How Much of Your Marketing Budget Is Actually Working?

John Wanamaker supposedly said that half of his advertising was wasted, he just did not know which half. That was in the 1800s.

More than a century later, most marketing organizations are in roughly the same position. The tools are more sophisticated. The dashboards are more colorful. The attribution reports are longer and more detailed. But the fundamental question, which dollars are actually doing something versus which ones are being absorbed by customers who were going to buy anyway, remains genuinely hard to answer.

For most brands, the honest answer is that between 30 and 60 percent of their marketing spend is generating little to no incremental return. Not because the marketing is poorly executed. Because it is reaching people who were already in the funnel, reinforcing decisions that were already made, and collecting attribution credit for conversions that had nothing to do with the ad.

The waste is not always obvious. It hides inside metrics that look healthy.

If you are running paid media at scale, you almost certainly have a reporting setup that shows ROAS by channel, cost per acquisition by campaign, and some version of return on marketing investment. These numbers tend to look reasonable. Some look quite good.

The problem is what they are measuring.

Ad platform metrics count attributed conversions. An attributed conversion is a purchase that happened after someone interacted with an ad, within whatever attribution window the platform uses. By that definition, every customer who clicks a retargeting ad on their way to a checkout they had already decided to complete is an attributed conversion. Every brand search from a loyal customer who would have found you organically is an attributed conversion. Every purchase from someone who was three days away from buying regardless is an attributed conversion.

The ad did not cause those purchases. It was present when they happened. The platform counted them.

This is not a small rounding error. For brands with meaningful organic demand, strong word of mouth, or high repeat purchase rates, the overlap between attributed conversions and conversions that would have occurred anyway is substantial. When you optimize your budget based on attributed metrics, you are optimizing for efficiency at claiming credit rather than efficiency at generating new revenue.

A channel can have excellent attributed metrics and be generating almost no incremental return. That is not a hypothetical. It is what controlled experiments routinely reveal when brands test their media spend rigorously for the first time.

Every business has a baseline: the revenue it would generate if it spent nothing on advertising. Loyal customers who reorder on a schedule. People who search your brand name and buy. Word of mouth conversions. Organic traffic that converts at a consistent rate.

This baseline revenue is not zero, and for established brands it is not small. It exists regardless of how much you spend on paid media. The job of advertising is to generate revenue above that baseline, by reaching new customers, pulling forward purchases from existing ones, or expanding purchase size. That incremental revenue, the amount above what would have happened anyway, is the real return on your marketing investment.

When you run retargeting campaigns that reach customers already in checkout, or branded search campaigns that capture searches from people already committed to buying, you are spending media budget on top of baseline conversions. The ROAS looks great because the denominator (spend) is small and the attributed revenue is high. The incremental return is close to zero because you were not the reason those purchases happened.

The practical test is simple to describe, though not always easy to run. If you paused this channel entirely for four weeks, how much revenue would you lose? For brand search campaigns on buyers who already know your brand well, the answer is often far less than the attributed revenue would suggest. For retargeting, same story. For prospecting campaigns reaching genuinely new audiences, the answer is usually closer to the attributed number.

That difference, between what attribution claims and what a pause would actually cost you, is where the waste lives.

Different channel types have different baseline overlap dynamics. These are generalizations, but they hold up across most brands that have run serious incrementality tests.

Retargeting is the most common source of inflated efficiency numbers. By definition, retargeting reaches people who have already been to your site or interacted with your brand. Many of them were already close to converting. The retargeting ad reaches them, they convert, the platform takes credit. For most brands, retargeting is worth some spend because it does accelerate some conversions and recover some that would have been lost. But the marginal return on retargeting spend diminishes quickly, and the attributed ROAS grossly overstates the actual incremental value.

Branded search captures demand that often exists independently. Someone who types your brand name into Google was already looking for you. The question is whether they would have found you organically if your paid brand search ad was not there. For most brands, the answer is: usually yes. Branded search has a real role in controlling your search real estate and preventing competitors from capturing brand queries. But the revenue attributed to it is largely baseline revenue, not incremental revenue driven by the advertising.

Broad match keywords and automated bidding expand reach into low-incrementality territory. When Google’s Performance Max or Meta’s Advantage+ campaigns optimize for conversions, they naturally find the easiest conversions to claim, which tend to be customers already in your funnel. The campaigns look efficient. The incremental returns are often much lower than the ROAS suggests.

Upper funnel and prospecting tend to be undervalued. Because prospecting reaches cold audiences, conversions are less immediate and attribution capture is lower. Time-decay models penalize early touchpoints. This leads most teams to underinvest in prospecting relative to what an incrementality-based view would recommend, because prospecting’s true contribution shows up in future conversion rates and brand awareness, not in the last 7-day click window.

The only way to establish whether a channel or campaign is generating incremental revenue is to run an experiment where you withhold it from some portion of your audience and measure what happens.

Geo holdout testing is the most practical approach for brands running national or regional campaigns. You split your geographic markets into a test group and a control group, run advertising in the test group only, and compare revenue outcomes. If the markets are well-matched going in, the revenue difference between them represents the incremental lift from your advertising. This approach is independent of cookies, attribution windows, and platform reporting. It measures actual revenue, not attributed revenue.

A well-run geo holdout test typically takes four to six weeks. The outputs are specific: this channel drove X dollars of incremental revenue at Y incremental ROAS. You now have a real number for what this channel is worth, not a platform’s claimed attribution.

Media mix modeling provides a complementary view at the portfolio level. Rather than testing one channel at a time, MMM uses historical spend and revenue data across all channels simultaneously to estimate where your incremental returns are coming from. It can identify diminishing returns curves for each channel, show you where you are over-invested relative to incremental returns, and model the expected outcome of budget shifts before you make them.

Holdout tests for specific campaigns can validate whether individual retargeting audiences, branded search campaigns, or prospecting segments are generating real lift. The setup is simpler than a full geo test: randomly split your audience, withhold the campaign from the holdout group, compare conversion rates. Most major platforms offer some version of this as a native test, though the methodological rigor varies.

When brands run their first serious incrementality tests, the results usually point in a consistent direction. Retargeting is over-invested relative to its incremental returns. Branded search spend is higher than what the incremental lift justifies. Prospecting is under-invested. Upper funnel channels like connected TV, audio, or display are undervalued because their impact is diffuse and their attributed metrics are weak.

A reallocation based on incrementality data typically involves three moves.

First, you set a floor on retargeting and brand search, enough to capture the lift that does exist without over-investing in conversions that would have happened anyway. The efficiency on these channels looks worse on attributed metrics when you cut them. Revenue holds up better than the attributed metrics would predict.

Second, you move that budget into prospecting and upper funnel channels that have demonstrated incremental lift. These channels will show lower attributed ROAS. They will add new customers and build the demand pool that retargeting harvests later.

Third, you set iROAS thresholds for each channel based on your actual test results and use those as the standard for budget allocation decisions going forward. Channels that clear the threshold get investment. Channels that fall below it get scrutinized or cut.

The disorienting part of this transition is that the dashboards get harder to read in the short term. Attributed ROAS may decline when you shift budget from high-credit retargeting to lower-credit prospecting. Revenue typically holds or grows because you are generating more actual demand. Learning to read through the attribution noise is part of what makes this transition uncomfortable and also part of what makes it worthwhile.

Most marketing budget conversations center on performance. Which channels are performing? What is the ROAS? What is the CPA? These are not bad questions, but they are not the right questions if performance is being measured through attribution.

The right question is simpler and harder: if we turned this off, what would actually happen?

That question forces the conversation toward causality. It is uncomfortable because most teams cannot answer it with data. They have attribution reports. They do not have holdout results. The attribution report tells them what would be attributed to the channel if it ran. It does not tell them what the business would lose if it stopped.

Every channel that cannot answer that question with experiment data is a channel where you are essentially trusting the platform to tell you whether the platform is worth paying for. That is a conflict of interest that no amount of sophisticated attribution modeling resolves.

Measured was designed to answer the question this article is built around: what is your marketing budget actually generating?

The platform runs continuous geo holdout tests across every channel in your media mix, producing channel-level iROAS benchmarks that are grounded in controlled experiments rather than attribution models. When a channel is over-invested relative to its incremental returns, the data surfaces it. When a channel is undervalued because its attributed metrics undercount its actual contribution, that surfaces too.

For most brands that come to Measured, the initial audit of their media mix turns up both kinds of surprises. Channels that looked efficient on paper that are generating thin incremental returns. Channels that looked average that are actually driving significant new revenue. The budget allocation that falls out of that picture is different from what attribution-based reporting would suggest, and in most cases the revenue outcomes improve when teams act on it.

The goal is not to spend less on marketing. It is to spend with confidence that you know what is working and why. Most measurement stacks cannot honestly claim to provide that. Incrementality testing can.

How much of a typical marketing budget is wasted? This varies significantly by industry, brand maturity, and channel mix, but research and real-world testing consistently find that a meaningful share of attributed marketing performance reflects baseline conversions rather than incremental ones. Brands running rigorous incrementality tests frequently find that 30 to 50 percent of attributed revenue in channels like retargeting and brand search would have occurred without the advertising. The “waste” is not always eliminated spend but rather spend that could generate more incremental return if redirected.

What is the best way to allocate a marketing budget? The most reliable approach is to base allocation on incremental return by channel, rather than attributed return. This requires running controlled experiments, typically geo holdout tests or conversion lift studies, to establish what each channel is actually causing versus what it is claiming credit for. Media mix modeling can provide a portfolio-level view to complement individual channel tests. Channels that demonstrate high incremental ROAS above your margin threshold should receive more investment; channels with low incremental ROAS should be reduced regardless of how their attributed metrics look.

What is incremental ROAS? Incremental ROAS, or iROAS, measures the revenue your advertising actually caused divided by what you spent on it. It is calculated using controlled experiments rather than attribution models. A channel with a 6x reported ROAS might have a 1.5x iROAS if most of its attributed conversions would have happened anyway. iROAS is a more reliable basis for budget allocation because it reflects genuine causal impact.

How do I know if my retargeting is generating real returns? Run a holdout test. Randomly split your retargeting audience and withhold ads from the holdout group for three to four weeks. Compare the conversion rate between the exposed and unexposed groups. If conversion rates are similar, the retargeting is mostly capturing conversions that would have occurred regardless. If there is a meaningful gap, the retargeting is generating genuine incremental lift. Most platforms offer some version of this test natively, though the methodology varies.

Should I cut branded search spend? Probably not entirely, but most brands are overspending on it relative to its incremental impact. The key question is how much of your branded search traffic would convert through organic results if paid brand search was paused. For many brands with strong organic rankings on their own brand terms, the answer is that a significant majority of branded search conversions are baseline, not incremental. A holdout test on a subset of markets or a targeted audience can give you an actual number to work with.

What is a marketing budget audit? A marketing budget audit is a structured review of where your spend is allocated and what each portion is actually generating in incremental returns. A rigorous audit goes beyond attributed performance metrics and incorporates incrementality test results, MMM output, and a systematic comparison of channel-level iROAS against margin thresholds. The output is a recommendation for how to reallocate budget toward channels with proven incremental return and away from channels generating primarily attributed revenue from baseline conversions.

Measured helps brands run the incrementality tests that a real marketing budget audit requires. If you want to know what your spend is actually generating, learn more at measured.com.