CTV Retargeting for Luxury Brands: How Cross-Device Audience Extension Turns a Streaming Impression Into a 90-Day Conversation

CTV retargeting is the practice of taking the households that were served a connected TV ad on Prime Video, Disney+, Netflix, Hulu, Peacock, Max or a publisher app, resolving those households to the phones, tablets, laptops and audio devices inside them, and continuing the message on those devices through display, online video, streaming audio, social and search. For a luxury brand, it is the difference between a $45 CPM impression that lands once and disappears, and a $45 CPM impression that opens a 60 to 120 day sequence timed to a sales cycle that is at least that long.

Most luxury CTV plans stop at the living room. The ad runs, completion rate is 96 percent, the brand lift study comes back positive, and then nothing follows the exposed household to the screen where it will actually research a jet card, request a private tour, book a fitting or schedule a consultation. The exposed household has been told a story and then left alone. Cross-device audience extension closes that gap, and when it is built correctly, measured against a holdout, it is the highest-return addition available to a plan that already has premium CTV in it.

Why CTV retargeting matters more for luxury than for anyone else

A DTC apparel brand can afford to treat a CTV impression as a branding event because its purchase happens on a phone within days. A luxury brand cannot. The sales cycles we plan against run 30 to 60 days for premium DTC and luxury goods over $5,000, 45 to 120 days for private aviation memberships and jet cards, 90 to 180 days for wealth management and private banking relationships, and 120 to 365 days for luxury real estate, private club memberships and fractional ownership. Across all of them the first exposure is almost never the last touch, and the device that carries the first exposure is almost never the device that carries the conversion.

The numbers behind that are consistent. In Stillwater's clean-room matching across private aviation, wealth management and luxury real estate clients, between 78 and 91 percent of households that eventually converted after a CTV exposure did so on a device other than the television, and the median gap between first CTV exposure and first site visit was 11 to 19 days. Nielsen's cross-platform research puts the share of adults 18 and older who use a second screen while watching television at roughly 85 percent, and among households with incomes above $200,000 the figure is higher still, because those households own more devices per person. The television reaches the household; the conversion happens on the devices the household carries.

That gap is the whole case for CTV retargeting. Without it, the plan relies on the household remembering the brand for two to four weeks, searching for it unprompted and finding it before a competitor bids on the same search. With it, the brand is present on the second screen the same evening, on the laptop at the office the next week and in the car on the commute, at a cost per household that is a small fraction of the original CTV impression.

How cross-device audience extension actually works

The mechanics have changed substantially since 2023, and understanding them is the difference between a program that reaches 60 percent of exposed households and one that reaches 15 percent.

Capturing the exposure

Every CTV impression served through a demand-side platform such as The Trade Desk, Amazon DSP, Yahoo DSP, DV360 or StackAdapt generates an impression log with a device identifier (a Roku ID, Fire TV advertising ID, Samsung TIFA, LG LGUDID or Apple IDFA where available), a truncated or full household IP address depending on the supply source, a timestamp and the deal ID. Publisher-direct buys on Disney+, Netflix and Prime Video can also generate exposure logs, but they arrive through a clean room rather than a DSP log, which changes the matching route described below.

Resolving the household

The exposure is then matched to the household's other devices through one of four routes, each with a different coverage rate and a different privacy posture.

The first route is the DSP's own household graph. The Trade Desk's graph, Amazon's device graph and the graphs licensed by StackAdapt and Yahoo from Tapad, Experian or LiveRamp connect devices seen on the same residential IP over repeated sessions, with additional deterministic signals where logins exist. Coverage of exposed households typically runs 55 to 75 percent in the United States, and the graph is the fastest route because the audience is built inside the DSP where the extension media will run.

The second route is Unified ID 2.0 and equivalent authenticated identifiers. Where a streaming platform passes UID2, EUID or RampID on the impression and the same identifier appears on the household's web and app traffic, the match is deterministic and highly accurate, but coverage is limited by authentication: 20 to 40 percent of exposed impressions currently carry a UID2 on the open supply we buy for luxury clients, higher on Disney, Paramount and Warner Bros. Discovery inventory where the platforms have adopted it.

The third route is clean-room matching for walled gardens. Prime Video exposures are matched to Amazon's device graph inside Amazon Marketing Cloud, and the resulting exposed-household audience can be activated in Amazon DSP for display, online video, Twitch, Fire TV and Alexa audio. Disney's Compass, NBCUniversal's One Platform and Netflix's LiveRamp and Snowflake clean-room integrations work the same way for their inventory. Match rates inside these environments are the highest available, often 70 to 85 percent of exposed households, but activation is limited to the platform's own DSP or a small set of approved partners.

The fourth route is IP-based extension, in which the household IP from the CTV impression is used directly to target display and online video on the same IP. It is the simplest and the least precise, since residential IPs rotate on a schedule of days to months depending on the provider and shared IPs in multi-dwelling buildings inflate the audience. We use it only as a coverage backstop with a 14-day maximum window.

Building the audience

The output of any of these routes is an exposed-household audience segmented by what the household was shown, how many times, and when. That segmentation is what separates CTV retargeting from generic retargeting: the household is not being followed because it visited a page, it is being followed because it received a specific message, which means the follow-up can be written to continue that message.

CTV retargeting channel sequence for luxury brands

The channels the exposed household is extended into, and the order, should follow the sales cycle rather than the media buyer's preference. The table below reflects the sequence we run for engagements with a 60 to 120 day cycle, with the benchmark ranges observed across Stillwater luxury accounts.

Two elements of that sequence are non-negotiable. The first is that search sits underneath every stage: the CTV impression manufactures branded and category search intent, and if the brand is not bidding on the terms it just created, the retargeting program is generating demand for whoever is. Our clean-room matches show branded search volume rising 18 to 40 percent among exposed households in the 14 days following a CTV flight. The second is that stage 5 is a second CTV flight, not a repeat of the first: the exposed non-converters at day 60 need a different creative, usually the proof or offer message, and running the same launch spot to them again is the most common form of wasted frequency we see in audits.

Retargeting windows and frequency for long sales cycles

Retargeting practice built for e-commerce uses 7 to 30 day windows and heavy frequency, and it fails luxury brands in both directions: too short to cover the cycle and too intense for an audience that is used to being courted, not chased.

The window should equal the median sales cycle plus one standard deviation, which in practice means 60 days for premium DTC, 90 to 120 days for private aviation and wealth management, and 120 to 180 days for real estate and club memberships. Within that window, the cumulative cross-channel frequency that produces the best measured lift in our tests is 12 to 20 exposures over the full period, front-loaded in the first 14 days and tapering. Cumulative frequency above 30 in a 90 day window shows flat or negative incremental lift for luxury categories, and above 45 we see measurable negative brand sentiment in lift surveys.

Frequency has to be managed at the household, not the device or the channel. A household that receives 3 display impressions, 2 audio impressions, 2 social impressions and 1 YouTube impression in a day has been served 8 times, and only a DSP or clean room that reconciles across those channels will know it. Where the extension runs across multiple platforms, we set per-channel caps at roughly 40 percent of the intended household cap, because cross-platform reconciliation is never complete.

Creative logic: what to say on the second screen

The failure mode of most CTV retargeting is showing the exposed household a cropped version of the television spot on a phone. The household already saw the spot. The second-screen creative has a different job at each stage, and the table below is the framework we brief against.

The inform stage carries the most weight and the least attention from creative teams. A private aviation prospect who has seen a 30-second brand spot on Netflix does not need another brand spot; they need to know whether the product is a jet card, a fractional share or on-demand charter, what the hourly rate is and what the minimum commitment looks like. Delivering that on native and long-form display in premium editorial, in the exposed household's evening browsing window, is where we see the largest lift in qualified inquiry rate.

What CTV retargeting delivers: benchmarks

The value of cross-device audience extension is only real if it is measured against the alternative of not doing it, which means an exposed-household holdout: a randomly selected 10 to 20 percent of CTV-exposed households that receive the CTV impression but no extension media. The table below summarizes the measured deltas across Stillwater luxury accounts running that design.

The last row is the important one for budgeting. Extension media typically costs a quarter to a third of the CTV budget it follows, and it lifts the return on that CTV budget by two-thirds or more, which is why the right way to think about CTV retargeting is as a multiplier on the CTV investment rather than a separate line.

Platform capabilities for CTV retargeting

Not every buying platform can build an exposure-based audience and extend it, and the ones that can differ in coverage and where the extension can run.

For most luxury engagements we run the extension through The Trade Desk for publisher-direct CTV and Amazon DSP for Prime Video, and reconcile household frequency between the two through a LiveRamp or Snowflake clean room. Running a third platform for extension rarely adds coverage worth the loss of frequency control.

Privacy and consent constraints

Cross-device audience extension is legal and routine in the United States when it uses pseudonymous identifiers and honors opt-outs, but the constraints tighten annually and a luxury brand's exposure to reputational risk is higher than most.

The practical rules are these. Extension audiences must respect the Global Privacy Control and state-level opt-out signals under the California, Colorado, Connecticut, Virginia, Texas, Oregon and other state privacy laws; the DSPs handle this at the bid level, but the advertiser's own site must honor the same signals or the search and site-side stages break the chain. Sensitive categories are excluded: wealth management and private banking extension cannot use health, precise financial or protected-class inference, and audiences should be built on exposure and behavior, not on inferred net worth passed through third-party segments. Frequency caps and windows are also a privacy control, not only a media one: a household that sees a brand 40 times across four devices in a month feels surveilled, and luxury buyers say so in lift surveys. Finally, extension into email and direct mail from a CTV exposure requires a deterministic match to a consented first-party record; matching an anonymous exposed household to a mailing address through a data broker is a line we do not cross for clients.

Common mistakes in CTV retargeting

The mistakes we correct most often when auditing luxury plans are consistent enough to list.

• Treating CTV as terminal. The plan buys premium CTV, reports completion rate and lift, and never builds the exposed-household audience. The most expensive impression on the plan is left to fade.
• Extending only into display. Display alone reaches the exposed household on one device type at one moment; the sequence above works because audio reaches the commute, social reaches the evening, and search catches the intent.
• Cropping the TV spot for the phone. The reinforce stage can echo the spot; the inform stage must not. Households that see the same message on every screen show lower inquiry rates than households that see a progression.
• E-commerce windows and frequency. A 14-day window and 40 impressions a week is built for a sneaker purchase and reads as harassment to a prospect deciding on a $250,000 membership.
• No search coverage. The retargeting budget manufactures branded search and a competitor's brand campaign harvests it.
• Measuring with platform attribution. Every DSP will claim the conversions of exposed households whether or not the extension caused them. Only an exposed-household holdout shows the extension's incremental contribution.
• Ignoring the second CTV flight. Exposed non-converters at day 60 are the warmest cold audience on the plan; a fresh CTV creative to that group is routinely the highest-lift line on a luxury schedule.
• Building the audience per platform. A Prime Video exposure audience in Amazon DSP and a Disney exposure audience in The Trade Desk with no cross-platform reconciliation produces double the intended household frequency for households reached on both.

How to implement CTV retargeting: a sequence

• Map the sales cycle by stage, with median and standard deviation from CRM data, to set the retargeting window and the stage timings.
• Choose the DSP pairing that covers the CTV supply on the plan, typically The Trade Desk plus Amazon DSP for luxury, and confirm household-graph coverage estimates for the target geography.
• Set up exposure logging on every CTV line, including publisher-direct PMPs and PG deals, and confirm deal IDs flow through to the audience builder.
• Build exposed-household audiences segmented by creative, frequency band and exposure date, and refresh them daily.
• Carve a 10 to 20 percent exposed-household holdout that receives no extension media, before any extension spend starts.
• Brief stage-specific creative for reinforce, inform, prompt and re-engage, and produce the second CTV flight before launch rather than after.
• Launch search coverage for brand and category terms with the exposed-household signal applied as an audience bid adjustment.
• Set household-level frequency caps in a clean room and per-channel caps at roughly 40 percent of the household cap.
• Read exposed-versus-holdout site visit, qualified inquiry and closed revenue at 30, 60 and 90 days, and reallocate extension budget by stage and channel against measured lift rather than platform-attributed conversions.
• Feed CRM stage outcomes back to the DSPs through offline conversion tracking so the extension lines optimize toward qualified inquiries and closed sales rather than clicks.

Where Stillwater Media fits

Stillwater Media plans and buys premium CTV on Prime Video, Disney+, Netflix and the publisher-direct marketplaces for luxury and high-consideration brands, and we treat every one of those impressions as the opening of a sequence rather than the end of one. We build the exposed-household audiences, run the cross-device extension across display, online video, streaming audio, social and search through The Trade Desk and Amazon DSP, reconcile household frequency in a clean room, and measure the extension against an exposed-household holdout so the lift we report is lift the media caused. We take a limited number of new engagements each quarter. If your CTV plan stops at the living room, [apply to work with us](https://stillwatermedia.io/apply).

---

*Stillwater Media is a selective performance media agency for luxury and high-consideration brands, based in Charlotte, North Carolina and working nationally. We plan and buy premium CTV, programmatic, digital out-of-home, streaming audio and YouTube Select for clients including JetLinx, W Hotels, PXG, FLY Exclusive and Financial Independence Group, and we measure everything against holdouts rather than platform-reported lift. Signal. Strategy. Scale.*