Our abandoned cart model gives an in-depth analysis of your customer journeys. By using the conversion window and intelligent inference, it lets us better understand details such as your carts' conversion rates, life cycles, and the optimal time for when a cart is likely to convert into a sale.

The conversion window is the time after your customer creates and abandons a cart where their cart is most likely to be recovered. By analysing millions of abandoned carts, we have determined that a reasonable default is seven days.

The default conversion window is customisable, but before you make changes, it's essential to understand that the abandoned cart life cycle is very short.

We identify any abandoned carts currently within the conversion window, as carts most likely to be recovered. These carts are where you should focus your engagement efforts, as these abandoned carts represent your best opportunities for potentially recoverable revenue. This is because customer journeys are likely to be still active, and purchase decisions might still be possible to influence. If you set a large conversion window, you are likely to reduce the accuracy of your reporting.

Our abandoned cart JavaScript tracks the status of abandoned carts using the cartPhase field. When we directly track a cart to an order taking place, we mark the cart phase as complete.

However, there are occasions where the customer journey is irregular and disrupted. For example, if a customer changes browser, device, or there was some other reason why we couldn't directly track the cart conversion to an order.

In these cases, the abandoned cart model is able to use inferred conversion. The model looks at each abandoned cart and attempts to find an order placed by the same contact within the conversion window for their cart. If a valid order is found, we define the customer journey as completed even if the cart was not tracked directly to a conversion. When this happens, we don’t change the actual cartPhase, but add a separate inferredPhase field and mark it as complete for these carts.

When we can't track an abandoned cart to a conversion, or find an inferred conversion within the conversion window, then we define a cart as lost. For lost carts, we set the inferredPhase field to lost, and the cartPhase field does not change.

By separating potentially recoverable carts and lost carts empowers us to give you a higher level of accuracy on your abandoned cart dashboard and analytics.

The JavaScript that powers our abandoned cart feature manages the status of these carts with the cartPhase property. The model's inference does not change this property in any way. Instead, it works alongside the script, with the conversion window concluding whether a cart is lost or converted.