Email Subject Line

Variant hashes are computed via Digest::SHA256.hexdigest(subject_string)[0..7].

How to run

Create the experiment

1. Create a new Linear ticket.

Ensure that you set the team to Notifications. This will allow you to choose the push copy experiment template.

2. Fill out the details of the ticket.

3. Assign the ticket to an engineer, that engineer can assign the ticket to Cursor

Note only a user who has Github access to the handshake repo can assign the ticket to Cursor and have Cursor create the PR

4. Review PR

Review the PR to make sure the copy is correct and experiment set up is correct.

5. Start the experiment

If using Launch Darkly and Eppo set the appropriate targets and create the Eppo experiment.g

Create the experiment

1. Create a new Linear ticket.

Ensure that you set the team to Notifications. This will allow you to choose the clean up push copy experiment template.

Notifications

Approve Job Posting

2026-03

• Eppo

Variant	Hash	Subject	Session Lift	Winner
control	431f6c1a	%{employer_name} just posted a role 🌟
variant_a	f061f7b7	New %{role_name} at %{employer_name}	2.09%	✅️
variant_b	ecdcd2ea	%{employer_name} is hiring: %{role_name}	1.69%
variant_c	2fb4a84b	Just posted: %{role_name} at %{employer_name}

Job Application Reminder

2026-02

• Eppo
• Linear

Variant	Hash	Subject	Session Lift	Winner
control	0fb01ae3	👀 Remember these jobs you checked out?
variant_a	24dc302d	%{employer} is still accepting applications	1.40%
variant_b	b5290df1	Still interested in this %{job_role} role?	2.39%	✅️

Job Recommendation Digest

2026-04

• Eppo
• Linear

Variant	Hash	Subject	Session Lift	Winner
control	0e8a27e1	%{employer_name} is hiring %{role} and more
variant_a	5a368bb6	%{first_name}, %{employer_name} is hiring %{role_name} and more

2026-01

• Eppo

Variant	Subject	Session Lift	Winner
control	New roles from %{employer_names} hiring now 🚀
variant_a	%{employer_name} is hiring %{role} and more	2.83%	✅️
variant_b	%{count} roles at %{employer_name} and beyond	−0.52%
variant_c	%{employer_name} may be looking for talent like you — plus more roles inside	1.01%
variant_d	Don't miss these roles from %{employer_names}	−0.55%

Saved Search / Digest Job Alert

2026-02

• Eppo
• Linear

Variant	Hash	Subject	Session Lift	Winner
control	e588997a	⚡️ Fresh jobs for "{saved_search_name}"
variant_a	c9f5a973	"%{saved_search_name}": %{employer_name} - %{job_role} and more	2.29%	✅️
variant_b	f855892a	"%{saved_search_name}": {%employer_name} and others are hiring	1.21%
variant_c	152208ca	New "%{saved_search_name}" jobs just posted	1.10%

Unread Inbox Message Weekly Digest

2025-02

• Eppo
• Linear

Variant	Hash	Session Lift	Winner
control	86bfc37a1748b72bd017ba36d452af9024a42daf724a7a0f30372423af9df5dd
variant_a	531c82085b4d16ebc4656f3544267d4c7d29912bbcd7bdb622c4affd1af7ec8e	2.42%	✅️
variant_b	269592639ef0448ade4de8e0546ddfd6e9d4fd54feabfe3144234da99d0a9029	1.61%
variant_c	bb2b0db2ab5bbf6dc4fcd398c629799e3c8938fb5ab710d815658914d928f891	1.98%

Hashes are long for this experiment because it was the first time we recorded it. Decided later to use the shorter hashes.

Variant	Condition	Count	Subject Line
variant_a	with_employer	one	%{employer_name} sent you a new message
variant_a	with_employer	other	%{employer_name} and others sent you %{unread_message_count} new messages
variant_a	without_employer	one	Someone sent you a new message
variant_a	without_employer	other	People sent you %{unread_message_count} new messages
variant_b	with_employer	one	New message from %{employer_name}
variant_b	with_employer	other	New messages from %{employer_name} and others
variant_b	without_employer	one	New message from someone
variant_b	without_employer	other	New messages from someone
variant_c	with_employer	one	%{employer_name}: %{unread_message_count} new message
variant_c	with_employer	other	%{employer_name} and others: %{unread_message_count} new message
variant_c	without_employer	one	%{unread_message_count} new message came in
variant_c	without_employer	other	%{unread_message_count} new messages came in