Compare one variation across two non-overlapping groups and decide how you will judge the result before launch.
1. Write a testable question
Choose one element, such as asking for a reply versus inviting a booking-link visit. Keep the business offer and other wording consistent. Define the main outcome: confirmed bookings, qualified replies, or another observable action. Use a measurement window that gives both groups the same time to respond.
2. Build comparable groups
Randomly divide eligible, consenting contacts using a spreadsheet or analysis tool. Confirm that no phone number appears in both groups. Check major differences such as recent purchase history or location. Do not compare your best customers with your least engaged contacts and call the difference a message effect.
3. Launch and measure manually
Create two drafts and two groups in Tiringo. There is no built-in split-test assignment or significance calculation. Only one campaign can be live per workspace, so plan sequential launches with comparable timing and check when messages actually reach recipients. Record the chosen variation and outcomes externally; unique link labels can help with website attribution.
4. Interpret the result cautiously
Compare counts as well as rates. Six bookings from 100 recipients versus four from 100 is a small observed difference, not proof of a durable winner. Decide whether to repeat with more observations before changing the whole playbook. Keep a test log containing the question, groups, dates, outcome, and next decision.
5. Keep an honest test record
Save both versions before launch and do not rewrite the question after seeing the outcome. Note operational differences such as delayed delivery or uneven support availability. Those details may explain the result. A repeatable record is more valuable than declaring a winner from a single dashboard number without knowing whether the comparison was fair.
- Save both original drafts.
- Record send and observation windows.
- Document disruptions before interpreting results.
