Is it common practice to integrate DataMilk with staging environment to test the integration before deploying onto the production site?
- Yes, this is common practice and DataMilk offers full support to assist with the integration if needed.
Can existing or upcoming CRO tests run alongside DataMilk without conflict?
- Yes, there can be no conflict as long as the same parts of the site are not touched by both parties.
- The changes DataMilk makes are independent - and so the effect of the changes will be independently measured.
Does DataMilk have a mobile app solution?
- This is on our product roadmap
Why does DataMilk ask for the gross profit of the entire order to be passed to the pixel instead of each product?
- DataMilk’s models aim to maximize the Gross Profit of the session, not the gross profit of the individual items.
- Passing the gross profit of the whole order enables us to know what the gross profit was and what the session looked like in terms of what the person did when making the order.
Use case on how Gross Profit impacts a smart component - Smart Navigation:
- Smart Navigation product categories can be sorted more intelligently when we have Gross Profit data because we can sort by expected additional gross profit per user.
- We aren’t sorting products by gross profit, we are sorting what the user did based on the expected incremental gross profit.
- The model looks at what millions of users did and finds similarities to the live user and sorts the product categories in the menu based on what should lead to maximize gross profit.
Can one see individual smart component performance on the dashboard?
- DataMilk.AImonitors user behavior and dynamically decides when each component is displayed for the best results. Therefore, there is no control group so we can’t get an uplift number per component.
- How DataMilk measures performance is by monitoring detailed data for each component and the component’s statistical effect on the overall results. The effect is labeled as good, learning, or auto paused and is displayed in the dashboard for each component. If a component is not driving results, it will be auto paused by the system. See smart component performance tags that are displayed in your dashboard.
How can one see how Smart Components are performing?
- The DataMilk Dashboard displays the most important analytics.
- The Dashboard displays performance of the Smart Visits (treatment group) against the Original visits (holdback/control group) to show the uplift.
Does DataMilk require gross profit data to be provided?
- No, DataMilk can optimize just for revenue
- For DataMilk to calculate a ROI, an assumed average profit margin is used.
- However, having a gross profit level understanding of the business is ideal.
How much business control does one have of Smart Components?
- The store owner has ultimate control over Smart Components
- The store owner can manually turn smart components on or off
- The store owner always has at least 5 days to Preview how a smart component will look and feel on their site and provide feedback make a request to make creative changes.
- Store owners have full creative freedom while DataMilk manages all of the backend and data engineering.
Does 50% of all traffic see all of the components from the start?
- No, DataMilk AI determines which slices of users to show components to and can start at a lower percentage within the treatment group while it is learning, and gradually increase as theres more data and certainty.
Can DataMilk still launch Smart Components if less than 500K visits were analyzed since integration?
Usually, DataMilk can launch Smart Components as soon as 2 weeks after integration is confirmed. However, this is not always the case as having enough data is a critical requirement to ensure that performance is measured accurately. Also, if Smart Components are launched with less than 500K visits, the Dashboard will likely say ‘Collecting Data’ for all KPI metrics.
Can one see how a Smart Component looks and behaves from the Dashboard before they’re launched live to the site?
- Yes, the Smart Components List on the dashboard includes a Preview link for every smart component.
- Every smart component has a 5 days grace period in the ‘Coming Soon’ status before being launched live onto the site automatically.
Why does it say ‘Collecting Data’ on my dashboard?
- If the confidence intervals are really wide the software identifies this and displays "Not enough data" because more time is needed to prove the uplift with statistical significance.
For all other cases we display the uplift value with a classification:
- "Negative" - the upper value of the confidence interval is < 0
- "Neutral" - 0 falls within the confidence interval
- "Positive" - the lower value of the confidence interval is > 0
- In addition to the above requirements, the software also analyzes user behavior each weekday twice from the first day smart components are launched live onto the site before coming out of the ‘Not enough data’ mode.
Should the same integration script tag be used for the staging site as the production site?
- Yes, DataMilk recommends you to just move the same tag over as the simplest and safest solution.
Can the Magic AI script and Pixel be moved from staging to production on different days?
- No, this should happen simultaneously.
Does DataMilk pass events to a store owner's Google Analytics?
- Yes, DataMilk passes events called ‘DataMilk’ to GA and can conversion uplift can be measured. However, DataMilk’s dashboard analytics are usually more accurate due to GA’s sensitive configuration settings.
Will Smart Components that are not improving conversion rate be disabled?
- Yes, DataMilk’s multi armed bandit algorithm automatically turns smart component traffic up and down based on its success.
- Only the most successful smart components will show on the right slices of traffic.
- The AI will 'Auto-pause' any Smart Components that are not contributing adequately to positive results.