The hash in one line
hash = base64( sha256( utf8( standardize(value) ) ) )- Standard Base64 with padding, not hex and not the URL-safe alphabet.
- No salt, no pepper, no HMAC. The same value always gives the same hash, at DROP and at every broker.
- Lowercasing is culture-independent. Never apply Turkish or other locale rules.
- If you already store hashes made with exactly these rules, compare them as they are; do not hash them again.
The rules below follow CalPrivacy's technical documentation, version 1.3.0 of 19 November 2026. Version 1.3.0 tightened several of them (NFKC for names, phone extensions, Unicode hyphens in ZIP+4, VIN length), so a routine written against an earlier version may miss matches.
Rules for each list
| List or field | Standardize | Example |
|---|---|---|
| Remove all whitespace, including inside the address; lowercase. Keep dots, plus signs and every other character. | [email protected] → [email protected] | |
| Phone | Remove any extension and its digits first. Keep digits only, then the last 10 (or all, if fewer). Do not add a country code stored elsewhere. | +1(212)555-0186 ext. 123 → 2125550186 |
| First and last name | NFKC, remove control and format characters, lowercase, transliterate, fold accents, keep letters and digits only. Each field separately. | O'Connor → oconnor |
| Date of birth | YYYYMMDD, four-digit year, no separators. | July 4, 1776 → 17760704 |
| ZIP | Drop a ZIP+4 suffix; keep letters and digits; lowercase; strip leading zeros; keep the first five characters. | 91790-3771 → 91790 |
| MAID | Keep hex characters, lowercase. Must be exactly 32 characters. | a3f1c2d4-5678-90ab-… → a3f1c2d4567890ab… |
| VIN | Trim, remove spaces and hyphens, lowercase. Other punctuation makes the value invalid. Must be 5 to 17 letters or digits. | 1HGCM82633A004352 → 1hgcm82633a004352 |
| CTVID | Keep letters and decimal digits, lowercase. Must be 8 to 32 characters. | SmartTV_9F8E7D6C → smarttv9f8e7d6c |
A value that fails a length or character rule (a 30-character MAID, a VIN with a slash) has no valid hash. Skip it rather than hashing whatever is left.
Two details trip people up. Email standardization does not fold Gmail-style dots or plus tags, so [email protected] and [email protected] are different hashes. Phone standardization takes the last ten digits, so a non-US number keeps its last ten digits and loses the rest: +84(90)123 4567 becomes 4901234567.
Names in detail
Names follow the longest rule, in this order:
- Unicode NFKC normalization (full-width letters become ordinary ones).
- Remove control and format characters, such as zero-width joiners.
- Lowercase, culture-independent.
- Apply DROP's transliteration table for Greek, Cyrillic and special Latin letters. For example
θ→th,щ→shch,ß→ss,ø→o,ł→l; the Cyrillic hard and soft signs are removed. - Convert accented Latin letters to plain ASCII:
María José→mariajose. - Keep only letters and decimal digits. Spaces, hyphens and apostrophes go.
Scripts without a transliteration, such as Chinese, Hebrew or Arabic, are hashed in their own script: 张 stays 张. Compound names stay in their own field (Juan Pablo → juanpablo); do not move a middle name into the last name.
Apply the transliteration table before the generic accent folding. Letters such as ø, ł and ß do not decompose under Unicode normalization, so folding first would leave them in place and the hash would differ.
ZIP codes in detail
The ZIP rule is the one most often coded wrongly, because the order of steps matters.
- Trim surrounding whitespace.
- If the whole value is a ZIP+4 (five digits, a hyphen or plus sign, four digits), drop the separator and the last four digits. Unicode hyphens and the full-width plus count as separators.
- Keep letters and decimal digits; lowercase.
- Strip leading zeros. A value of all zeros becomes
0. - Keep the first five characters.
| Input | Standardized | Why |
|---|---|---|
91790-3771 | 91790 | ZIP+4 suffix dropped |
00712345 | 71234 | Zeros stripped first, then five characters kept |
00300-9999 | 300 | ZIP+4 dropped, then zeros stripped |
M1B 1A1 | m1b1a | Non-US postal codes are allowed |
Note that stripping zeros means a ZIP such as 02134 is hashed as 2134. The DROP regulations describe the step as keeping “the first 5 characters”, and the November 2026 proposal reads as if it trims before removing zeros. DROP builds its hashes with the technical specification, so follow the specification and watch for the final rule text.
Composite hashes
NDZ and NameVIN combine several fields:
- NDZ = first name hash + last name hash + date-of-birth hash + ZIP hash
- NameVIN = first name hash + last name hash + VIN hash
Hash each standardized field on its own, join the four (or three) Base64 strings exactly as they are, each with its trailing = and no separator, then hash that joined string. Do not decode to bytes before joining. Every field must be present. Here is CalPrivacy's webinar example:
first "Jane" -> jane -> gfj23eiDZfOSh5bseqU/coILBtuGZPX+dqfrE+JFRqI=
last "Doe" -> doe -> eZ75KhGvkY4/t0HfQpNPO1aO0tk6wd908bjUGieTKm8=
dob "7/22/1990" -> 19900722 -> 0oJHwSqomfyhaatmj6yKGJfv+q7SrChsjxL7siNkmM4=
zip "94103-3857" -> 94103 -> kdwlGeqYxQAs8gkeahK3cur9zp3KYY5ibX07gnU2F4k=
NDZ = base64(sha256( first + last + dob + zip ))
= X7qUgPCyldidyYW0lpVyXUQ1nfrZdsb3LDH02zC0rfQ=People with several values
Consumers can give DROP more than one value for every identifier except date of birth. CalPrivacy's webinar showed DROP building every combination of first name, last name and ZIP with the single birth date, so Jane and Jane-Ann Doe with two ZIP codes produce four NDZ hashes.
Do the same on your side: for each person, hash every combination of the names and ZIP codes you hold, including past addresses. The regulations ask brokers to apply any standardization they know will make a match more likely (11 CCR § 7613(a)(1)(A)).
Test vectors
Run these through your routine before processing a real batch. Every hash below was recomputed from the standardized value. Two printed values in CalPrivacy's material are wrong, and the correct hashes are shown; they are marked.
| List | Input | Expected hash |
|---|---|---|
[email protected] | KA18MT/ph6IHYjzT9zwETySDQyvSh87YuoSBpOQtkhE= | |
[email protected] | mKDnDvwF2inxrKcK1hJN2TRkxPfL6kzNNTtU12eH8Bw= | |
[email protected] | OQhS9Ardy54oaPGAtHHvQxWjQGPxRUR9c/UfXlrR70c=A webinar slide prints a different hash; this is the hash of the standardized value. | |
| Phone | +1(415)555-9317 | vGM7y5n+hBXRSEAklhHDPCbysyNgYTmXdMcagGUOY8E= |
| Phone | +84(90)123 4567 | ptzVkgbv9DonwvPCHmXmJ2SEOaolSh37z3ZzY/Gmm+U= |
| Phone | +1(212)555-0186 ext. 123 | IEWV38qOhXizX6uXmPYydSfrglkBOIfpSou2bt0lqcc= |
| Name | María José | 3VsywV/Fa+hXC9SFqmuszWhThL0Dyf6UelEDdsxJeKY= |
| Name | O'Connor | eg+/30DL65dCm9uI9RLOzSfw1Et+V5hqtXhCO7YZNqw= |
| ZIP | 91790-3771 | 2FPZucR4x7U8KlM+SFAX4LPGhwNz/PIZUCSUdDh0o/s= |
| VIN | 7F01C123456 | B4tPoGDMN2UMZSKr42kfXfbdN3uycsiuw4sgMLesNHs=The documentation prints a stray leading character in the standardized value. |
| MAID | a3f1c2d4-5678-90ab-cdef-1234567890ab | 250KY6lOgzYUB3EHrkbDCE2kMEZQE69SF38muhoDudI= |
| CTVID | SmartTV_9F8E7D6C | d9fxuTIrKksBXYDlaWsvQbcyUUfCbUx7v78gd+KP+tc= |
| NDZ | Danielle · Johnson · 1985-07-04 · 91790 | PQOfn1RffEKmqMmNAzDKKaoZCwxWbQZkQzPWmQo9REA= |
| NDZ | Jane · Doe · 7/22/1990 · 94103-3857 | X7qUgPCyldidyYW0lpVyXUQ1nfrZdsb3LDH02zC0rfQ= |
| NameVIN | Eve · Genesis · 1HGCM82633A004352 | rtnDuXIe63jXYQQXW5r07GJ7lSsrib8+46QuKFwkOmk= |
The hash checker runs the same rules in your browser for any value you type. Nothing you enter leaves the page.
A minimal implementation
The hashing step itself is short. This Node.js sketch handles email and the composite step; the work is in the standardization rules above.
import { createHash } from "node:crypto";
const b64sha = (s) => createHash("sha256").update(s, "utf8").digest("base64");
const email = (v) => v.replace(/\s+/gu, "").toLowerCase();
b64sha(email("[email protected]"));
// "KA18MT/ph6IHYjzT9zwETySDQyvSh87YuoSBpOQtkhE="
const ndz = (first, last, dob, zip) =>
b64sha(b64sha(first) + b64sha(last) + b64sha(dob) + b64sha(zip));
// arguments already standardized: "jane", "doe", "19900722", "94103"Use toLowerCase(), not a locale-aware variant, and test with names that contain Turkish İ, Greek with accents and Cyrillic signs; the published vectors do not cover them.
Where matches get lost
- Hex output instead of Base64, or URL-safe Base64.
- Phone numbers stored with an extension in the same field, or with the country code stored elsewhere and added back.
- Dates of birth stored as text in mixed formats, converted with the wrong day and month order.
- ZIP+4 values with a Unicode dash pasted from a document, which an ASCII-only rule does not recognize.
- Composite hashes built from raw bytes, or with separators between the parts.
- Spreadsheet software stripping leading zeros from ZIPs or turning long IDs into scientific notation before you hash.
CalDrop Compliance applies these rules in a background worker in your browser, so your customer file is hashed on your machine and never uploaded. How browser matching works.
Questions
What hash does DROP use?
SHA-256 over the UTF-8 bytes of the standardized value, written in standard Base64 with padding: 44 characters ending in '='. There is no salt or HMAC.
How do I hash an email address for DROP?
Remove all whitespace, lowercase it, keep dots and plus signs, then take the SHA-256 of the UTF-8 bytes and Base64-encode it. [email protected] hashes to KA18MT/ph6IHYjzT9zwETySDQyvSh87YuoSBpOQtkhE=.
How is the NDZ hash built?
Standardize and hash the first name, last name, date of birth (YYYYMMDD) and ZIP separately. Join the four Base64 strings with no separator, then SHA-256 and Base64-encode the joined string.
Why does DROP strip leading zeros from ZIP codes?
The technical specification strips leading zeros before keeping the first five characters, so 02134 is hashed as 2134 and 00712345 as 71234. Follow the specification, because DROP builds its hashes with it.
Do I need to store my data hashed?
No. You only need to standardize and hash it when you compare it with a DROP list. Storing it hashed is your choice.
Which version of the DROP specification is current?
Version 1.3.0, dated 19 November 2026, in CalPrivacy's technical documentation.
Related
Sources
- CalPrivacy: DROP operations and API documentation (hashing rules, transliteration, test vectors)
- CalPrivacy: DROP technical reference (version history)
- CalPrivacy: Processing DROP requests
- DROP regulations, 11 CCR §§ 7600–7622 (PDF)
- CalPrivacy: DROP audits (proposed regulations)
Written by the CalDrop Compliance team at Preptics LLC and last reviewed 22 November 2026. This guide explains the rules; it is not legal advice. CalDrop Compliance is an independent product of Preptics LLC. It is not affiliated with, endorsed by or operated by the State of California or the California Privacy Protection Agency (CalPrivacy).
