Definition

Incrementality measures the additional conversions caused by advertising that would not have happened without ads. It answers “Did this channel create new demand?” rather than crediting ads for organic or direct conversions.

Detailed Explanation

Methods include geo holdout tests (switch ads off in one city), PSA ghost ads, and marketing mix modeling (MMM). Incremental ROAS is often lower than platform ROAS.

Platform attribution over-credits retargeting and branded search; incrementality tests expose true lift before budget reallocation.

Design tests with statistical power — run 4–8 weeks, control for seasonality (avoid Dashain-only windows unless testing festival incrementality).

Nepal Context

Holdout tests in Pokhara vs Kathmandu work well for national brands with uneven spend. Cooperatives and MFIs use branch-level geo experiments before scaling NPR 500k+ monthly Meta budgets.

Smaller sample sizes lengthen test duration — partner with analytics teams to avoid underpowered conclusions.

Practical Examples

  1. Beginner: Pause brand Search for 2 weeks in one city, compare total leads (ads + organic) vs control city.

  2. Intermediate: Run Meta lift study on 15% holdout for lead campaigns, review incremental CPA.

  3. Advanced: Annual MMM with 2+ years of spend, weather, and festival dummy variables to guide channel mix.

Key Takeaways

  • Incrementality = true causal lift from marketing.
  • Platform ROAS ≠ incremental ROAS — test before scaling.
  • Geo holdouts are practical for Nepal multi-city brands.
  • Account for festivals and seasonality in test design.
  • Use results to set MER targets, not platform dashboards alone.

Common Mistakes

  1. Calling off tests early when results look noisy.
  2. Running holdouts during major festivals only.
  3. Ignoring organic brand search spike when pausing paid brand.