--- title: 3. Audience & Experimentation slug: lf-lab/3-audience-and-experimentation docTags: createdAt: 2026-05-19T07:38:57.343Z --- ### Narrative context Two days have passed since the welcome message. Most of the simulated contacts will go cold — they were curious, not committed. The brand should not keep spending on people who will not convert. Before the flagship offer email goes out, the journey filters down to the hot leads only. And for those that stay, it tests whether to strike while iron is hot (24 hours) or to give a longer pause (72 hours). ### Build the audience Before the journey can branch on *hot leads*, the audience itself has to exist. - In the left menu, under **Customer**, click **Audiences**. - Click **Create audience** → **Build rule**. - Set the **Name** to `[{{LASTNAME}}] Mortgage hot leads`. - Set the **Evaluation Method** to `Batch`. The first condition is about the propensity score towards the conversion: - On the left, in **Attributes**, search for `propensityScore`. - Drag **Propensity Score** (with capital letters) onto the canvas. - Set the operator to `is greater than or equal to`. - Set the number to `60`. The second condition is about the mortgage simulation: - On left, click on the **Events** tab. - Search for `simulation`. - Drag and drop **Simulation Completed** onto the canvas. - Then search for `amount`. - Drag and drop Amount (`Demopotemea > Simulation > Amount`) over the **Simulation Completed** card in the canvas. - Set the operator to `is greater than or equal to`. - Set the number to `1500000`. In plain terms: profiles with an high propensity to convert and that have done a simulation for more than 1.500.000 SEK. - Click **Publish**. :::hint{type="info"} **Why this matters**: the audience captures intent + fit in one definition - a recent high-value simulation combined with a propensity score that flags the profile as worth a second touch. The journey does not have to know how "hot" was decided - it only asks, at runtime, whether the profile is currently in the audience. ::: ### Add the audience condition The **Mortgage hot leads** audience you just published will now gate the rest of the journey. Customers who do not qualify exit here. Go back to **Journey Optimizer** using the 9-dots icon :inlineImage[]{src="https://api.archbee.com/api/optimize/1Jp0iolWv35Z_mkI7oUiT/1NRPAUqiUtAEHu1vuZLs_-20260514-132131.jpg" alt caption width="34" height="35"} on the top right corner. - Open the journey with your lastname. - In the left panel, click the **Orchestration** tab. - Drag a **Optimize** node onto the canvas, after the 2-day convergence wait. - In the Properties panel, set the **Label** to `Hot leads only`. - Click the textbox under **Expression** to open the expression editor. - On the left , open **Audiences** and drag and drop the audience **LF - Mortgage hot leads** onto the canvas. - Click **Ok** to save the condition. - Click **Ok** to close the window. - Rename Path 1 as `Hot leads`. - Click **Save**. The default fallthrough path becomes the exit path for non-qualifying profiles. ![](https://api.archbee.com/api/optimize/1Jp0iolWv35Z_mkI7oUiT/SryR_sxEh9azEkrbA1hua_image.png) ### Add the path experimentation The same offer can perform very differently depending on *when* it is sent. LF does not assume — it tests. Half the hot leads get the recap email after 24 hours; the other half after 72 hours. The platform measures which timing converts better and then scales every future profile onto the winning path automatically. - In the left panel, click the **Orchestration** tab. - Drag **Optimize** onto the canvas, after the audience check node. - In the Properties panel, set the **Label** to `24h vs 72h test`. - Set the **Method** as `Experiment`. - Click on **Create experiment**. - Leave Success metrics as **Indirect Clicks (AJO)**. - Configure the two paths: - Click on **Add treatment**. - Rename Treatment A as `24h`. - Rename Treatment B as `72h`. - Enable **Auto-scale**. - Leave the rest of the configuration as it is. - Click **Create**. Once you're back to the canvas, click **Save**. Your canvas should now look like this: ![](https://api.archbee.com/api/optimize/1Jp0iolWv35Z_mkI7oUiT/-RhLuWnG9mW1V13KA9nBM_image.png) :::hint{type="info"} Auto-scale means: once the experiment reaches statistical significance, every *new* profile that enters this node is routed onto the winning path. Profiles already in flight finish on the path they were assigned to. No one gets re-routed mid-journey. ::: ### Add the dynamic waits Each path now needs its own waiting period before the flagship email fires. - In the left panel, click the **Orchestration** tab. - Drag **Wait** onto the path labeled **24h**. - In the Properties panel, set the label to `Wait 24h`. - Set **Duration** to 24 hours. - Click **Save**. - In the left panel, click the **Orchestration** tab. - Drag **Wait** onto the path labeled **72h**. - In the Properties panel, set the label to `Wait 72h`. - Set **Duration** to 72 hours. - Click **Save**. :::hint{type="info"} We use two separate wait nodes rather than one parameterised wait because the experimentation report tracks performance *per branch* — keeping the waits side-by-side makes the timing variable visible in the canvas and in the report. ::: ![]() ### Business punchline You just made the journey *spend more carefully*. The audience filter respects the brand's budget — no offer goes to a customer unlikely to act on it. The experimentation node respects the brand's curiosity — instead of guessing the right timing, you have set up a structured test that the platform will resolve on its own. By the end of the campaign, every new hot lead is automatically routed to whichever timing has proven to work best.