Ask any CMO in Australia what keeps them up at night in 2026 and "measurement" will be in the top three. Budgets are under pressure, media mixes are more fragmented than ever, and the old shorthand of last-click attribution has quietly stopped answering the only question that really matters: what would have happened if we hadn't run this campaign at all? That question is the whole point of incrementality testing — and it's fast becoming the measurement discipline every omni-channel marketer needs to understand.
This guide is written for marketers who have heard the term "incrementality" thrown around in agency decks and vendor pitches but haven't had it explained cleanly. We'll cover what it is, why it matters more in an omni-channel world, the main methods you can use, common pitfalls, and how to get started without needing a data science team of ten.
What incrementality testing actually is
Incrementality testing measures the causal lift a marketing activity delivers — the sales, visits, or brand outcomes that would not have happened without it. It answers a very specific question: if I hadn't shown this ad, run this DOOH burst, or activated this retail media placement, would the outcome have been different? Everything else — reach, impressions, click-through, even last-touch conversions — is just correlation dressed up as insight.
The mechanic is simple in theory. You compare a group exposed to your marketing (the test group) to a matched group that wasn't exposed (the control group). The difference in outcomes is the incremental lift. In practice, building a genuinely comparable control group is where most brands fall over, because omni-channel media exposures rarely respect neat boundaries.
Why omni-channel makes incrementality both harder and more important
In a single-channel world — say, paid search — incrementality was already contested but at least tractable. Google's own studies over the years suggested that a meaningful share of branded search clicks are not incremental at all: consumers were going to come anyway. That's a useful data point, but it only covers one channel.
Omni-channel campaigns explode the complexity. A shopper might see a DOOH billboard on their commute, get retargeted on connected TV that night, receive a retail media banner inside the Coles or Woolworths app the next morning, and finally convert in store. Which of those exposures was incremental? Which was the icing on a decision that had already been made? Traditional multi-touch attribution divides the credit; incrementality testing asks whether the credit should exist at all.
Recent industry benchmarks tell the story. Multiple ANZ media effectiveness studies over the past 18 months have shown that when brands run properly designed incrementality experiments, the measured lift from channels like DOOH and connected TV is often materially higher than last-touch attribution suggests, while some lower-funnel digital channels look less heroic than their dashboards imply. It's uncomfortable reading for anyone whose planning has been anchored to attribution reports — but it's also where the budget-reallocation opportunities live.
The four methods most marketers should know
There's no single "right" way to run an incrementality test. The right method depends on the channel, the outcome you're measuring, and how much scale and time you have. Four approaches cover the vast majority of what Australian brands actually run today.
Geo-based holdout tests: split your media plan by geography, holding out a matched set of markets from a specific channel or campaign. Ideal for DOOH, TV and broad-reach channels where user-level randomisation isn't possible.
User-level randomised control trials (RCTs): platforms like Meta, YouTube and some retail media networks can randomly assign users to test vs. control cells and report incremental outcomes. High rigour, but limited to walled gardens.
Ghost bids and ghost ads: in programmatic, the DSP records bids or impressions it would have served to a random subset of users but suppresses them. This creates a clean control cohort inside a live campaign.
Matched-market and synthetic control methods: statistical techniques that construct a "synthetic" version of your test market from a weighted combination of other markets. Useful when you can't cleanly hold out geographies.
For most brands starting out, geo-based holdouts and platform-native RCTs are the practical entry points. They're methodologically defensible, they don't require you to instrument every touchpoint, and they produce numbers your CFO can actually engage with.
Where incrementality testing usually goes wrong
The failure modes are surprisingly consistent. First, tests are run at insufficient scale. If your campaign is small or your conversion volume is low, the statistical noise will drown any real signal — and you'll conclude either that nothing works or that everything works, both of which are wrong. Second, control groups aren't genuinely comparable: they differ in seasonality, competitive activity, or media weight from other channels. Third, tests are stopped too early, before the delayed effects of upper-funnel channels like DOOH and CTV can show up in the outcome window.
The fourth and most human failure mode is measuring the wrong outcome. If your only KPI is same-week online conversions, you'll systematically under-credit channels whose job is to shift consideration or drive future in-store visits. Incrementality only works when the outcome you're measuring is the outcome the campaign was actually designed to move.
Incrementality testing isn't a measurement upgrade — it's a mindset upgrade. Once a team starts thinking in terms of what would have happened anyway, every media conversation gets sharper. That's the shift we're helping brands make." — Eric Fan, CEO, Lumos
How to actually get started in the next 90 days
You don't need a rebuild of your measurement stack to run your first incrementality test. Pick one campaign — ideally something with meaningful budget and clear objectives — and design one geo-holdout study around it. Choose a channel where you genuinely suspect the attributed number is either flattering or unfair, so the result actually changes how you plan next quarter.
Agree the primary outcome up front, in writing. Lock the test and control markets before the campaign starts, not after. Run for long enough to capture both the exposure and the reasonable outcome window — a month is usually the floor, not the ceiling. And commit, before the results come in, to acting on what they show. The most expensive incrementality tests are the ones that get filed away because the answer was inconvenient.
Where DOOH sits in the incrementality conversation
Digital out-of-home is one of the channels where incrementality testing has quietly changed the narrative most. Because DOOH exposures are hard to tie to individual users, it has historically been under-credited in attribution models built around cookies and clicks. Well-designed geo-based holdouts consistently show DOOH driving meaningful lift on brand and behavioural outcomes — footfall, search interest, in-app engagement, and downstream conversions — that traditional attribution simply couldn't see.
At Lumos, this is the work we spend a lot of time on with clients: designing DOOH plans that are measurable by construction, not retrofitted with measurement after the fact. Increasingly, that means baking geo-holdouts, brand lift studies and privacy-safe audience matching into the plan from day one, so the incrementality question can actually be answered when the campaign wraps.
The bottom line for 2026
Attribution will still exist in 2026 — it's useful for pacing, optimisation, and quick reads. But the strategic decisions about where to put next year's budget increasingly need to be grounded in incrementality, because that's the only measurement discipline that answers the counterfactual. For omni-channel marketers, that shift isn't optional anymore; it's the price of making confident calls in an increasingly fragmented media environment.
If you're planning a DOOH or omni-channel campaign in the next quarter and want to build incrementality measurement into it from the start, we'd love to talk. Reach out to the Lumos team at spotlumos.com and we'll walk you through what a properly designed test looks like for your brand.
Lumos is a data-driven programmatic DOOH advertising platform operating across Australia, New Zealand and international markets. We help brands and agencies plan, activate and measure omni-channel campaigns with confidence — combining audience intelligence, mobility data, and measurement frameworks built for how media actually works today.
[SCHEMA_ORG]{"@context":"https://schema.org","@type":"Article","headline":"Incrementality Testing in Omni-channel Campaigns: A Beginner's Guide","description":"Incrementality Testing in Omni-channel Campaigns: A Beginner's Guide","image":"https://images.unsplash.com/photo-1590650046871-92c887180603?w=1200&q=80&fm=jpg","datePublished":"2026-07-15","dateModified":"2026-07-15","author":{"@type":"Organization","name":"Lumos","url":"https://spotlumos.com"},"publisher":{"@type":"Organization","name":"Lumos","url":"https://spotlumos.com","logo":{"@type":"ImageObject","url":"https://spotlumos.com/logo.png"}},"mainEntityOfPage":{"@type":"WebPage","@id":"https://spotlumos.com/au/resources/incrementality-testing-omnichannel-campaigns-beginners-guide-2026"},"keywords":"programmatic DOOH, digital out-of-home, audience intelligence, media planning","about":{"@type":"Thing","name":"Programmatic DOOH Advertising"},"isPartOf":{"@type":"WebSite","name":"Lumos","url":"https://spotlumos.com"}}
