Mixture stage forecasts

The first forecast of this sub-team is the aggregate-level gross sales forecast. With this challenge, we forecast the gross sales for the upcoming X weeks, each on the weekly and each day ranges. To provide a little bit of context round aggregation, one potential stage of aggregation may very well be the gross sales of the corporate as an entire. Such a forecast might help with making company-level selections and dealing on setting targets and expectations. One other potential stage can be gross sales that come by means of the warehouses of bol, which is essential for operations and workforce allocation.

An vital frequent attribute of most aggregate-level forecasts in our workforce is that in addition they rely upon the gross sales forecast (making them downstream forecasts), as gross sales are sometimes the first driver of many different metrics that we’re forecasting.

This leads us to a different essential forecast, which is the buyer help interplay forecast. With this challenge, we offer an estimate of what number of interactions our buyer help brokers can count on throughout the subsequent weeks. This forecast is essential for the enterprise, as we don’t wish to over-forecast, which might result in overstaffing of buyer help. Then again, we additionally don’t wish to under-forecast, as that might result in prolonged ready instances for our clients.

To ensure that our providers (webshop, app) scale properly throughout the peak interval (November and December), we additionally present a request forecast, that’s, what number of requests the providers can count on throughout the busy intervals.

Lastly, we offer a spread of logistics-related forecasts. Bol has a number of warehouses through which we retailer each our personal objects, and the objects of our companions who wish to use bol’s logistical capabilities to make their enterprise function easily. As such, we offer a couple of completely different forecasts associated to logistics.

The primary one is logistics outbound forecasts, that’s, a forecast indicating what number of objects will depart our warehouses within the coming weeks. Equally, we offer an inbound forecast, which focuses on objects arriving in our warehouses. Moreover, we additionally present a extra specialised inbound forecast that additional divides the incoming objects by the kind of package deal they arrive in (for instance, a pallet vs a field). That’s vital as these completely different sorts of packages are processed by completely different stations throughout the warehouses and we’d like to verify they’re staffed appropriately.

Merchandise stage forecasts

The second sub-team focuses on item-level forecasts. Bol provides round 36 million distinctive objects on the platform, and for many of these, we do want to supply demand forecasts. These predictions are used for stocking functions. This manner, we attempt to anticipate the wants of our clients and order any objects they could require properly upfront in order that we will ship it to them as quickly as potential.

Moreover, the workforce gives a devoted forecast that may deal with newly launched objects and pre-orders. With this forecast, the stakeholders can anticipate what number of objects will promote at some point earlier than the discharge and throughout the subsequent month after the discharge. This manner, we will ensure that we’ve sufficient copies of FIFA or Stephen King’s newest novel.

Lastly, our workforce additionally developed a promotional uplift forecast, which helps to guage the uplift in gross sales of a given merchandise primarily based on the worth low cost and the period of the promotion. This forecast is utilized by our specialists to make higher, data-driven selections with regards to designing promotions.