Incrementality: Assessing Real Marketing Impact Beyond Attribution Metrics

Traditional attribution models in marketing report what happened after an interaction but rarely clarify what would have occurred without it. This distinction sits at the centre of the incrementality debate: attribution tallies up conversions touching a channel or partner, while incrementality isolates the actual lift a marketing action delivers, offering a credible answer to what marketing activities really cause.

Incrementality: What It Measures That Attribution Misses

Incrementality quantifies the additional conversions, customers, or revenue resulting from a specific marketing action—relative to a baseline that excludes the activity. This approach avoids over-crediting channels simply for being present in the conversion pathway. For example, if a customer clicks an affiliate’s link and buys five minutes later, attribution software records that as a partner-driven conversion. In reality, the customer may have already resolved to purchase, inflating the partner's perceived value.

Modern marketers need incrementality to answer the central questions attribution cannot:

• Did this campaign actually create new sales, or just intercept ones set to occur?
• Which partners drive genuinely new demand versus those that primarily capture what already existed?
• Are high ROAS figures masking limited incremental growth?
• If spending increases on a channel, will that yield new conversions, or are we above the point of diminishing returns?

To illustrate, if an affiliate program is attributed with 10,000 sales, but a controlled incrementality test shows only 1,000 more sales than a non-exposed group, then the true incremental gain is closer to 1,000—illustrating the risk in crediting all attributed activity as causal.

Attribution vs. Incrementality: Causality Matters

The key operational difference: attribution assigns credit, incrementality tests causality. Last-touch and multi-touch models, no matter how sophisticated, simply distribute credit based on rule sets. They do not ask whether the marketing interaction changed the outcome.

Google’s Conversion Lift documentation makes this line explicit: attributed conversions derive from rules and tracking, while incremental conversions require contrasting treatment and control groups.

"One measures involvement. The other measures cause and effect." — Trackier

The risk in ignoring this distinction is blunt: marketing teams may misallocate spend, rewarding channels inflating numbers rather than growing the market.

Why Combine Attribution and Incrementality for Smart Budgeting

Debate about which measurement should prevail is largely a false dichotomy in practical campaign management. Attribution remains valuable for mapping customer journeys and tracking near real-time campaign performance. But incrementality adds a crucial causal layer, grounding those numbers with evidence.

For example, a content partner may show $500,000 in attributed revenue. Only after running an incrementality test does a team know if exposures to that partner truly deliver higher conversion rates than a matched, unexposed cohort. If so, that partner deserves recognition; if not, their reporting value is inflated.

Why Incrementality Is Gaining Ground in Marketing

The proliferation of marketing channels—organic, social, affiliate, paid, influencer—means a single conversion often appears in multiple reporting systems, multiplying credit instead of clarifying cause. At the same time, marketing leaders are held to tougher standards for budget justification.

Incrementality is central to this accountability. Research supports its practical value: a 2026 Nielsen study of Pinterest campaigns in Canada’s CPG sector found 80% of campaigns produced measurable incremental sales, providing a real-world benchmark for what “working” means from a causality lens.

For operators running affiliate or partner marketing programs, prioritising incrementality enables:

• Tighter budget allocation
• Clearer partner evaluation (who drives true growth versus simply intercepting conversions)
• Reduction in cannibalisation between channels
• More accurate commission and bonus frameworks
• Quicker identification of diminishing returns

Incrementality in Affiliate and Partner Marketing

Affiliate and partner marketing presents some unique incrementality challenges. Consider the wide variation:

• Content publishers may spark initial brand discovery.
• Influencers build consideration.
• Comparison sites guide product selection.
• Coupon aggregators or cashback platforms close out conversions, sometimes for buyers already set to purchase.

While every partner can drive attributed conversions, their incremental impact varies. Partner A may log 10,000 orders, but testing finds little lift compared to baseline. Partner B produces only 6,000 ordered conversions but a substantial rise in new customers. Incrementality reveals growth drivers hidden by surface-level attribution.

Evaluating partner performance requires going beyond conventional metrics to review:

• Incremental conversion rate
• Incremental revenue and new customers
• Incremental CPA and ROAS
• Average order value
• Customer lifetime value

This multidimensional assessment sharpens investment priorities far more than dashboards reporting only credited conversions.

How to Test and Measure Incrementality

Incrementality testing hinges on the quality of your experimental setup—especially audience segmentation and outcome measurement. Before running a test, marketers need clear definitions for iROAS (incremental ROAS), distinct from conventional ROAS. iROAS answers, “How much revenue truly resulted from the money spent?” by subtracting conversions in an unexposed control group from those in an exposed treatment group. Divide this incremental lift by ad spend to calculate iROAS.

Common incrementality test formats include:

• Randomised Holdout Test: Split your audience; one group sees the marketing (treatment), the other doesn't (control). The conversion gap is your incremental impact. Google’s Conversion Lift operates on this logic. Example: Retaining 10% of users from an affiliate promo, then comparing conversion rates and attributing the difference to true lift.
• Geographic Test: Assign regions as treatment/control. Track parallel outcomes and adjust for baseline differences. Particularly helpful when user-level segmentation isn’t feasible.
• Scale Test: Examine what happens when spend increases—helpful for detecting diminishing returns. If incremental revenue rises at a much slower rate than spend, the channel’s efficiency is declining.

Practical Considerations and Failure Points With Incrementality Testing

Three recurring pitfalls can trip up even tightly controlled incrementality tests:

1. Control Group Contamination

Users in the control group sometimes interact with the marketing by other means—organic search, independent publisher discovery, or sharing. This contaminates the test group, reducing the clarity of any observed lift.

2. Seasonality

Periodic demand shifts (holidays, paydays, competitive campaigns) can mask or exaggerate incremental effects. If a test overlaps with a peak period, it can mislead on real lift.

3. Sample Size

Statistical significance matters. Small samples often deliver random, unreliable outcomes. Calculate sample size requirements before the test, not after, to safeguard decision confidence.

"A striking relative lift with minimal conversions is likely just noise, not signal." — Oliver Grant

Key Takeaways for Performance Marketers

The ultimate question is not “who got credit” but “what additional value did marketing create?” Attribution identifies associations; incrementality probes causality. This only becomes valuable when test results meaningfully influence campaign and budget decisions.

A pragmatic decision matrix:

• High attributed & high incremental performance: Consider scaling spend.
• High attributed & low incremental performance: Look for cannibalisation or inflated credit.
• Low attributed & high incremental: Investigate under-attribution.
• Low attributed & low incremental: Reassess the channel entirely.

Operators should combine attribution, experimental analysis, and business metrics—not pit them against each other. The goal isn’t picking a “winner” from the attribution dashboard but understanding where additional marketing investment will drive true growth.

For more insights on optimising campaigns and evaluating partners, see our B2B technology and news coverage.