Did TV Create The Sales... Or Just Get Credit For Them?

What would happen to your sales if you turned the TV commercials off?

If you can't answer that, you don't actually know how many of the sales you're crediting to TV were incremental.

Because your dashboard can show you what happened after the commercial aired.

It can't show you the version of that same night where the commercial never ran.

And that's where TV attribution gets uncomfortable.

One company we work with had built what seemed like a perfectly reasonable workaround.

Roughly 35% of its users were unidentified, making it impossible to follow every viewer through to a purchase.

So the team established a normal sales baseline across its website and Amazon.

A commercial would air.

They'd watch sales for the next 30 minutes.

Anything above baseline?

Credit TV.

Simple.

Logical.

And potentially very expensive.

Because “after” and “because of” are not the same thing

The team was spending heavily on linear TV.

Some individual 30-second spots cost $7,000.

“When you're advertising a 30 second spot for $7,000, you need to sell a lot of units to get there.”

And the sales were showing up.

The problem was when they were showing up.

Much of the TV budget was concentrated between 7 p.m. and 10 p.m.

Which also happened to be...

“the prime time, shopping time.”

Now the baseline method has a problem.

Imagine normal sales are 100 units.

A commercial airs at 8 p.m.

Sales jump to 130.

The obvious conclusion is that TV created 30 additional sales.

Except you don't actually know that.

Maybe sales would have climbed to 115 because that's what normally happens around 8 p.m.

Maybe another campaign was driving demand.

Maybe customers had already decided to buy earlier in the day and simply happened to complete their purchase after the commercial aired.

Maybe TV really did create all 30.

The sales spike doesn't tell you which explanation is true.

That's the problem hiding inside what I think of as The Counterfactual Gap.

The number you really want doesn't exist in your dashboard

The question isn't simply:

How many people bought after the commercial?

It's:

How many people bought because the commercial ran who would not have bought otherwise?

That's incrementality.

And it requires comparing what actually happened against something you can never directly observe:

What would have happened without the ad?

That's the counterfactual.

You don't get to run August 12th twice.

Once with the TV campaign.

Once without it.

Then compare Shopify and Amazon at the end of the night.

So every measurement approach is trying, in one way or another, to get closer to that missing answer.

And that's why a post-ad sales spike can be so seductive.

You can see it.

You can put it on a chart.

You can point to the commercial and then point to the revenue.

It feels causal.

But proximity is not causality.

And the more expensive the channel becomes, the more expensive that assumption becomes too.

The real danger shows up in your next media plan

Suppose that $7,000 commercial is repeatedly followed by strong sales.

You conclude the spot is working.

So you buy more of it.

Then more.

Soon you're not merely giving TV credit for sales it may not have created.

You're funding tomorrow's TV budget with yesterday's attribution assumption.

That's when a measurement problem becomes an allocation problem.

Because the important question is no longer:

Did TV get credit?

It's:

Did spending more on TV create enough additional business to justify spending more on TV?

That's a different question.

And now that this company is a Blueprint client, it's starting to produce some very interesting answers.

Then the TV spend moved

In a recent review, the team put linear TV spend and purchases on the same view.

And something jumped out almost immediately.

There were periods where linear spend was roughly $11,000–$12,000.

Then roughly $24,000.

Then approximately $60,000.

You'd expect a movement that large in spend to produce a pretty noticeable movement in purchases if the additional TV dollars were producing a similarly large incremental effect.

But purchases didn't move anything like the spend did.

“The spend went up huge and there's the tiniest bump up.”

Another described what the graph appeared to show even more simply:

“Big spend, same result. Small spend, same result. Big spend, same result.”

But here's the important part:

It doesn't prove TV isn't working.

And pretending it does would simply replace one attribution mistake with another.

A weak relationship is evidence... not a verdict

There may still be meaningful value coming from TV.

Certain networks could perform differently.

TV could create effects that don't appear immediately as a purchase.

Timing matters.

Other marketing activity is happening simultaneously.

And the team is still working through the TV data and its timing before drawing stronger conclusions.

But that's precisely why the observation matters.

The evidence doesn't have to answer everything to challenge something.

In this case, it challenges the assumption that a large increase in TV spend should automatically receive credit for sales occurring around the same period.

If you can dramatically increase spend without seeing a comparable movement in the business outcome you're trying to create...

You have a reason to investigate.

Not a reason to declare victory.

Not a reason to declare the channel dead.

A reason to ask a better question.

This is where correlation becomes useful without pretending it's causation

There's a strange tendency in marketing measurement to treat correlation as either proof or useless.

It's neither.

Correlation can tell you whether two things are moving together strongly enough to deserve attention.

If TV spend repeatedly rises and purchases rise with it, that's evidence worth investigating.

If TV spend moves dramatically while purchases barely respond, that's evidence worth investigating too.

What correlation cannot do by itself is magically eliminate every other possible explanation and tell you exactly what would have happened without the spend.

That's why the useful progression isn't:

Correlation → certainty.

It's:

Signal → hypothesis → decision → stronger test when necessary.

That distinction matters enormously when millions of dollars are moving through the media plan.

Better measurement should tell you what the evidence allows you to conclude

This is where we think marketing measurement has to become more useful.

Blueprint brings linear TV into the same measurement environment as the rest of the media mix so teams can examine how changes in TV spending relate to changes in the business outcomes they actually care about.

Not just:

What happened in the 30 minutes after the commercial?

But:

When this investment changes, what changes in the business?

That gives you a better signal for whether the channel appears to be earning its budget.

And sometimes that evidence will be strong enough to support an allocation decision.

Sometimes it won't.

If you're considering a major reduction in TV spend and the observational evidence still leaves too much uncertainty, the answer isn't to manufacture more certainty from the same data.

The answer may be to run a stronger incrementality test.

Create a cleaner baseline.

Introduce enough controlled variation to see what happens when the spend changes.

Then see whether the business changes with it.

The goal isn't to make every chart sound definitive.

It's to know what the evidence is strong enough to tell you.

Because eventually this isn't really a TV question

It's a capital allocation question.

If another $100,000 goes into linear TV, what evidence suggests the business will get enough back to justify it?

And if the evidence doesn't support that next $100,000...

Where does it have a better chance of producing growth?

Maybe that's still TV.

Maybe it's a different part of the TV buy.

Maybe the evidence points somewhere else entirely.

That's the decision Blueprint is designed to help make.

Not by pretending every offline impression can be perfectly tied to an individual sale.

And not by claiming statistical relationships magically prove causation.

But by putting your media spend and business outcomes into one decision environment...

Showing you where meaningful relationships appear—or fail to appear...

And helping you determine where the evidence says the next dollar deserves a closer look.

Because sometimes the most valuable thing your measurement can tell you isn't:

“This channel drove $X.”

It's:

“You've been assuming this spend is creating the result. The data is giving you a reason to test that assumption.”

And when you're spending thousands of dollars for 30 seconds...

That's a question worth answering before you buy the next spot.

Want to see whether your biggest media investments are actually earning the budget?

FAQ

What is TV attribution?

TV attribution attempts to connect television advertising with downstream business outcomes such as purchases, leads or revenue. Because linear TV often lacks the user-level tracking available in digital channels, marketers may use baselines, statistical relationships, geographic tests or other measurement approaches to estimate its contribution.

Why isn't a sales spike after a TV commercial enough to prove ROI?

Because timing alone does not establish causation. Some customers may have purchased even if the commercial had never aired. Other marketing activity, natural shopping patterns and existing demand can also contribute to the observed increase.

What is incrementality in TV advertising?

Incrementality is the additional business outcome generated because the advertising occurred compared with what likely would have happened without it. The second scenario is known as the counterfactual.

What's the difference between attribution and incrementality?

Attribution assigns credit for an observed conversion. Incrementality asks whether the marketing activity actually caused additional outcomes that would not otherwise have occurred. A channel can receive attribution for a sale without necessarily being responsible for creating an incremental sale.

Can correlation prove that TV advertising caused sales?

No. Correlation can reveal useful relationships between changes in TV activity and business outcomes, but correlation alone does not eliminate other possible explanations. It can provide evidence for an allocation hypothesis and help determine when stronger testing is warranted.

When should a company run an incrementality test?

An incrementality test becomes especially valuable when a consequential budget decision requires stronger causal evidence than existing observational data can provide. The appropriate test depends on the available data, channel, business and ability to introduce meaningful variation.

How does Blueprint help measure linear TV?

Blueprint can bring linear TV spend into the same measurement environment as other paid media and business outcomes, helping teams examine how changes in TV investment relate to changes in results. Those signals can inform budget decisions and identify where further testing may be warranted rather than assuming every sale occurring after a commercial was caused by it.

Does Blueprint perfectly attribute every linear TV sale?

No measurement approach can directly observe the alternate reality in which the same customer journey occurred without the advertising. Blueprint's role is to improve the evidence available for the decision, identify meaningful relationships in the data and help teams determine when that evidence supports action or when stronger testing is appropriate.