From Point-of-Sale to RecommendationEngine
Within the Schemas menu, a somewhat unique option can be found. If RecommendationEngine is picked as the target during Schema creation, the OrderLine Schema can be selected. This Schema cannot be edited, only used as-is - being designed for a specific purpose: Transfering Point-Of-Sale Data into Raptor's proprietary RecommendationEngine. As such, all you need to do in order to run it is to pick a suitable name...
There are two main uses for OrderLine. Firstly, it can be used to feed legacy sales-data from before signing up for Raptor's services into RecommendationEngine, to give it a lump sum of figures to start working with. A one-time affair, generally, since you'll subsequently want to set up a regular pipeline for such data using one of the many other flexible options offered by DataManager. Secondly, it can be used to provide RecommendationEngine with regular influxes of POS data from brick-and-mortar stores in a straightforward and easy-to-use manner.
The Key to Efficient Data Management
For the most part, using the OrderLine Schema is no different from any other - for more details on how to establish a Dataflow and selecting a Schema for it, see the Introduction to Data Manager step-by-step guide. However, there is one unique trait to this Schema - the mandatory Key column. The purpose of this feature is to prevent RecommendationEngine from being choked with duplicate purchases, leading to inaccurate results. If your dataset already contains a unique identifier for each purchase, excellent! Simply map that to the Key column, and you're golden.
If not, you'll have to create one using the Combine feature. Simply click on the New Transformation button, and select two or more datapoints that, in combination, would be unique to a given purchase. For example, this could be an ItemID and a checkout timestamp - the source is ultimately irrelevant, so long as it's unique. Click past the Transform stage by selecting Skip Transformation, to arrive at the Combine-stage. You will now need to pick the order of the columns, but for the purposes of this particular exercise, this is actually irrelevant - the Key Generator is specifically set up to ensure that you can't accidentally create two 'unique' keys from the same datapoints by mixing up the order of the columns. So, just pick whatever order you like - it won't make a difference.
Finally, select the Key Generator Combination Type, and hit Combine before moving on to the final step. Here, you simply select String from among the various formats - right at the bottom of the drop-down -and hit Convert. You can now save your new Key as a Custom Column! Map the results to the Key column in the OrderLine Schema, and you're set. Simply fill out the rest of the Schema as normal.
Whether you're making a one-time transfer of legacy purchase data, or setting up a regular pipeline for data from physical points of sale, the OrderLine Schema will get it done quick and easy. Just turn the Key, and away you go!
Please sign in to leave a comment.