visit, which day the household acted.

No other instrument is this precise or this quick.

But it has a blind spot, and the blind spot is the whole reason this conversation exists.

IP matching tells you the household saw the ad and then converted. It does not tell you the ad caused the conversion. The household might have converted anyway. The match shows you the path. It does not show you the counterfactual.

It also degrades. Consent banners suppress it. Privacy regulation narrows it. Shared IP addresses, carrier-grade network translation, and the slow migration of the internet's plumbing all chip away at the precision over time.

The depth sounder is essential. It is also not enough. It tells you what is under the hull. It does not tell you where the harbor is.

The Second Bearing: Incrementality

This is the instrument that proves direction. The one that tells you which way is true north — not magnetic north, not the direction that feels right, but the actual causal truth.

Incrementality answers the only question that matters in the end. Would those orders have happened anyway?

You take a market. You turn CTV off. You leave it running everywhere else. You wait six weeks, eight weeks, twelve. Then you compare the dark market against the lit ones.

If orders fell where the ad went dark, the ad was causing them. Not correlated with them. Causing them.

This is the gold standard. It is the bearing that cannot lie, because it is built on a controlled experiment instead of an observed pattern. The 2025 Stella benchmark ran this experiment 225 times across major channels and put CTV at the top — 3.30x incremental return, higher than Meta, higher than Performance Max, while branded search came in at 0.70x. The instrument that proves causation put CTV first and put the channel last-click was crediting nearly last.

But this bearing has a blind spot too.

You cannot take it continuously. You cannot run a holdout on every campaign, in every market, every week. It is episodic by nature. You take the bearing, you get a fix, and then you sail on the strength of that fix until you take the next one.

It tells you which way is true north. It does not tell you your exact position at every moment in between.

For that, you need the third bearing.

The Third Bearing: Media Mix Modeling

This is the instrument from far away. The view from the chartroom, not the deck. It does not see any single household. It sees the whole coastline.

Media mix modeling looks at everything at once. Every channel. Every dollar. CTV and linear and search and social and radio and the billboard on the interstate. It pulls in seasonality, promotions, pricing, weather, the economy. It models how the entire portfolio produces sales over time.

It is the only instrument that sees offline. The only one that captures the long tail — the brand effect that shows up months after the impression. The only one that does not depend on tracking a single user, which means it survives every privacy change coming down the road. There is no cookie to break. No IP to suppress. No consent banner to dodge.

It is durable in a way the other two are not.

But it is coarse. It works in broad strokes. It can tell you CTV contributed this much to the quarter. It cannot tell you which creative, which week, which household. It needs years of data to speak with confidence, and by the time it speaks, the moment to act on a single campaign has often passed.

It tells you how it all fits together. It does not tell you what happened on Tuesday.

Where the Three Lines Cross

Now look at what each instrument cannot do, and watch the pattern.

IP matching is fast and precise but cannot prove causation.

Incrementality proves causation but cannot run continuously.

Media mix modeling runs continuously and survives privacy but cannot see the single campaign.

Each blind spot is covered by another instrument's clearest view.

IP matching cannot prove the ad caused the order — but incrementality can. Incrementality cannot run every week — but IP matching can. Neither one sees offline or the long tail — but the mix model does. The mix model cannot see the single household — but IP matching can.

This is not a coincidence. It is the entire argument.

No single instrument locates you. The compass alone runs you aground. The depth sounder alone leaves you lost at sea. The chart alone cannot tell you what is under the hull right now.

But three bearings, taken together, cross at exactly one point.

That point is the truth about your CTV.

IP matching draws the line of what happened. Incrementality draws the line of whether it would have happened anyway. The mix model draws the line of how it fits the whole. Where those three lines meet is the only place the answer actually lives.

One instrument lies. Three instruments agree. The agreement is the measurement.

Why Brands Sail on One Instrument Anyway

If triangulation is so obviously right, why does almost nobody do it?

Because one instrument is easy and three instruments are work.

Last-click comes free in the dashboard. It is already on. It produces a number by lunchtime. It requires no holdout, no model, no patience. It asks nothing of you except that you believe it.

Triangulation asks for all three. It asks you to run a holdout you have to defend to a CFO. It asks you to build or buy a mix model that takes time to mature. It asks you to stitch IP-matched conversions into the same view as the other two.

It asks you to navigate instead of guess.

Most brands choose the guess. They sail on the compass alone, and they call the heading wherever the compass happens to point. Then they wonder why they keep ending up somewhere other than where they meant to go.

The brands that triangulate do not wonder. They know their position. And knowing your position is the entire advantage, because you can only steer toward a destination you can actually locate.

The Final Take

Every instrument on the bridge will lie to you eventually.

The compass drifts. The log accumulates error. The satellite drops signal. Last-click points to the wrong channel. IP matching mistakes a path for a cause. Incrementality goes quiet between tests. The mix model speaks too slowly to save a single campaign.

None of them is the answer. All of them are.

The navigator does not pick the one true instrument, because there is no one true instrument. The navigator takes three bearings and finds the point where they cross.

That is what CTV measurement actually is. Not a dashboard. Not a single number. Not last-click attribution standing alone on the bridge, drifting a few degrees off and carrying the whole ship with it.

It is closed-loop IP matching for what happened. Incrementality for whether it would have happened anyway. Media mix modeling for how it all fits together.

Three lines. One crossing point. One truth.

One instrument lies.

Three instruments tell you where you are.

Cory Poccia CEO, CS & Co. Marketing Studio™