On this page
We take care of it for you
Taking personal details out of company data is careful, skilled work, and it has to be right before anything is licensed. Rightmark takes it off your hands as a service: we set the rules for your data, bring in a specialist team to do the work inside your own systems, and test the result.
You stay in control the whole way. You keep ownership and the key, you approve before and after samples, and you pay nothing upfront.
Rather run it yourself? Rightmark is building the Clean Room, a container that does the cleaning inside your own systems, with you at the controls.
Five techniques, chosen field by field
No single technique fits every field. A buyer training a support model needs the conversation, not the customer, so each kind of detail gets the technique that removes the person and keeps the record useful.
John becomes ****
Masking
Hides a value behind repeated symbols, so the field keeps its place and the value is gone. In images and video, faces, car plates, tattoos and other distinctive marks are blurred the same way.
Jane Doe becomes Person 0417
Pseudonymization
Swaps names and other direct identifiers for artificial IDs. The same person gets the same ID in email, chat and the CRM, so threads and workflows still connect.
37 becomes 35 to 44
Generalization
Lowers precision until a value stops pointing at one person: an exact age becomes an age band, and coordinates become a city.
1,240.00 becomes 1,262.00
Swapping and noise
Shuffles values between records, or adds small random noise to numbers, so no row leads back to one person while totals and trends stay usable.
jane.doe@example.com becomes [EMAIL]
Type tokens
Replaces each detail it finds with a token that names its type, such as [EMAIL], [URL] or [IMAGE], so a reader still knows what stood there. Detection matches known patterns and lists of words and phrases, your own list of client and project names included.
What we find and replace
We look for personal details in structured fields and in free text alike, because a name can hide in a greeting, a signature or a forwarded thread. These are the kinds of detail we find.
People and contact details
- Person names
- Email addresses
- Phone numbers
- Postal addresses
- Dates of birth
Identity numbers
- National ID, Social Security and tax numbers
- Passport and ID card numbers
- Driver's license numbers
Money and access
- Payment card numbers
- Card security codes
- Bank account numbers and IBANs
- Passwords
- API keys and access tokens
- One-time passcodes
Online identifiers
- IP addresses
- Account IDs, usernames and customer IDs
- Links that carry personal details
Health details that turn up in passing, such as a diagnosis mentioned in a support chat, are removed too. Rightmark does not take health data yet (patient records, medical imaging or veterinary records), until a HIPAA program and vendor agreements are in place.
Every kind of data you hold
Direct identifiers
Names, Social Security and other national ID numbers, home addresses, email addresses and direct phone numbers, wherever they sit: a CRM field, an email signature or the middle of a chat.
Images, video and biometric details
Faces, car plates, distinctive skin marks and tattoos, and any image or video feed with people in it. They are blurred before footage such as dashcam video is licensed.
Text and conversations
Support chats, tickets, email, records and documents that mention private people, from any operational system. Free text is read in full, not only the fields around it.
Numbers and attributes
Ages, income levels, exact dates and map coordinates. One of them rarely names anyone, but a few together can, so they are generalized or given small noise.
How your data is protected
You keep ownership
Your data stays yours. Buyers license the de-identified copy on terms agreed in writing before you sign.
It runs inside your systems
The work runs inside your own systems or your own cloud account, so raw data stays in your environment. Only the de-identified copy you approve goes to a buyer.
Role-based access
Only the people assigned to your project can open your data, and only for the job in front of them.
You hold the key
Where values become pseudonyms, the key that maps them back is encrypted and kept by you. Rightmark and buyers never get it.
Tested before you sign off
We re-scan the output, test a sample for anyone who could still be picked out, and show you before and after pairs. You approve the result before any of it is licensed.
Your rules, your counsel
We work to the privacy requirements set for your project. Which laws apply to your data, and what they ask, is for your own counsel to decide.
How it works
- Tell us what you hold: the systems, the kinds of data and roughly how much. Words only, no files yet.
- We map the fields and agree the rules with you: a technique for each kind of detail, and your list of client, staff and project names to replace.
- We de-identify a sample inside your systems, and you review before and after pairs.
- Once you approve the sample, we run the full set and check it again.
- Buyers license the de-identified copy. You pay nothing upfront: buyers cover extraction, scrubbing and transfer.
Questions
What does the service cost me?
Nothing upfront. Buyers cover extraction, scrubbing and transfer, and who covers what is set out in writing before you sign.
Does my raw data leave my systems?
No. The work runs inside your own systems or your own cloud account. Only the de-identified copy you approve goes to a buyer.
Is the result anonymous?
We call it de-identified, because that is what we can stand behind: direct identifiers are removed, the rest is generalized or replaced, and a sample is tested for anyone who could still be picked out. Whether data counts as anonymous in law depends on the data and on who holds what, so that judgment is for your counsel, with our test results in hand.
Who holds the key that turns pseudonyms back into names?
You do. It is encrypted and stays on your side. Rightmark and buyers never get it.
Can I give you names to remove?
Yes, and it makes the result better. Client names, project codenames and staff names you list are replaced wherever they appear, in fields and in free text.
Do you de-identify health data?
No. Rightmark does not take health data yet (patient records, medical imaging or veterinary records), until a HIPAA program and vendor agreements are in place. Health details that turn up in passing elsewhere, such as in a support chat, are removed with the rest.
How is this different from the Clean Room?
The Clean Room is software Rightmark is building for you to run yourself, inside your own systems. The service is a specialist team doing the cleaning for you, inside your systems too, with Rightmark setting the rules and checking the result.
Sources
- Rightmark: Trust center