What Is Crowd Anonymity?

Crowd anonymity is Apple's control over how much a SKAdNetwork postback reveals. A data tier sets which campaign digits and conversion values can be included.

Crowd anonymity is Apple's privacy control on SKAdNetwork postbacks. Apple assigns each attributed install a data tier and then removes fields that would make the crowd behind that postback too small to hide in. The postback still has no user or device identifier. The tier decides how much campaign and conversion detail is left.

Advertisers do not choose the tier, and they do not receive the formula.

How do the tiers work?

Apple describes four data tiers for SKAN 4. The first postback (conversion window covering days 0 to 2) can carry more than the second and third.

TierFirst postbackLater postbacks
0Two-digit source-identifier. No conversion value.Not sent
1Two-digit source-identifier and a coarse conversion valueTwo-digit source identifier and a coarse value
2A two-, three-, or four-digit source-identifier and a fine conversion-valueTwo-digit source identifier and a coarse value
3The tier 2 fields, plus source-app-id or source-domainTwo-digit source identifier and a coarse value

The crowd Apple considers includes the source app or source domain, the advertised app, the install country, and the hierarchical source identifier. A source identifier that splits one campaign into many rare values can shrink the crowd and drop the tier. Apple's own diagrams of the cutoff are illustrative. There is no published rule such as a fixed number of installs per day.

postback-sequence-index tells you which window produced the postback (0, 1, or 2). A null conversion-value on sequence 0 at tier 0 is expected. A null fine value on sequence 1 or 2 is also expected, because those postbacks use coarse-conversion-value only. See conversion value for the windows and the update API.

A concrete example

An ad network writes a four-digit source-identifier. The first two digits are the campaign, and the last two are an ad set.

A postback comes back with source-identifier of 14 (two digits), no conversion-value, and no second postback. That matches tier 0. The ad set digits were not returned. Reporting those installs at ad set level would invent a split Apple did not send.

Another postback for a larger campaign returns source-identifier 1407, conversion-value 20, and postback-sequence-index 0. That is a finer first postback. A later postback for the same pattern may still arrive as source-identifier 14 and coarse-conversion-value high. The report should show the digits that were present, not pad 14 back out to 1407.

What are the limits?

You cannot buy your way to tier 3 with a setting. Volume, overlap, and how finely you slice the source identifier all feed a calculation Apple does not publish. Quoting a threshold you found in a diagram will not match every campaign.

Missing fields are easy to misread as tracking bugs. Check the ad network IDs, the signature, the conversion update, and the postback endpoint first. If those are healthy and the value is still null, treat it as the tier.

Crowd anonymity does not restore a person-level path, and it is not a substitute for App Tracking Transparency. ATT governs permission for cross-company tracking and the Identifier for Advertisers. SKAN postbacks are available without that permission because they are aggregate. They also cannot answer questions that need user-level joins. When those questions matter, teams use aggregate methods such as the ones in what's left when you can't track anyone, alongside the postbacks, not instead of checking them.

Modeled metrics from an ad network are a third column. Do not add them to raw postback counts.

How Hardal handles it

Hardal does not change Apple's tier and does not fill redacted fields. Store the postback as it arrived, including null conversion-value and a short source-identifier, next to the first-party app events for the same period. A dashboard that replaces null with zero will show a conversion that did not happen.

The SKAdNetwork guide lists every postback field and the tier table with Apple's documentation links.

Related terms: SKAdNetwork, Conversion value, App Tracking Transparency

Frequently asked questions

What are the SKAdNetwork crowd anonymity tiers?

Apple assigns a data tier from 0 to 3. Tier 0 returns a two-digit source identifier and no conversion value, and it does not send the later postbacks. Higher tiers can add a coarse or fine conversion value and, at tier 3, the source app or domain.

Does Apple publish the install count that unlocks a tier?

No. Apple's diagrams are illustrative. The tier depends on the crowd around the source app or domain, the advertised app, the country, and the source identifier. There is no public cutoff you can quote as a guarantee.

Is crowd anonymity the same as App Tracking Transparency?

No. App Tracking Transparency asks permission before an app tracks people across other companies' apps and websites or reads the advertising identifier. Crowd anonymity limits the fields on a SKAdNetwork postback, which does not contain a user or device identifier.

What should a report do with a missing conversion value?

Keep the null. Do not store it as zero unless zero is a real mapped value and the field was actually present. A missing field is a privacy outcome until you have ruled out an integration fault.

Hardal is the first-party data and agentic analytics platform.

We help enterprises send accurate signals to their ad platforms and grow revenue, using server-side data from their websites, mobile apps and AI agents.