Almost every claim a CTV vendor makes about targeting and attribution rests on ACR data. Automatic content recognition is the technology inside smart TVs that fingerprints what is on the screen, second by second, and reports it back to the manufacturer or a licensed data partner. It is how a platform can tell you that a household saw your competitor's linear spot, that your CTV campaign reached homes your broadcast buy missed, or that exposed households visited your website at 2.4 times the rate of unexposed ones.
ACR data in CTV advertising is useful, and for some jobs it is the only practical tool. It is also a panel, not a census, and the panel is thinnest exactly where luxury and high-consideration brands need it most: in the top few percent of households by wealth. This guide explains how ACR works, who holds the footprints, what the data can legitimately support, where it misleads, and how we use it inside plans for private aviation, wealth management, luxury real estate and premium automotive clients.
How does ACR data actually work?
A smart TV with ACR enabled samples the audio or video signal on the screen at a set interval, typically every one to several seconds, and converts the sample into a compact fingerprint. The fingerprint is matched against a reference library of linear programming, ads and streaming content. When a match is found, the TV logs the content ID, the timestamp, the input source and the device identifier, and sends the log to the manufacturer's servers or a licensed partner.
Three details determine what the data is worth:
- Input coverage. ACR reads whatever is displayed, so it captures linear cable, over-the-air, streaming apps, gaming consoles and HDMI-connected set-top boxes. That is its main advantage over app-level measurement, which only sees its own app.
- Opt-in. ACR is governed by a consent screen at TV setup. Opt-in rates are high in practice, generally above 80 percent by industry estimates, because the prompt is bundled with other setup consents, but the population that declines is not random.
- Identity. The TV reports a device ID. Turning that into a household that can be matched to a programmatic identifier, a CRM file or a wealth segment requires an identity graph and an onboarding step, and every step loses records.
The output is a household-level viewing log with reasonably precise timing. It answers "what was on this screen" very well. It answers "who was watching" only by inference.
Who owns the ACR footprints?
There is no single ACR dataset. Each footprint belongs to a manufacturer or operating system, and each licenses it differently. The table below summarizes the major U.S. sources as we work with them in 2026. Footprint figures are approximate and drawn from public statements and our own planning experience; each vendor reports its own numbers and definitions.
| Source | Where it comes from | Approx. U.S. opted-in footprint | How advertisers access it | Notes for luxury planners |
|---|---|---|---|---|
| Samba TV | Licensed ACR from multiple TV brands (Sony, Philips, Sharp, TCL and others) | 25–30M devices | Direct measurement contracts; segments in most DSPs | Multi-brand mix; strongest independent measurement offering |
| Vizio (Inscape) | Vizio smart TVs; Vizio Ads now under Walmart | 20–25M devices | Vizio Ads; segments via DSPs and clean rooms | Value-priced hardware skews mid-market; limited reach at top of wealth curve |
| Samsung Ads | Samsung Tizen TVs | 40M+ devices | Samsung Ads direct; Samsung DSP | Largest single-brand footprint; premium panel share better than average |
| LG Ad Solutions | LG webOS TVs | 20–25M devices | LG Ads direct; select DSP integrations | Strong OLED share means over-indexes in higher-income homes |
| Roku | Roku OS TVs and players (ACR on Roku-branded TVs) | 80M+ streaming households (ACR subset smaller) | Roku Advertising, OneView | Platform data richer than ACR alone; app-level plus ACR on newer sets |
| Comscore / iSpot / EDO | Aggregate licensed ACR plus set-top-box data | Panels of 20–40M households | Measurement subscriptions | Used for currency and ad-occurrence measurement, not targeting |
Two consequences follow. First, no footprint covers the whole market, and the footprints do not overlap cleanly, so "we measured 100 percent of your CTV campaign" is never literally true. Second, the composition of each footprint depends on who buys that brand of television, which is a wealth-correlated decision.
What ACR data in CTV advertising can legitimately do for a luxury brand
We use ACR data in four ways. Each has a defensible logic and a specific failure mode.
Linear-to-CTV retargeting and incremental reach
If a brand is still running linear television, ACR can identify households that saw the linear spot and either suppress them from the CTV buy (to extend reach) or add frequency against them (to reinforce). For luxury brands that run a small amount of linear, usually golf, tennis, Formula 1 or financial news, this is the highest-value application. In our plans, ACR-driven suppression typically shifts 15 to 30 percent of CTV impressions from already-reached households to unreached ones, with no change in CPM.
The failure mode is the footprint gap: households outside the panel are neither confirmed reached nor confirmed unreached, and the DSP treats them as eligible. In practice, that means the suppression is partial, and the "incremental reach" figure a vendor reports describes the panel, not the buy.
Competitive conquesting
ACR logs ad occurrences, not just programs, so a private aviation company can target households that were exposed to a competitor's spot within the past seven, fourteen or thirty days. This works. Response rates on conquest segments run 1.3 to 2.0 times the campaign average in our aviation and wealth-management work, because the competitor has already done the job of identifying an in-market household.
The failure mode is scale. A competitor with a modest linear or CTV plan generates a small exposed pool, and after identity matching and frequency limits the segment may be 40,000 to 150,000 households. That is fine for a premium brand with a $5,000-plus customer value. It is not a reach vehicle.
Cross-platform frequency management
Because ACR sees every input, it can tell you that a household received four exposures across Netflix, Peacock, a cable news channel and a YouTube session on the living-room set. Without it, each of those platforms counts frequency independently. For luxury brands, over-exposure is a brand-perception problem as much as a waste problem; a wealth-management prospect who sees the same spot eleven times in a week draws a conclusion about the firm's discipline. We use ACR-informed frequency data to set cross-platform caps, which we cover in more depth in our piece on CTV reach and frequency planning.
Exposure-based measurement
This is where most of the money and most of the confusion live. ACR provides an exposed group (households the panel saw receive the ad) and can be joined to outcomes: site visits via pixel and IP match, CRM conversions via clean room, or foot traffic via location partners. Comparing exposed to unexposed households produces the lift figures that vendors report.
The problem is that the unexposed group is not a control. Households that did not see a CTV ad differ systematically from those that did in viewing hours, platform mix, ad-tier subscription status and device brand. A naive exposed-versus-unexposed comparison overstates lift, often by two to four times against a randomized holdout in the tests we have run side by side. ACR exposure data is valuable as an input to a properly designed test; it is not a substitute for one. We explain the difference in our guide to holdout testing and our post on true incremental lift measurement.
Where ACR data in CTV advertising misleads luxury advertisers
Most of the ACR literature is written for mass-market brands, for whom a 30-million-household panel that skews slightly mid-market is an acceptable approximation. For brands whose buyer is in the top 2 to 5 percent of households by wealth, the same panel has structural blind spots.
The panel under-represents the top of the wealth curve
Television brand choice correlates with income. Value-tier brands with high ACR opt-in and aggressive data licensing over-index in the middle of the distribution. Premium sets, custom installations, projection systems and commercial-grade displays in second and third homes are either lightly represented or absent. In audits we have run for wealth-management and private aviation clients, matching a first-party customer file to major ACR footprints yields a match rate of 25 to 40 percent, against 45 to 60 percent for the same footprints on a general-population file. The households a luxury brand most wants to measure are the households ACR is least likely to see.
Multi-home households break the model
An ultra-high-net-worth household may have four or more televisions across two or three residences, several of them different brands. ACR will treat each set as a separate household. The result is fragmented exposure records, inflated reach counts and frequency that looks lower than it really is. Our post on the everywhere-millionaire audience goes into why multi-residence behavior confuses most household-level systems.
Opt-outs are not random
Households that decline ACR consent, install privacy-focused streaming devices, or route content through an Apple TV or a dedicated media player into a "dumb" display are disproportionately technically sophisticated and affluent. Apple TV does not provide ACR data to advertisers at all. In affluent ZIP codes, the share of streaming that flows through such devices is meaningfully higher than average, which removes a slice of the target audience from the panel entirely.
Content recognition is not audience recognition
ACR knows the screen showed a golf tournament. It does not know whether the person watching was the household principal, a teenager or a housekeeper. Co-viewing, which we discuss in our post on CTV co-viewing measurement, means that even a correctly identified household exposure may not have reached the decision-maker.
Benchmark ranges for ACR-driven tactics
These ranges come from luxury and high-consideration campaigns across private aviation, wealth management, luxury real estate and premium automotive over the past 24 months. They should be treated as planning guidance, not guarantees.
| Tactic | Typical segment size (households) | CPM premium vs. standard PMP | Performance vs. campaign average | Notes |
|---|---|---|---|---|
| Linear-exposed suppression | N/A (suppression, not target) | $0 | +15–30% unique reach | Requires brand's own linear schedule loaded |
| Competitive conquest (30-day) | 40K–150K | +$4 to +$10 | 1.3–2.0x site-visit rate | Diminishes quickly past 30-day window |
| Category-viewer targeting (e.g., golf, financial news) | 500K–3M | +$3 to +$8 | 1.1–1.4x | Broad; useful for prospecting layers |
| ACR-informed frequency capping | Full campaign | $0 to +$1 (data fee) | 8–15% reduction in wasted impressions | Value grows with number of platforms in plan |
| Exposure-based lift measurement | Panel-dependent | $15K–$60K study fee | Reports 20–60% "lift" typically; discount heavily | Not a substitute for a holdout |
How we use ACR data without being misled by it
The methodology below is what we apply on Stillwater plans. It treats ACR as an instrument with known bias, and corrects for the bias rather than pretending it away.
- Audit the footprint against the first-party file before committing budget. Match the client's customer and prospect file to the ACR-backed segment in a clean room or through the onboarding partner. Record the match rate. If it is under 30 percent, ACR-based tactics stay in the plan for suppression and conquest only, not for reach or measurement.
- Use ACR for suppression first, targeting second, measurement last. Suppression carries no CPM premium and the downside of a footprint gap is simply a slightly less efficient buy. Targeting carries a data fee that should be justified by conquest lift. Measurement carries the greatest risk of a wrong conclusion.
- Never report ACR exposed-versus-unexposed lift as incremental lift. Label it as observational. If the client wants an incrementality number, run a geo experiment or a household-level holdout, which we cover in our post on geo-experiment design. ACR exposure data can then be used to confirm that treatment-group households actually received the ad, which improves the test's precision.
- Weight the panel to the target, not to the population. Vendors weight ACR panels to U.S. Census demographics. For a luxury brand, re-weight to the wealth distribution of the target audience using the first-party match. This usually reduces reported reach and increases reported frequency, which is the honest direction.
- Reconcile reach across sources. Compare ACR-reported reach to platform-reported reach from Netflix, Prime Video, Disney+ and Peacock and to any independent measurement in the plan. Discrepancies above 20 percent indicate either a footprint gap or a household definition mismatch, and should be understood before anyone builds a story around the number.
- Set a decay window on conquest. A household that saw a competitor's private aviation spot 45 days ago is barely more valuable than a random affluent household. We use 14-day windows for aviation and automotive, 30 days for wealth management and real estate, and refresh segments weekly.
Common mistakes with ACR data in CTV advertising
- Treating "ACR-verified reach" as total reach. It is panel reach projected outward. Ask for the panel size, the projection method and the match rate to your file.
- Buying ACR-based "affluent viewer" segments as if they were wealth data. These segments infer affluence from programming (financial news, golf, luxury real estate shows). Viewing a program about wealth is not the same as having it. Our post on wealth-based audience segmentation covers what deterministic wealth data looks like.
- Accepting a vendor's exposed-versus-unexposed study as proof of ROI. It is proof of correlation between being in the panel, being exposed and converting.
- Ignoring Apple TV and non-ACR devices in affluent homes. Plan for a portion of the target audience that no ACR panel will ever see, and reach it through platform-direct PMP deals and first-party matched audiences instead.
- Running conquest against competitors who barely advertise. The segment will be too small to matter, and the data fee will not be recovered.
- Letting ACR frequency data override common sense. If ACR shows a frequency of two but the client's CEO is complaining about seeing the ad every night, the panel is missing a device in that home. Believe the CEO.
Where ACR fits in a luxury CTV plan
For a private aviation, wealth-management, real-estate or premium automotive brand, ACR data earns a place in the plan as a suppression layer, a conquest layer and a diagnostic input to measurement. It does not earn a place as the primary targeting engine, which should be built on deterministic first-party data and wealth-qualified segments, or as the primary measurement system, which should be built on randomized holdouts. Used that way, it improves the efficiency of a premium CTV buy by a measurable margin and never gets to author the ROI story on its own.
Stillwater Media builds CTV plans for luxury and high-consideration brands on Netflix, Disney+, Prime Video, Peacock and other premium streaming inventory, with ACR footprints audited against each client's first-party data before a dollar is committed. We take a limited number of new engagements each quarter. If your brand's customer is worth more than $5,000 and your CTV vendor has been reporting lift numbers you cannot reconcile, apply to work with us.
