Your media plan says the audience is five million people. Your buying stack can find maybe a third of them. That gap has a name: addressable reach.
A new explainer from PPC Land breaks down the metric that quietly governs frequency capping, attribution and CPM pricing, and why the number on your dashboard is almost always more flattering than the delivery behind it.
What addressable reach actually measures
Addressable reach is the count of people, households or devices inside a target audience that a buyer can identify and serve an ad to. It is expressed as a raw number or as a percentage of the target.
It is always smaller than the audience your brief describes, because recognising the same person across a browser, a phone and a TV requires a chain of matches, and records fall out at every link.
Adobe Audience Manager puts the arithmetic on paper: Segment Addressable Audience divided by Total Segment Population. Its worked example takes a segment of one million devices and reports 700,000 addressable, a 70 per cent match rate. Two caveats come with it: mobile app audiences are delivered but never counted, because the metric depends on cookie syncs apps do not perform, and Safari traffic drops out entirely thanks to Intelligent Tracking Prevention.
The multiplier problem
Here is the part planners underestimate. Addressable reach is not a lookup. It is a chain of sequential multipliers:
- Does an identifier exist in that environment at all?
- Does your identity graph resolve it?
- Does consent permit transmitting it?
- Does it survive the trip to the bidder?
- Do you win enough auctions to turn an addressable impression into a delivered one?
Score a respectable 80 per cent at each of those five steps and you address roughly a third of your target. That is the maths behind every campaign that hit its impression goal and missed its reach goal.
Not all matches are equal
In real-time bidding, identity travels in the eids array of the OpenRTB user object. OpenRTB 2.6, published in April 2022, added fields describing provenance rather than value: who inserted the identifier, who matched it, and the mm method code.
That method list is effectively a quality grade. A value of 3 means authentication such as an email login. A value of 5 means inference from something like an IP address. Both bid requests carry an identifier. Both count toward the same addressable reach figure. Only one of them is likely to be the right person.
The accuracy reckoning
Independent research has been brutal on the inferred end of that scale. CIMM and Go Addressable published Truthset work on 5 November 2025 examining nearly a billion IP records from six providers. IP-to-postal linkages were accurate 13 per cent of the time. IP-to-email hit 16 per cent. Providers agreed on the same household linkage in just 6.4 per cent of cases.
FreeWheel translated that into a warning that IP-based targeting can miss up to 87 per cent of households, while disclosing that IP accounts for 17 per cent of audience targeting on its own platform. Adstra and InterMedia Advertising reported in July 2026 that 23 per cent of residential IPs reached the intended geography.
Why buyers still pay for it
Because the incrementality case holds. The February 2024 Go Addressable and CIMM study of 145 addressable TV campaigns found addressable delivering around 40 per cent target audience reach regardless of brand penetration, holding between 37 and 45 per cent whether the brand sat under 20 per cent penetration or over 70. It also put addressable 20 to 30 per cent ahead of linear on efficiency once linear reach curves flatten.
Infrastructure caught up too. The Video Advertising Bureau’s June 2026 guide, citing an April 2025 EMARKETER forecast, reported 92 per cent of US pay TV households as addressable-enabled. A Go Addressable survey of 300 professionals in April 2026 found 78 per cent expecting addressable TV in 2026 upfront talks, up from 67 per cent a year before.
What to do with this
Three practical moves for growth and media teams.
Stop planning against universe size. Ask your DSP or partner for addressable reach against your specific audience and destination, not a blended platform total. IAB Europe’s transparency guidance tells buyers to request monthly active addressable users split by Safari, Firefox and Edge rather than one headline number.
Interrogate the match method, not the match rate. Ask what share of your addressable pool sits on authenticated signals versus IP inference. Vendor scale claims, such as Viant’s 95 per cent US household match rate via TransUnion or Yahoo’s 232 million ConnectID logged-in users, count recognisability, not correctness. Neither is independently audited.
Watch for definitional gaming. Addressable reach rises when the universe narrows. A stack that quietly excludes hard-to-match inventory reports a prettier ratio while reaching fewer humans.
And keep the terms straight. Addressable reach is a ceiling, not a delivery figure. Classical reach is net unduplicated audience actually exposed. Incremental reach is what a channel adds beyond another. High addressable reach with heavy overlap produces almost none of it.
Source: PPC Land
