Why OOH Incrementality Is Undervalued in Most Media Mix Models

Every media mix model (MMM) we have seen that includes OOH allocates it less budget than the channel's actual sales contribution warrants. Not because outdoor advertising is ineffective - the evidence from incrementality testing says otherwise - but because MMMs are structurally biased toward channels that produce trackable last-click signals. OOH does not produce last-click signals. This is a modeling problem, not a channel problem.

How Media Mix Models Handle OOH (Poorly)

Media mix models use historical spend and sales data to estimate the marginal return from each channel. The accuracy of these estimates depends heavily on the variation in spend within the historical data - channels that have been turned on and off, varied in spending levels, and tested in different markets produce more reliable estimates than channels where spending has been consistent and continuous. Most brands that use OOH have run consistent flight-based campaigns without systematic variation, which means the MMM has limited data to isolate the OOH contribution from baseline sales trends.

Additionally, MMMs typically model OOH using TAB gross rating points (GRPs) as the input variable. GRPs are a trailing metric based on traffic counts, not a measure of audience engagement or reach quality. A well-placed billboard in a high-quality audience location and a poorly-located billboard on a low-traffic road will show similar GRP values if their traffic counts are similar. The model cannot distinguish between effective and ineffective OOH spend if the input variable is this crude.

The Attribution Gap Between OOH and Digital

Digital channels - search, social, display - generate user-level data: impressions served, clicks, sessions, conversions, and in some cases individual-level customer IDs. This data feeds attribution models that can estimate the contribution of specific ads to specific conversions. OOH generates placement-level data: estimated impressions per location per time period, with no individual-level identifiers and no conversion signal attached to the impression.

The consequence is that digital channels get credited for conversions they observe directly while OOH gets credited only for the statistical residual that the MMM cannot explain with other channels. When a consumer sees an OOH ad, searches for the brand, and converts through paid search, the paid search channel gets the click attribution. The OOH exposure that prompted the search gets nothing, even though it was the first touchpoint in the funnel.

This asymmetry systematically undervalues channels that work at the top of the funnel - awareness, consideration, intent - and overvalues channels that capture intent already generated by other channels. OOH is primarily a reach and awareness vehicle. It generates searches, drives direct traffic, and increases baseline brand recall. None of these effects are captured by last-click or even data-driven attribution models.

Incrementality Testing as the Correction

The correct methodology for measuring OOH incrementality is matched-market testing: run OOH in a set of test markets, withhold it from a set of carefully matched control markets, and measure the difference in sales or other outcome metrics between the two groups. This is the same methodology used for TV incrementality testing and is considered the gold standard for measuring channel contribution when individual-level attribution is not available.

Running a valid matched-market OOH test requires a minimum of 6 to 8 market pairs, a test period of at least 8 to 12 weeks, pre-period baseline measurement of 12 to 24 weeks, and outcome measurement that can be matched at the market level. For a brand with physical retail distribution, sales scanner data provides a clean market-level outcome. For direct-to-consumer brands, web analytics segmented by geography serve the same purpose. The OOH My Media platform is adding matched-market test design tools to help brands structure these experiments within their existing campaign infrastructure.

What Incrementality Testing Consistently Shows

Brands that have run rigorous OOH incrementality tests consistently find that the channel's contribution to sales is 30 to 50 percent higher than their MMM estimates. Procter and Gamble, which runs some of the most sophisticated marketing measurement programs of any advertiser, has published findings showing that OOH contributes to awareness and consideration metrics at rates that are not captured in their attribution data. Independent research from Nielsen has found that OOH advertising generates a higher online activity rate per advertising dollar spent than any other traditional media channel - approximately 4x the rate of television and 2x the rate of print.

The mechanism is search and direct site visits prompted by OOH exposure. When a consumer sees a compelling outdoor ad, the most common response is a search query - the consumer looks up the brand they saw. This search-generated traffic converts at high rates because the consumer has already been primed by the OOH exposure. But the conversion gets attributed to branded search, and the OOH ad that generated the search intent receives no credit. Correcting for this effect substantially increases the estimated return on OOH investment.

How to Build OOH Into Your Attribution Model More Accurately

There are three practical approaches to correcting for OOH undervaluation in attribution models. First, add branded search volume as a leading indicator of OOH effectiveness. Compare branded search volume in OOH markets versus non-OOH markets during and after campaign flights. Lift in branded search volume is a reliable proxy for OOH reach driving consideration. Second, use geo-based conversion lift measurement: compare conversion rates in markets where OOH is running against matched markets without OOH, with adequate pre-period baseline to control for existing trends. Third, integrate OOH delivery data into your MMM as a separate variable distinct from GRPs - using actual impressions delivered (from Geopath or OOH My Media's reporting) rather than estimated GRPs improves model accuracy because it reflects variation in actual delivery quality rather than just traffic count proxies.

None of these approaches fully solves the attribution gap, but each incrementally improves the quality of data available to the MMM. The goal is not perfect attribution - which does not exist for any channel - but a more accurate representation of OOH's true contribution so that budget allocation decisions reflect reality rather than a structural modeling bias. As we described in our article on real OOH attribution methodology, the measurement tools available today are substantially better than what existed five years ago, and they should be used to update legacy MMM frameworks that were built before this data existed.

The Practical Implication for Budget Allocation

If you are a brand with $5M+ in annual media spend and you have never run a rigorous incrementality test on your OOH allocation, you are making budget allocation decisions based on a model that is structurally underestimating OOH's contribution. The cost of running a matched-market incrementality test is typically $30,000 to $80,000 including research design, data collection, and analysis. The cost of misallocating $500K to $2M in annual media spend because your MMM is wrong is substantially higher. The test pays for itself if it surfaces any meaningful allocation adjustment.