Anything above baseline went to TV.

That was the rule.

Take the normal level of Shopify and Amazon sales.

Run the commercial.

Watch the next 30 minutes.

If orders climb above normal...

Credit the difference to television.

It was a practical workaround for a difficult problem.

Roughly 35% of their users couldn't be identified well enough to follow them through to purchase.

So without a clean customer-level trail, the team looked at what happened to sales immediately after the commercial instead.

And it seemed to work.

Until someone on the team pointed out one inconvenient detail:

“It also happens to be the prime time, shopping time.”

Suddenly, the same sales spike could tell two completely different stories.

The commercial aired...

And convinced people to buy.

Or...

The commercial aired during the exact 30 minutes people were already most likely to buy.

The orders look identical either way.

The economics don't.

Because 30 seconds can cost $7,000

As one marketing leader put it:

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

And that turns a seemingly academic attribution question into a very expensive one.

Suppose a commercial airs at 8:00 PM.

Orders normally spike around that time.

But tonight they spike even higher.

Your attribution model assigns everything above baseline to TV.

The spot looks profitable.

So you buy it again.

Maybe at the same price.

Maybe you bid more for the time slot.

Maybe the network gets more budget next month.

Except there's a number missing from that entire calculation:

How many of those customers would have purchased anyway?

That's the number you can't see.

And it's the number that determines whether the commercial actually created incremental growth.

The most dangerous sales aren't necessarily the ones you can't attribute

Sometimes they're the ones you attribute too easily.

Because there's something incredibly convincing about sequence.

Ad airs.

Sales rise.

Therefore, ad drove sales.

But the first two statements are observations.

The third is a conclusion.

And when an advertisement happens to run during a naturally high-conversion period, that conclusion becomes much harder to make.

Imagine running an ice cream commercial on the hottest afternoon of the year.

Sales surge.

Was it the commercial?

The temperature?

Both?

How much would you have sold without the commercial?

That's the problem this team had run into with prime shopping hours.

Their baseline could tell them what sales normally looked like.

Their 30-minute window could tell them what happened next.

But neither could reveal the missing alternate reality:

What would have happened during those same 30 minutes if the $7,000 commercial had never aired?

That's the Counterfactual Gap.

And traditional attribution has a difficult time closing it.

A sales spike proves timing... not incrementality

This distinction matters because marketing teams often ask attribution systems to answer two very different questions as though they're the same.

The first is:

What happened after we advertised?

The second is:

What changed because we advertised?

Those aren't interchangeable.

If 100 orders arrive after a commercial but 90 would have happened anyway...

The advertising didn't create 100 incremental orders.

The economically interesting number is closer to the difference.

And that's why simply making the post-commercial window more sophisticated doesn't necessarily solve the problem.

You could monitor five minutes instead of 30.

You could establish a better baseline.

You could build a prettier dashboard.

You could get the sales data faster.

But you're still fundamentally asking:

What happened after the ad?

What you actually need is evidence that changes in the advertising are associated with changes in the business.

And that's where Blueprint approaches the problem differently.

Instead of watching the spike... watch what happens when the spend moves

Blueprint's impact methodology doesn't start by automatically claiming the orders that appeared immediately after an offline ad.

It looks for variation.

Because advertising spend is rarely perfectly constant.

One day a channel spends more.

Another day it spends less.

Campaigns change.

Networks change.

Other channels move at the same time.

And purchases move too.

Blueprint statistically evaluates those small changes in media investment against the changes occurring in purchases across the business.

Then it looks for the relationship.

When spend here changes... do results reliably move with it?

That sounds like a subtle difference.

It's not.

The 30-minute model essentially asks:

“Sales rose after TV aired. How much credit should TV get?”

The impact model asks:

“When TV changes, what happens to the business?”

Now you're no longer relying on one convenient sales spike to prove the investment worked.

You're looking for a pattern that continues to show itself as the investment changes.

The size of the budget doesn't determine the answer

This is particularly important when one channel already consumes an enormous amount of money.

If you're spending heavily on TV, you don't want a measurement system that effectively reasons:

TV received a huge amount of spend... therefore TV must have created a huge amount of the result.

That turns the conclusion into an assumption.

Blueprint's impact coefficient isn't based simply on how much money went into a channel.

It's based on the variability in that investment and the corresponding movement in business results.

Think about the difference.

Imagine TV receives $900,000 this month.

Meta receives $300,000.

TV shouldn't automatically receive three times the importance simply because it received three times the budget.

Instead, you want to observe what happens as those investments move.

When TV receives more...

Do purchases meaningfully move?

When Meta receives more...

Do purchases meaningfully move?

When either receives less...

What happens then?

Over enough observations, those movements begin creating a measurable relationship between investment and outcome.

Blueprint uses that relationship to build an Impact Score.

Now you're measuring evidence of impact...

Rather than assuming impact from proximity or spend.

TV can finally compete against the rest of the media mix

This matters because the commercial wasn't running in a vacuum.

While TV was airing...

Meta was running.

Google was running.

YouTube was running.

Other advertising could be creating demand.

Customers could have interacted with previous campaigns.

And some customers were simply going to buy.

Blueprint can bring offline media into the same measurement environment as the digital channels already competing for budget.

For offline media, structured information such as the day, spend and an identifier for the advertising can be brought into the system.

Now the team isn't limited to asking:

Did sales rise after TV?

They can begin comparing the measurable impact of TV with what is happening across the rest of their paid media.

And that's where the $7,000 question gets much more interesting.

Because maybe TV really is creating substantial additional demand.

If the relationship between TV investment and purchases keeps showing up, that's evidence worth paying attention to.

But maybe another channel is moving purchases more efficiently.

Maybe TV is working overall, but the relationship is stronger on one network than another.

Maybe additional spend stops producing the same movement in results after a certain point.

Those are allocation questions.

And they're much more valuable than simply deciding who gets credit for the orders that already happened.

More data can make the answer more specific... but only when the data earns it

There's another important principle behind Blueprint's approach:

Don't manufacture precision the data can't support.

Initially, TV may need to be evaluated as a broader channel.

But with enough spend variation and enough purchases, the analysis can potentially become more granular.

Blueprint refers to this as event density.

The more meaningful events you have—in this case, purchases—and the more useful variation there is in the media activity surrounding them, the more opportunity there is to examine what is happening below the channel level.

That could eventually mean comparing networks.

Or campaigns.

And where sufficient data supports it, potentially identifying particular spots producing outsized impact.

But that's very different from declaring:

“The commercial aired at 8:07 PM and created exactly 23 sales.”

Blueprint doesn't need to pretend it knows something the data can't actually establish.

Because false precision is exactly how you end up back where this team started:

Taking an observable event...

And turning it into a causal claim the evidence doesn't support.

Sometimes “certain enough to act” is more valuable than pretending to be certain

Measurement doesn't need to produce an immaculate answer about every historical sale before it becomes useful.

It needs to produce enough evidence to make the next decision better.

If the Impact Score suggests TV is moving results more than its traditional attribution suggests...

That's a hypothesis you can test with budget.

If another channel appears to have substantially more impact than it's receiving credit for...

That's a hypothesis you can test too.

Move money.

Observe what happens.

See whether the relationship holds.

And let each new allocation create additional evidence.

That's a very different feedback loop from:

Commercial aired → sales spiked → buy another commercial.

Because now your next investment isn't merely the consequence of the model.

It becomes another test of the model.

Which finally gets you to the question that matters

This team didn't really need to know whether TV deserved credit for an order at 8:17 PM.

They needed to know whether spending thousands of dollars on 30 seconds of television was creating enough additional business to justify doing it again.

And if they had another $7,000 available...

Whether it belonged in that commercial at all.

That's where impact measurement becomes allocation.

Blueprint can use the evidence accumulating across channels to help evaluate where additional budget appears to have the greatest opportunity.

The system looks for the productive range it can currently see... the point before additional investment appears to become less attractive.

Internally, Blueprint describes this as finding the “bend in the knee.”

So the question progresses again.

Not:

“Did TV get the sale?”

But:

“Is TV moving the business?”

Then:

“How strongly?”

Then:

“Compared with what else we could fund?”

And ultimately:

“Where should the next dollar go?”

That's the question traditional attribution often struggles to answer because it's obsessed with dividing up yesterday's conversion.

But yesterday's conversion is already gone.

The expensive decision is what you do tomorrow.

The $7,000 sales spike might be completely real

That's worth emphasizing.

This isn't an argument that TV didn't work.

Maybe it worked extraordinarily well.

Maybe the commercial created a huge portion of that spike.

Maybe TV is generating demand that gets credited somewhere else entirely.

The point is that the spike itself can't tell you.

Because when the commercial and natural customer demand peak at the same time, correlation becomes incredibly easy to mistake for causation.

So instead of asking:

“How many orders arrived after we spent the $7,000?”

Start asking:

“When we change what we spend, what actually changes with it?”

Do that across enough observations...

Across TV...

Across Meta...

Across Google...

Across the rest of the media mix...

And you start building something far more valuable than a story about who deserves credit.

You build evidence for where the money is actually moving the business.

Because the $7,000 question was never really:

Did sales happen after the spot?

They clearly did.

The question was:

How many of those sales wouldn't have happened without it?

That's the difference between paying for sales...

And paying for customers who were already on their way.

FAQ

Does a sales spike after a TV commercial prove the commercial worked?

No. A post-commercial sales spike establishes a temporal relationship, but it doesn't by itself establish that the commercial caused the increase. If the ad runs during a naturally high-conversion period, some of those customers may have purchased even without seeing the commercial.

What is the counterfactual in advertising measurement?

The counterfactual is what would have happened if the advertising had not occurred. If sales rise after an ad, the incremental value of the advertising depends on how many of those sales would not otherwise have happened.

How does Blueprint approach TV advertising measurement?

Rather than automatically assigning sales immediately following a TV spot to television, Blueprint can statistically examine variations in media spend alongside variations in purchases and other business results. When a sufficiently strong relationship emerges, that information can contribute to an Impact Score used to evaluate the channel.

What is Blueprint's Impact Score?

Impact Score is Blueprint's method for expressing the statistical relationship between changes in an advertising channel and changes in business outcomes. It is designed to help evaluate channels whose effect may not be fully captured through deterministic digital attribution.

Can Blueprint measure individual TV networks or commercials?

Potentially, depending on the available event density. Blueprint can begin with TV at the broader channel level and may become more granular when sufficient purchase volume and media variation support the analysis. Exact hourly or individual-spot attribution should not be assumed where the underlying data cannot support that precision.

Why does spend variability matter when measuring advertising impact?

Variation creates opportunities to observe whether business outcomes move as advertising investment changes. Rather than assuming a channel is important because it receives a large budget, the analysis can examine whether changes in that investment repeatedly correspond with changes in results.

How can impact measurement improve media allocation?

Impact measurement can help marketers compare evidence of how different channels affect business results and use those signals to inform where additional budget may have the greatest opportunity. Those allocation decisions can then create additional evidence about whether the measured relationship continues to hold.