Marginal review
A managed Bayesian MMM service that pairs channel analysis with incrementality testing.
Reviewed by iTechGuides Editors · Editorial team · Updated Oct 2026
Marginal is a managed media measurement service for brands that combines Bayesian media mix modeling with incrementality experiments. It is aimed at teams that want to understand channel contribution—including brand and upper-funnel investment—and use experiments to inform budget planning. The cloud service is available through the web and lists fit across small, mid-market, and enterprise organizations. Its mix of modeling and test design suits teams seeking an incrementality-calibrated view rather than a self-serve analytics workspace.
The modeling work includes channel contribution decomposition with uncertainty estimates, saturation curves, and budget reallocation scenarios. Marginal also designs geo-holdout and conversion-lift tests, then uses their results to calibrate subsequent model runs. Test-design power analysis is included, which makes experimental planning part of the offering rather than a separate add-on. Outputs arrive as a structured report and working file intended for use in existing planning workflows. This approach may suit teams that can incorporate delivered analyses into their own planning process; those seeking native ad-platform integrations should account for the requirement to provide exported spend and outcome data.
According to the vendor's pricing description, the first three model runs are free, with individually scoped pricing afterward. There is no free plan, and the service is not positioned as a self-serve subscription. The website describes refresh as ongoing and says subsequent runs incorporate new data, but does not specify a cadence. Marginal is a fit for teams prepared to supply data and engage with a managed measurement service, especially when incrementality testing matters alongside modeled channel contribution. Teams that need native connectors or a stated recurring refresh schedule should compare alternatives before choosing it.
Marginal pros and cons
- Where it wins
- Estimates channel contributions with uncertainty and saturation curves
- Uses geo-holdout and conversion-lift tests to calibrate later model runs
- Delivers structured reports and working files for planning workflows
- Where it doesn't
- Customers provide exported data; native ad-platform connectors are not offered
- Pricing is individually scoped after the first three model runs
- Model refresh cadence is not specified
Marginal fact sheet, pricing and score →
Advertiser disclosure: iTechGuides is reader-supported. Vendors can pay for top positions in our rankings and for a place on other products' pages, and we may earn a commission when you click some links. How we rank.
Last updated · How we research and update