A/B Testing your Flows

The platform features a Split Randomly action, which directs contacts down evenly distributed, randomized pathways. This function is ideal for evaluating the performance of different flow segments. Furthermore, a built-in simulator is accessible directly within the flow builder, enabling you to test logic during the construction phase.

Externally, we advise sending actual messages to gauge flow performance and to carry out A/B testing.

A/B testing—alternatively referred to as split or multivariate testing—is a technique for comparing two or more variants of a message, flow, or campaign to determine which one most effectively achieves a specific goal. This controlled experiment is performed on comparable audience segments, where the original version acts as the control and any subsequent versions are considered treatments.

Getting Started

• Establish clear objectives: Determine the primary goal of your flow. Is it to disseminate information, gather data, or accomplish another purpose?

• Formulate a hypothesis: Predict which flow version you expect to outperform the others and outline the reasoning behind your prediction.

• Identify variables for testing: Compile a list of elements you wish to evaluate. Examples include overall flow length, individual message size, word choice, and the initiation method (e.g., a contact-triggered keyword, a phone call, or a scheduled start).

Conducting an A/B Test

• Isolate individual variables: Ensure you are measuring the impact of only a few changes at a time. For instance, if Version A personalizes greetings with a contact’s name and Version B does not, this singular difference makes it straightforward to attribute any variance in results.

• Utilize comparable audience samples: Distribute the different flow versions to groups that are similar and representative of your entire target audience. Showing Version A only to seasoned contacts and Version B solely to new ones would skew your findings. The ideal approach is to split your target population randomly and evenly.

• Allow sufficient time for results: While the platform allows for rapid deployment—you can register, connect a channel, and begin messaging in minutes—a measured approach is crucial for meaningful tests. Since SMS is a significant extension of an organization’s brand, larger deployments require thorough testing to ensure flows are positively received and effective.

Creating flow versions

Generating a new flow variant is straightforward: within the flow editor, click the ☰ menu icon and choose “Copy.” This will create a perfect duplicate for you to modify and label. You can then employ a Split Randomly action to link and distribute contacts between these versions.

You can link your versions using a Split Randomly action.

Evaluating the results of an A/B test