Data minimisation: collect less data

Data minimisation means collecting only the personal data you need for a clear purpose. Data you never collect cannot leak, be misused or need deleting later.

A sign-up form for a newsletter needs one thing, an email address. Many forms also ask for a phone number, a birth date and a postcode, and every extra field is a risk.

Write down why you collect the data. Then ask of each field: could we do this without it?

If the answer is yes, do not collect it. A field kept "in case it is useful" has no purpose yet, so it should not be there.

A purpose also limits use. Data collected to send a parcel is not free to be used for something else.

Not everyone in a company needs every column. The team that sends the newsletter does not need phone numbers.

A view is a saved query. People who read the view see its columns and nothing else. The table underneath stays as it is.

Use the first two lines from last time if your terminal is new, then make a view with three columns.

printf '[client]\nssl-verify-server-cert=0\n' > ~/.my.cnf

alias sql='mariadb -h db -u student -pstudent shop -t -e'

sql "CREATE OR REPLACE VIEW student_mailing AS SELECT id, first_name, country FROM customers"

sql "SELECT * FROM student_mailing ORDER BY id LIMIT 3"

sql "SELECT email FROM student_mailing"

The first read prints customers 1, 2 and 3: Lucas in the United Kingdom, Ines in France and Elena in Ireland. Only id, first name and country exist in the view.

The second read fails with an error that says the column email is unknown. The view has no such column, so the team cannot see it.

Sometimes a team needs a rough value. A birth date can become a birth year, or an age band. A full postcode can become its first part.

The rough value still answers the question, such as "how many readers are under 30?", but it points at far fewer people.

-- Keep a broader version of two precise values.
SELECT id,
       SUBSTR(birth_date, 1, 4) AS birth_year,
       SUBSTR(postcode, 1, 3) AS area
FROM signups
ORDER BY id;
-- Change 1, 4 to 6, 2 to see the birth month instead of the year.

sql "DROP VIEW student_mailing"