Campaigns – A/B Testing

Compare two campaign texts by creating separate campaigns for different guests. Vendion has no automatic A/B feature that splits an audience, measures a winner or sends the winning version to the remaining guests.

Decide what to learn Change one thing at a time: the greeting, heading, length or wording of the same offer. If both price and wording change, you cannot separate their effects.

Decide how to follow up the result. Vendion shows sending results, but not measured opens, link clicks or bookings caused by the campaign. You could use a clearly documented manual follow-up of replies or bookings. Choose the observation period before starting.

Prepare two groups

1. Build a relevant segment under Marknadsföring → Gäster → Segment and check contact details and consent.
2. Create manual tags Test A and Test B under Gäster → Taggar.
3. Open Gäster → Register, choose the segment and assign guests to the tags. Use bulk tagging for several guests at once.
4. Check nobody belongs to both groups. A segment requiring both tags can help find overlap.
5. Allocate as randomly and evenly as possible. Splitting by how long someone has been a guest creates groups with different behaviour and can bias the comparison.

Create and send

1. Open Marknadsföring → Kampanjer → Ny Kampanj and use a clear name, such as “Autumn menu – A”.
2. Choose the channel and text A. Under Mottagare, select Taggar → Test A.
3. Review recipients, text, consent choices and SMS cost.
4. Create version B with Taggar → Test B, keeping other conditions the same.
5. Schedule comparable sending times. Send each version only to its own group.

Compare after the chosen period Open both campaigns and check sending results. If many messages failed in one group, the comparison is less fair. Then compare the outcome you chose to track yourselves, using recipient counts as the denominator.

Beräknad ROI in the campaign view is not a reliable revenue measurement. It uses the guest’s latest visit and lifetime spending, and may use a limited sample.

Small differences can be random. No recipient count always guarantees a winner; consider both group size and actual responses. Record the version, time, audience, sending results and observations. Use the learning in a future relevant campaign without repeating the same offer to people who already received it.