Add up reach channel by channel and you will always feel like a genius. One person who saw your ad on linear TV, in a streaming app and in a social feed shows up three times in three reports — and once in the real world.
Incremental reach is the correction. As PPC Land explains in its breakdown of the metric, it counts the people a channel, publisher or placement reached that no other part of the same campaign reached. It is subtraction, not addition: that channel’s audience, minus everyone already exposed elsewhere on the plan.
Why the question is commercially awkward
Total reach asks how many people a buy touched. Incremental reach asks how many of them nothing else touched. That is a much harder sell.
A streaming service that mostly reaches households the broadcast schedule already covered is selling duplication. One that reaches new households is selling growth. On a media plan, both look identical until you deduplicate.
How the number is actually produced
Per the source, you need three things: an exposure record per channel at person or household level, a way to recognise the same person across those records, and an agreed universe to express percentages against.
The arithmetic is simple — and order dependent. Deduplicated reach across two channels minus the reach of the first channel gives the second channel’s incremental contribution. Flip the order and both numbers change, because the base channel absorbs all the overlap. Google’s Cross-Media Reach reporting sidesteps the argument by fixing the base explicitly, comparing digital video against a matching linear TV campaign using licensed third-party TV data — available only in selected countries, and in some cases only to allowlisted accounts.
Where one data source sees every channel, overlap is observed. Where it doesn’t, overlap is modelled. The oldest model — the Sainsbury formula, or random duplication, in routine use since the mid twentieth century — assumes exposures are independent. It needs nothing but individual reach figures, which is why it survives. It also overstates combined reach and understates frequency whenever audiences correlate, and they correlate constantly: heavy viewers of one medium tend to be heavy viewers of another. Academic comparisons found a demographically weighted variant more accurate in 82% of 9,680 within-medium cases and 77% of 968 across-media cases.
The seller problem
Here is the part performance marketers should tattoo somewhere: almost every published incremental reach figure comes from the party selling the inventory.
- Amazon released an Exclusive Reach Rate for streaming TV inventory on 23 December 2024, and reported advertisers using frequency group controls saving up to 26% of budget while gaining up to 21% incremental reach.
- TikTok said internal 2025 data put TopReach at 59% incremental reach versus TopView alone — while an iSpot-measured telecom case study returned 30%, a gap the company did not explain.
- TikTok Pulse analysis of 519 US campaigns run July–October 2025 claimed each 1% of budget delivers 2.4% exclusive reach.
- DAZN Media+ said streaming added 36% reach to AFL campaigns and 27% to NRL campaigns — without disclosing sample size, window or method.
Hybrid measurement is the commercial answer: census-level platform delivery data plus an independent consented panel that supplies the overlap census data can’t see. AudienceProject built its business on that model, published a white paper on 22 April 2026 audited by PwC and Mediasense among others, opened Polish reporting on 2 March 2026, and signed with Omnicom Media Italy on 27 May 2026.
Where it breaks
The identity layer is the real constraint. Truthset research for CIMM and Go Addressable, published 5 November 2025, examined close to a billion IP records from six providers and found probabilistic linkage weaker than its pricing implies. When recognition fails, overlap becomes invisible — and incremental reach gets overstated.
Definitions diverge too. Universes differ between vendors, household versus person counting produces different answers, and co-viewing adds people device-based counting can’t see (YouTube now reports co-viewed and unique reach separately).
What to do with it
Three practical moves for planners and growth teams:
1. Stop summing channel reach. Your plan overstates audience by exactly the size of its overlap.
2. Build a cost per incremental person. That is the only fair way to compare a CTV line against a social line without double counting.
3. Don’t confuse it with incrementality. Incrementality is causal — sales or conversions versus a counterfactual, established through holdouts and geo tests. Incremental reach counts people. A channel can add plenty of new humans who do absolutely nothing.
Ask sellers for the base channel, the universe, the method and the window. If they can’t produce all four, treat the percentage as marketing collateral.
Source: PPC Land
