← All playbooks·Cross-media measurement
Playbooks · Cross-media measurement
Deduplicate reach across channels into net reach
This job turns three separate reach claims, linear, CTV and streaming, into one deduplicated net reach number for the plan. It feeds the decision on whether the channel mix needs rebalancing to lift net reach or whether a channel is now just stacking frequency on people already reached.
What you need first
- Target universe definition and size (e.g. adults 18-54) from the census or the currency provider's universe file
- Single-channel reach for linear, CTV and streaming, each from its own source: linear panel/return-path data, CTV platform or ad-server delivery, streaming panel or platform reporting
- Pairwise and triple overlap matrix across the three channels from the market's single-source cross-media panel or hybrid currency (e.g. Kantar, GfK, national JIC)
- A common campaign time window applied to all three reach figures
- Currency provider's methodology notes on coverage gaps or adjustments (co-viewing, non-addressable inventory) if the raw numbers need correcting before use
The procedure
- Confirm all three channel reach figures share the same universe, audience definition and time window; convert any that don't before touching the overlap matrix
- Pull single-channel reach for linear, CTV and streaming from their respective sources, expressed as both persons and % of universe
- Pull the three pairwise overlaps and the one triple overlap from the cross-media panel or currency
- Compute net reach via inclusion-exclusion: sum the three single-channel reaches, subtract the three pairwise overlaps, add back the triple overlap
- Convert the net reach figure to a percentage of the universe
- Compare net reach % against the naive sum of the three single-channel reach %s to quantify duplication removed
- Recompute net reach with each channel dropped in turn to find each channel's marginal contribution, and flag any channel whose marginal lift is small relative to its budget share
Worked through with numbers
Universe: adults 18-54 in a mid-size European market, 6,000,000. Single-channel reach: linear 2,400,000 (40%), CTV 1,800,000 (30%), streaming 1,500,000 (25%); naive sum 5,700,000 (95%). Overlap matrix from the panel: linear∩CTV 720,000 (12%), linear∩streaming 600,000 (10%), CTV∩streaming 900,000 (15%); triple overlap 360,000 (6%). Net reach = 2,400,000+1,800,000+1,500,000-720,000-600,000-900,000+360,000 = 3,840,000, or 64% of universe. Dropping CTV: net(linear,streaming) = 2,400,000+1,500,000-600,000 = 3,300,000 (55%), so CTV's marginal contribution is 9 points. Read it as: net reach of 64% against a naive sum of 95% means close to a third of the combined reach claim was duplicate exposure, and CTV is adding only 9 net points despite carrying 30% reach on its own, which is the number to weigh against its budget share before adding more CTV weight.
Where it goes wrong
- Summing single-channel reach percentages as if they don't overlap inflates net reach by the full duplication amount; subtract the actual overlap matrix instead of applying a rough discount
- Applying one flat overlap percentage across every channel pair misstates net reach because linear-CTV overlap and CTV-streaming overlap come from different viewing behaviors; use pair-specific figures from the panel
- Dropping the triple-overlap add-back term undercounts net reach; inclusion-exclusion needs the sum, the three pairwise subtractions, and the triple add-back, not just the subtractions
- Mixing reach numbers built on different audience definitions or campaign windows before deduplicating produces a net reach figure that isn't real; match universe and window across all three channels first
How to know it is right
Net reach must land strictly between the largest single-channel reach (40%) and the naive sum of all three (95%); recompute the inclusion-exclusion arithmetic once more from the same overlap matrix and confirm it matches before passing the number on.
Terms used