Quick answer
Second-party data is another company's first-party data that you receive directly through a partnership or data-sharing agreement, such as a retailer sharing its loyalty audience with a brand. Its provenance and consent terms are known — that agreement separates it from third-party data.
Formula: Match rate = matched records ÷ partner records
What is second-party data?
Second-party data is another company's first-party data that you receive directly from it under a data-sharing agreement. Microsoft's Xandr glossary says it in one line: "Second-party data is another company's 1st party data." The records did not change — their owner did. A retailer's loyalty audience is first-party data for the retailer and second-party data for the brand that receives it.
One hop, one known collector: you can ask how records were collected, when they were refreshed and what the people in them were told. That is also the exposure. The value and the risk of second-party data both live in the sharing agreement — which fields you get, what you may do with them, and what the partner's users consented to.
Three partnership patterns recur:
- Retailer to brand. A retailer shares its loyalty audience with a brand that sells through it, to reach category buyers and suppress existing purchasers.
- Software company to reseller agency. Salesforce's example: the company shares its customer list with an agency that resells its products, which uses it to find similar prospects.
- Airline to hotel partner. An airline shares a booking audience with a hotel group in the same destinations — complementary, not competing, customer bases.
If your own collection is thin, start with a first-party data strategy.
Zero-, first-, second- and third-party data compared
The four types differ less in their fields than in who collected them and how far they travelled. The consent column decides what you may do next.
| Type (who collects it) | How you get it | Consent provenance | Typical cost model | Main risk |
|---|---|---|---|---|
| Zero-party — the visitor, volunteered to you | Preference centre, survey, quiz | Explicit, given to you for a stated purpose | Your own form and storage | Small scale; self-reported answers drift from behaviour |
| First-party — you, on your own properties | Your script or SDK, logins, server events | Your own notice and consent record | Cost of collection and storage | Coverage gaps: anonymous or non-consenting visitors |
| Second-party — one known partner | Direct agreement, co-marketing, clean room query | The partner's consent, limited to agreed purposes | Negotiated fee, revenue share or audience swap | Purpose mismatch; paying for overlap you already own |
| Third-party — aggregators, from many sources | Purchase from a broker or marketplace | Often unclear; not collected for your purpose | Per record, licence or CPM | Provenance and freshness you cannot audit |
Xandr's third-party definition is the contrast: aggregators buy first-party data from many owners, pool it and resell it, so the original collector is out of the conversation. With second-party data they are still in it. See zero-party data and third-party data.
How do you get second-party data?
Amplitude lists the channels: strategic partnerships with complementary customer bases, publisher collaborations, industry data-sharing groups and data clean rooms where partners analyse joined data without exposing raw rows. A deal usually starts as one of four conversations:
- Direct partnership. Two companies with adjacent audiences exchange or license a defined segment.
- Co-marketing. A joint webinar or report produces co-registered contacts; the registration form sets what each side may do with them.
- Retail media or publisher deal. You buy access to an audience they collected, usually activated inside their platform, not delivered as a file.
- Clean room or hashed-ID match. Both sides contribute hashed identifiers; only matched counts and aggregates come back.
Whatever the channel, the agreement is the product. Seven terms to settle before a file moves:
- Fields and format. Exact fields, hashed or raw, file or API, refresh cadence, and a preview before you pay — Salesforce recommends asking for one.
- Permitted purpose. Named uses only, such as suppression, lookalike seed or measurement; anything else needs written approval.
- Consent scope and provenance. What the partner's notice said, which purposes and regions it covered, when records were collected, who is controller.
- Retention and freshness. How long you may keep the file, the maximum record age, the re-delivery schedule.
- Deletion and data subject requests. How opt-outs and deletions travel both ways, and your deadline, for example 30 days.
- Onward transfer. No brokers, sub-processors or re-identification of hashed IDs without written consent.
- Audit and exit. Provenance checks, match-rate reporting, security and breach notification, and what happens to the data at termination.
Match rate and incremental reach: a worked example
Match rate = matched records ÷ partner records. Always name the denominator: the same join gives two percentages, depending on whose file you divide by.
Sample data: your identified base is 40,000 hashed email addresses, the partner offers a file of 25,000, the join returns 6,000 matches, and the partnership fee is $3,000.
| Metric | Computation | Result |
|---|---|---|
| Match rate, partner side | 6,000 ÷ 25,000 | 24% |
| Match rate, your side | 6,000 ÷ 40,000 | 15% |
| Incremental reach | 25,000 − 6,000 | 19,000 net-new people |
| Cost per matched record | $3,000 ÷ 6,000 | $0.50 |
| Cost per incremental record | $3,000 ÷ 19,000 | ≈ $0.16 |
The two cost figures differ about threefold because they answer different questions: the 6,000 matched records are what you can join, suppress or measure; the 19,000 unmatched records are new reach. For enrichment, $0.50 is your number; for reach, $0.16.
Price the alternative. Sample assumption, not a market rate: a third-party list at $0.60 per record matching at 10%. The same 6,000 matches need 6,000 ÷ 0.10 = 60,000 records, so 60,000 × $0.60 = $36,000 — twelve times the fee.
Decision rule, written before the negotiation:
- If the partner-side match rate is below 15% and you bought the file to enrich or suppress your base, most of it cannot be joined — renegotiate or walk.
- If you bought it for reach, set a ceiling from margin: $40 contribution margin per converted customer × 2% expected conversion = $0.80 per incremental record. $0.16 clears it; $1.20 does not.
- If the partner cannot state what their users consented to, no price is low enough.
When partner numbers don't match yours
The campaign ends: the partner reports 900 conversions from the shared audience, your goal report shows 750 for the same 30-day window. Sample data: agreement = 750 ÷ 900 = 83.3%, a gap of 150. Neither number is automatically wrong. Work through five causes in order:
- Conversion definition. Does the partner count any order, including unpaid carts, while your goal fires only on a completed purchase?
- Attribution window. Their 60-day click window against your 30-day lookback creates a gap by construction; TapCub's lookback defaults to 30 days, configurable from 1 to 90.
- Bot filtering. TapCub identifies bots and counts them separately by default, so crawler traffic never enters your denominator.
- Consent-tier coverage. EEA and UK visitors who withhold analytics consent are still recorded, but without a persistent ID they cannot be joined to the partner's list, which may still count them through its own tags.
- Deduplication. Orders versus people: three purchases by one customer are three conversions in one report, one in the other.
If the gap survives all five, ask for a timestamped sample of the disputed records and join on the agreed key — then write the definition, window and dedup rule into the agreement.
Consent and compliance: GDPR and CCPA
General information about what the texts say, not legal advice; have counsel review your own deal.
GDPR defines consent in Article 4(11):
"'consent' of the data subject means any freely given, specific, informed and unambiguous indication of the data subject's wishes by which he or she, by a statement or by a clear affirmative action, signifies agreement to the processing of personal data relating to him or her"
Article 6(1)(a) makes consent purpose-bound: processing is lawful where "the data subject has given consent to the processing of his or her personal data for one or more specific purposes". Canonical text: EUR-Lex. It blocked our automated request, so the wording was checked on the Article 4 and Article 6 mirrors at gdpr-info.eu.
CCPA is broader than most marketers expect. Civil Code §1798.140(ad)(1) defines sale as transferring personal information to a third party "for monetary or other valuable consideration". Section 1798.140(ah)(1) defines sharing as the same transfer "for cross-context behavioral advertising, whether or not for monetary or other valuable consideration, including transactions between a business and a third party for cross-context behavioral advertising for the benefit of a business in which no money is exchanged." Both quotes come from the CPPA's official statute PDF.
If-then rules that follow from those texts:
- If the partner's notice never named your purpose, their consent does not cover your processing: restrict the use, collect your own consent, or work with aggregates.
- If the file is hashed emails, it is still personal data and the same obligations apply.
- If you send your audience to a partner for cross-context behavioral advertising, that is sharing under CCPA even with no payment, so opt-outs must be applied before the file leaves.
- If the deal is an unpaid audience swap, "other valuable consideration" may still make it a sale.
- If you share only daily aggregates with no personal data, these definitions do not attach to that file — do not join it back to individuals.
See GDPR for the regulation itself.
Common misconceptions
- "Second-party data is third-party data with fewer steps." The difference is auditability: one known collector you can question about method, freshness and consent, versus a pooled broker file you cannot.
- "The partner collected it lawfully, so we can use it for anything." Consent is specific and informed (Article 4(11)) and tied to stated purposes (Article 6(1)(a)). A new purpose needs its own basis.
- "Hashed emails are anonymous." Article 4(5) defines pseudonymisation as processing where data "can no longer be attributed to a specific data subject without the use of additional information", kept separately. Hashing is pseudonymisation — and matching is the act of re-linking a record to a person.
- "A bigger file is a better deal." Above, 25,000 records delivered 19,000 net-new people at about $0.16 each; a larger file with a worse match rate can deliver fewer usable people for more money.
Preparing your side of the deal in TapCub
Run the overlap example on your own base before signing. Boundary first: TapCub collects first-party data from your own website, app and mini-program. It is not a data broker, marketplace or clean room: it does not sell, exchange or match audiences, so the join itself happens in your environment or a clean room. What TapCub can do is tell you what your side of the agreement actually contains.
- Know how big your matchable base really is. Only identified people can be joined. With identified mode on (Pro plan or above),
tapcub.loginafter sign-in puts people in All-platform analytics › Users › People with a day-by-day timeline. Anonymous visitors have no profile, so they were never part of the 40,000 denominator, and the match rate on your side should be computed against identified people only. - Check consent coverage before promising reach. In Configuration › Privacy center, the default collect tier is 2; the default EEA and UK rule is tier 2 with consent required, and a hit without analytics consent falls to tier 1 (the salt rotates daily, no persistent ID). Pageviews and custom events are still stored, but those visitors can never be joined to a partner file. Quote the matchable share in the agreement, not total traffic.
- Line up retention and deletion with the agreement. The same page holds the retention day counts and data subject requests (access, export, restrict processing, erasure), located by email, login ID, device ID or person ID. Detail past retention is already deleted, so do not promise records you no longer hold; when the partner forwards an erasure request, file it here.
- Share aggregates when aggregates are enough. For campaign reporting, a read-only link from the Public sharing tab under Configuration › Settings can open just the Overview and Sources modules, with a password and a day cap; the visitor list and sessions can never be shared that way. Daily aggregates contain no personal data, which is the safer file to send. Person-level rows leave only through an async export job in All-platform analytics › Data › Data export, and only within the fields and consent scope the agreement names.
- Reconcile on the same window. For the 750-versus-900 dispute, read conversions in Web analytics › Goals for the exact 30 days, split the campaign by UTM in Web analytics › Sources, and note that recognized bots are counted separately in Web analytics › Bots, so they are already out of your denominator. Then work through the five causes above with the partner.
The thesis in practice: TapCub can show what you hold, who consented to what, and what it adds up to; the value and the risk of the partner's data still sit in the agreement you sign.
Sources
- Microsoft Xandr. Online Advertising and Ad Tech Glossary — first-, second- and third-party definitions. Checked 2026-11-15.
- Amplitude. Understanding First-, Second-, Third-, and Zero-Party Data — second-party channels, partnerships and clean rooms. Checked 2026-11-15.
- GDPR, Regulation (EU) 2016/679, Article 4 and Article 6 — consent, pseudonymisation, lawful basis. Canonical EUR-Lex text blocked automated access; wording verified on the gdpr-info.eu mirror. Checked 2026-11-15.
- California Privacy Protection Agency. CCPA statute, Cal. Civ. Code §1798.140 — (ad) sale, (ah) sharing. Checked 2026-11-15.
- Salesforce. What is First Party, Second Party and Third Party Data? — software-company-to-agency example. Checked 2026-11-15.
Was this explanation clear?
We use your answers to fix and expand this page.
Couldn’t send. Please try again in a minute.
Thanks for the feedback. We’ll use it to improve this page.You’ve already sent feedback on this page. Thanks.

Solutions Engineer
Covers SDKs, API integrations, and data implementation, with an emphasis on validation, system interoperability, and troubleshooting.