Viant vs The Trade Desk: how does each DSP find the household?
Viant vs The Trade Desk comes down to two household layers: Viant's patented Household ID, and The Trade Desk's Identity Alliance, with UID2 as its deterministic ID framework. Both buy connected TV, and Viant says its Household ID can match 95% of U.S. adults. They differ on who can read the ID.
This is for whoever signs a CTV demand-side platform (DSP) contract and must defend it to finance. By the end you can name the dimension that should decide it. Our programmatic CTV services page covers how we plan CTV buys.
Our view: the identity model shapes whether a frequency cap holds across the screens in one home and whether exposure can be joined to your sales data. Your own CRM match rate is the test, so run both platforms on it before the contract, not after.
Treat it as a measurement question, which puts the work with analytics and insights. This post narrows our people-based DSP buyer's guide to the two names CTV buyers ask about most.
"Identity resolution transforms reach and frequency from estimates into accountable metrics."
Keith Petri, SVP, Identity and Data, Viant Technology, Viant blog, 20 January 2026
How do Viant Household ID and UID2 resolve a household?
Viant resolves households with its patented Household ID, which it calls deterministic; The Trade Desk offers UID2 as a deterministic ID framework and Identity Alliance as its household layer.
Viant's fiscal 2025 annual report describes Household ID as "our proprietary innovation that combines digital and personal identifiers into a normalized household profile". The same filing says the technology "cuts through more than 1.8 billion IP addresses and other identifiers and translates them into 125 million households." On 28 January 2025 Viant announced a partnership with TransUnion to enrich Household ID with its TruAudience identity data. The same 10-K says: "Our DSP has exclusive access to the HHID." Viant's 20 January 2026 blog adds that "interoperability with alternative ID frameworks like Unified ID 2.0 allows brands to maintain significant scale".
The Trade Desk's path has two layers. UID2, in the words of its official documentation, is "a framework that enables deterministic identity for advertising opportunities on the open internet". The UID2 FAQ says a UID2 is generated from an email address or phone number, and that a publisher can use single sign-on providers to retrieve the email address for this. The framework "does not have centralized storage for personal data". The household layer is Identity Alliance.
The Trade Desk describes Identity Alliance as "a coalition of the industry's leading identity graph solutions" that "unifies people, households, and devices". The same page says CTV and streaming device IDs "are associated with specific apps used by multiple people", and that identifiers are "grouped by deterministic and probabilistic associations". Viant's blog calls Household ID deterministic, "rooted in verified, real-world identifiers that are tied directly to people and devices, not probability scores". Neither publishes its matching logic, so ask each how a household is formed.
Viant vs The Trade Desk compared on eight identity dimensions
On the companies' own pages the sharpest differences are where the household is built, who can read the ID and what each says about markets.
| Dimension | Viant | The Trade Desk | Last verified |
|---|---|---|---|
| Identity source | Household ID: "digital and personal identifiers", including IP addresses, device IDs and CTV IFAs; TransUnion TruAudience enrichment announced 28 January 2025 | UID2, generated from an email address or phone number; Identity Alliance, a coalition of identity graphs | 27 Oct 2026 |
| Household resolution | "Normalized household profile", which Viant calls deterministic; 125 million households per the FY2025 10-K, no geography stated | Identity Alliance "unifies people, households, and devices", grouped by "deterministic and probabilistic associations"; UID2 is "more closely tied to single users" | 27 Oct 2026 |
| CTV frequency | Reach and frequency anchored to deterministic IDs (Viant blog) | "Manage frequency to advertise more efficiently" (identity solutions page) | 27 Oct 2026 |
| Measurement | Closed-loop measurement; incrementality testing named in the blog FAQ | Reporting on "frequency savings, lower-funnel conversions, and channel-level performance" | 27 Oct 2026 |
| Who can read the ID | 10-K: "Our DSP has exclusive access to the HHID"; blog: UID2 interoperability "allows brands to maintain significant scale" | UID2: Apache 2.0 licensed, open to participants who sign the agreement. Identity Alliance: called "interoperable"; export terms not on the page read | 27 Oct 2026 |
| Privacy posture | Privacy Center page lists a cookie opt-out, a non-cookie opt-out form, Global Privacy Control, NAI connected TV choices and a rights-request portal; it does not say how they apply to Household ID | No central store of personal data; Transparency and Control Portal opt-out | 27 Oct 2026 |
| Markets | "Currently, our focus is primarily on the U.S. market" (10-K) | 3,843 employees in 21 countries; UID2 operators hosted in Sydney and Singapore AWS regions | 27 Oct 2026 |
Read the "who can read the ID" row twice. Viant's 10-K says its DSP has exclusive access to the Household ID; its blog says UID2 interoperability lets brands keep scale. The UID2 FAQ says the same email or phone, sent at the same time, gives the same raw UID2, and "The result is the same, regardless of the Operator". Whether either household layer exports to a second DSP or clean room is a contract question. How often CTV apps hold an email or phone for a UID2 is our inference, not a company figure.
This is not legal advice. Hashed email or phone is still personal data under the privacy laws of our five markets, so settle consent and notice with counsel before any CRM file is uploaded.
Our note on the Privacy Sandbox wind-down covers why cookie replacements stopped being the question.
What do the 95% and 125 million figures actually measure?
Viant says its Household ID can match 95% of U.S. adults aged 18 and over; that is a stated capability against a reference file, not a share of CTV impressions you will reach.
Viant announced it with TransUnion on 28 January 2025 and its SVP of Identity and Data repeated it on 20 January 2026: "matching Viant's Household ID to 95% of all U.S. adults 18+." The base is people in a reference file. Your reach depends on which of those households stream on inventory Viant can buy, a number nobody has published.
The Trade Desk states no comparable coverage figure on the identity pages we read, so the chart ranks nothing. Its connected TV page does say advertisers can "reach more than 185 million households streaming premium content around the world"; that describes inventory, not identity resolution.
The 95% base is U.S. adults, and the 10-K says Viant's focus is primarily the U.S.; CTV was "46% of total advertiser spend on the platform" in the fourth quarter of 2025. The Trade Desk 10-K puts 37% of its workforce outside North America, and its 63% North America share includes Canada. Headcount is not identity coverage, and the 10-K's 125 million households carry no stated geography.
A match rate against a reference file, a household count and campaign reach are three different measures. Ask each vendor for one number: the share of your own CRM file that resolves to a household their DSP can bid on in your market, measured on the same day.
Which DSP fits which CTV brief, and what are the Viant alternatives?
Choose by where the audience comes from and where the money is spent: a U.S. household-frequency brief leans to Viant, a multi-market CRM brief leans to The Trade Desk.
These are Leapbuzz decision rules for a brand buying programmatic CTV through one primary DSP: starting positions for the pilot.
- U.S. reach and frequency across the screens in a home, little CRM data: start with Viant. Its 10-K says Household ID combines digital and personal identifiers; it does not describe a dependency on your logins.
- Large hashed-email CRM file, and you want the same audience in other DSPs or a clean room: start with The Trade Desk and UID2, which is open to any participant who signs the agreement; ask what Identity Alliance exports.
- Budget split across Singapore, Malaysia, Australia, the US and Canada: The Trade Desk reports staff in 21 countries and UID2 operators in Sydney and Singapore; neither states identity coverage in your market. Viant's 10-K says its focus is "primarily on the U.S. market", so ask in writing how Household ID resolves a household in your other markets.
- Sales measured on your own data: rank portability above match rate, and read our first-party data strategy notes first.
On "Viant alternatives": for household-level bidding, the alternative is another DSP's household layer. The Trade Desk's is Identity Alliance, with UID2 as its deterministic ID framework. Whichever you pick, ask in writing what can be exported.
Supply matters too: the deals you already hold belong in the same media integration review as the identity test, and our CTV PMP and programmatic guaranteed setup guide covers the deal side.
If you cannot say what share of your CRM file each DSP resolves to a household in your markets, the contract is a guess. Leapbuzz runs the side-by-side test and hands back a match-rate and reach-overlap map you can take to either vendor; ask us for it.
How do you run a four-week pilot of Viant vs The Trade Desk?
A four-week pilot gives both DSPs the same budget and frequency cap, then scores them on numbers you can check outside either platform.
| Week | Test | Metric | Measured by |
|---|---|---|---|
| 1 | CRM match rate | Share of the same file resolved to a household each DSP can bid on, per market | Each DSP, same file, same day |
| 2 to 3 | Reach and overlap | Unique households reached; overlap between the two | A clean room or neutral measurement partner |
| 2 to 3 | Frequency | Share of reached households above the agreed weekly cap | Independent exposure log, not the DSP report alone |
| 4 | Incrementality | Sales or site visits versus a CRM holdout group | Your own data |
| 4 | Portability | Can the audience and exposure IDs be exported and reused? | The contract, in writing |
Send the same hashed file to both DSPs in week one for the match test. Then split it once: a random holdout that goes to neither DSP, and one disjoint treatment cell per DSP. Buy the same publishers where you can, so reach reads identity, not supply.
Score = 0.25 × match rate + 0.20 × (100 minus share over cap) + 0.15 × portability score + 0.40 × cost score. Cost score = 150 minus the cost-per-incremental-household index (your plan = 100, so on plan scores 50, an index of 50 or lower scores 100, 150 or higher scores 0). Portability is your 0 to 100 reading of the written answers. Reach is reported, not scored. A gap under 5 points is a tie.
DSP A scores 58.5 and DSP B scores 59.0. A gap of 0.5 points is a tie: decide on supply, market coverage and contract terms.
Illustrative example, not client data and not a benchmark: DSP A matches more of the file, DSP B costs less per incremental household and exports more, and the pilot does not separate them. Change the weights if your brief is pure U.S. reach, and fix them in the pilot plan before results arrive.
Do this before the contract
Send both vendors the same four written questions this week, and sign nothing until the answers sit beside the week-one match rates.
- What share of our hashed CRM file resolves to a household you can bid on, in each market we buy, measured on the same day?
- Can our audience and the exposure IDs be exported to another DSP, a clean room or our measurement partner, and under what terms?
- How is a household cap counted when one CTV app is watched by several people, and can we audit it from a log?
- For Viant: how does Household ID resolve a household outside the U.S.? For The Trade Desk: which Identity Alliance graphs cover our markets?
Today: count the rows of the hashed CRM file you would upload, by market, and put that number in both RFP emails.