Level 8 · Social-media shoppers on iPhone email

One mask per rule

Jordan’s ticket has three rules, and each one becomes a mask. Naming them keeps the final filter short and makes it obvious which rule does what:

is_consumer = customers["segment"] == "consumer"
from_social = customers["referral_source"].isin(["instagram", "tiktok", "pinterest"])
on_icloud = customers["email"].str.endswith("@icloud.com")

audience = customers[is_consumer & from_social & on_icloud]

isin is the clean way to say “is one of these values”. The alternative is three equality tests joined with |, inside their own parentheses, which is longer and easy to group wrongly.

str.endswith checks how a piece of text ends. It’s the right tool for an email domain: "@icloud.com" must be the very end of the address. str.contains("icloud") would also match a Gmail address that happens to include “icloud” in the name part.

What about missing values?

referral_source is missing (NaN) for 666 customers. isin simply returns False for them, so they drop out, which is what we want: an unknown source isn’t a social referral. str.endswith would return NaN for a missing email, but every customer has one, so the mask is clean here. On a column with gaps you’d add na=False, just as with str.contains.

Sorting with a tie-breaker

Several customers can sign up on the same day, so sorting by signup_date alone could list them in any order. Adding customer_id as a second sort key makes the order fully predictable:

audience.sort_values(["signup_date", "customer_id"])

The list has 50 customers. The oldest signup is customer 67 (2022-08-26, via TikTok), and the newest is customer 1999, who signed up on 2025-12-31. Instagram brings in half of them (25), TikTok 15 and Pinterest 10, which matches Instagram being the biggest social source in the whole table (218 customers).

Common mistakes

• Writing customers["referral_source"] == "instagram" | "tiktok", which fails. Use isin.
• Using str.contains("icloud") when the rule is about how the address ends.
• Joining masks with and instead of &.
• Sorting only by signup_date, so customers who signed up on the same day can come out in a different order than the one asked for.