Why High-ROAS Campaigns Do Not Always Deserve More Budget

Why High-ROAS Campaigns Do Not Always Deserve More Budget

When a campaign shows a 6x ROAS and the next best shows 2x, the default decision is obvious: move budget from 2x to 6x. In many cases this is correct. In enough cases to be a systematic problem, it is wrong, and the extra budget spent on the 6x campaign produces less incremental revenue than the 2x campaign would have with the same dollars.

ROAS (Return on Ad Spend): revenue attributed to ads divided by ad spend. A 4x ROAS means $4 of attributed revenue per $1 spent. ROAS measures attribution efficiency, not incremental value, a distinction that becomes important when a campaign is already capturing demand that would have converted without the ad.

Key takeaways

1. ROAS measures attributed revenue per dollar spent, not incremental revenue. A branded search campaign can show 15x ROAS while generating near-zero incremental lift because those users would have converted organically.
2. The three conditions that create a high-ROAS-but-low-incremental-value trap: branded keyword capture, retargeting of already-warm audiences, and bottom-of-funnel campaigns that harvest demand created elsewhere.
3. The marginal ROAS concept, what additional revenue does the next $1 in this campaign produce, is more useful than average campaign ROAS for budget allocation decisions.
4. The correct diagnostic: run an incrementality test or compare ROAS against organic baseline traffic. If organic traffic drops when the high-ROAS campaign pauses, the campaign is additive. If organic holds, the campaign is substituting.
5. Prooflytics flags campaigns where high ROAS coincides with declining impression share or narrow audience size, both signals that the campaign is harvesting a fixed pool rather than expanding it.

The mechanism: why ROAS and incremental value diverge

The operational pain this creates: a B2B SaaS performance team has three campaign types: branded search (12x ROAS), retargeting (6x ROAS), and prospecting (1.8x ROAS). Every quarterly budget review moves money toward branded and retargeting and away from prospecting. Pipeline slowly shrinks as the acquisition funnel narrows. After 6 to 9 months the high-ROAS campaigns have also degraded because they have exhausted the audience that prospecting used to replenish.

The core mechanism: branded search campaigns and retargeting campaigns capture intent that was created elsewhere, by organic content, word of mouth, a sales conversation, a product review. These campaigns get attribution credit for conversions that would have occurred anyway via direct navigation or organic search. Their ROAS is high precisely because they intercept already-converted demand. Adding budget to them does not create more demand; it adds more spend to the interception layer, which dilutes ROAS until the budget increase becomes unprofitable.

Prospecting campaigns create demand from audiences that have never engaged with the brand. Their ROAS is lower because they are working harder, converting cold traffic. But each incremental dollar in prospecting adds a genuinely new user to the pool, which compounds into future branded and retargeting conversions.

What the data shows: the three-layer analytics test

The ICP problem this creates for performance teams: most analytics infrastructure lives at Layer 1, descriptive reporting of what happened (spend, clicks, ROAS). Budget decisions made at Layer 1 look logical but miss the causal story underneath. Scaling a high-ROAS campaign is a Layer 1 decision; evaluating whether that ROAS reflects incremental value requires Layer 2 reasoning (predictive: what would have happened without the campaign?) and Layer 3 (prescriptive: what is the optimal budget given that answer?).

The three-layer analytics framework defines:

• Layer 1 (Descriptive): "Yesterday we spent $5K on branded search and got 80 conversions at 12x ROAS", the dashboard view
• Layer 2 (Predictive): "Without branded search spend, 65 to 70 of those users would have converted via direct or organic navigation", requires holdout testing or MMM
• Layer 3 (Prescriptive): "The incremental branded search budget should not exceed $800/day; above that, each dollar cannib alizes organic conversion at a worse net revenue per dollar than prospecting"

Industry research across 252 companies representing $53B in combined annual marketing spend found that 53% do not use forward-looking metrics (like incremental lift or predictive models) in campaign planning. They operate at Layer 1 and make budget decisions that look analytically rigorous but are structurally blind to incrementality.

Prooflytics surfaces the Layer 2 diagnostic: when a high-ROAS campaign shows declining impression share (audience is saturating) alongside stable or growing organic traffic, the briefing flags it as a potential cannibalization signal and recommends an incrementality check before the next budget increase.

Prooflytics

Make the call with the whole picture

Briefs are daily; the understanding compounds.

Start free trialSee pricing

14 days free · no credit card

01. When to scale a high-ROAS campaign

Scale is justified when:

• Impression share is below 60 to 70%: the campaign is not capturing all available demand. Adding budget increases reach rather than frequency.
• Audience size is large relative to current reach: retargeting audiences with millions of potential users can absorb budget increases with proportional returns. Retargeting audiences of 5,000 users cannot.
• Incrementality test confirms lift: a holdout test (running the campaign for 70% of eligible users, withholding from 30%) shows that the 30% who did not see the ad converted at materially lower rates. If conversion rates are similar in both groups, the campaign is harvesting, not creating.
• CPA is stable or improving as budget increases: if adding 20% more budget produces 20% or more conversions, marginal ROAS is not degrading.

02. When high ROAS is a warning signal

High ROAS combined with any of these conditions suggests harvesting rather than growing:

• Audience size below 10,000 in retargeting: frequency builds quickly on small audiences. A 6x ROAS on a 5,000-person retargeting list likely has a frequency above 8 to 10, meaning budget is being spent showing ads to people who have already converted or decided not to.
• Branded keyword share above 40% of search campaign budget: if most of your high-ROAS search spend is on your own brand name, competitors are not bidding aggressively on your brand (good), but the campaign is capturing navigational intent that would convert without the ad (and without the CPC cost).
• Organic traffic drops when the campaign pauses: run a 7-day pause test. If organic search and direct traffic both increase during the pause, the campaign was substituting for organic conversion, not supplementing it.
• Impression share above 85%: the campaign is already capturing nearly all available demand in its targeting parameters. Budget increases produce frequency, not reach.

03. The marginal ROAS framework for budget decisions

Average campaign ROAS is a snapshot of historical performance under current budget. Marginal ROAS, what the next $1 produces, is what determines whether a budget increase is rational.

Estimating marginal ROAS without a formal incrementality test: increase budget by 15 to 20% for 14 days and observe whether ROAS holds, improves, or degrades. If ROAS drops more than 15% relative to the budget increase, you are past the marginal efficiency point for that campaign.

The formula that connects these concepts: if a campaign spends $10,000 at 5x ROAS ($50,000 revenue) and an additional $2,000 produces only $5,000 in additional attributed revenue (2.5x marginal ROAS), the average ROAS of the campaign now looks like $55,000 revenue on $12,000 spend = 4.6x. The blended average obscures the fact that the last $2,000 underperformed significantly.

For DTC Shopify brands where contribution margin matters more than gross ROAS, the relevant threshold is not 1x ROAS but the break-even ROAS: 1 divided by (1 minus COGS percentage). A brand with 40% COGS needs at least 1.67x ROAS to break even on ad spend, meaning a campaign at 2x ROAS is marginally profitable but a campaign at 1.5x is burning cash regardless of what the dashboard shows as "positive ROAS."

Bottom line

• High ROAS does not mean the campaign is creating incrementally more revenue. It often means the campaign is efficiently capturing demand that already existed.
• The three conditions that create high-ROAS-low-increment campaigns: branded keyword capture, small retargeting audiences with high frequency, and bottom-of-funnel harvesting of demand created by prospecting or organic channels.
• The correct budget allocation framework uses marginal ROAS (what the next dollar produces) not average ROAS (what historical spend produced). These diverge when campaigns hit audience saturation.
• Incrementality tests, holdout experiments comparing conversion rates with and without ad exposure, are the only reliable way to confirm that ROAS reflects real lift.
• Prooflytics monitors impression share, audience size, and ROAS trends daily and flags campaigns where high ROAS coincides with saturation signals before the budget decision compounds the error.

You can read independent reviews of Prooflytics on G2 and compare it to alternatives in the marketing analytics category.

Connect your Google Ads and Meta Ads to Prooflytics and see which of your high-ROAS campaigns show saturation signals in tomorrow's briefing.