Forecasting (lab)
Last updated
Last updated
This feature is flagged to "LAB" (aka Laboratory feature), it means that interfaces are not definitive and may evolve in the future.
The prediction line represents your future performance over the next 9 days, notably by framing it with a minimum and a maximum confidence curve (light orange band). Our Data Scientist team has refined a predictive model that learns your own seasonality based on a mix on last weeks activity and last year at the same period. So, to have a better accuracy you must have at least 12 month of data.
The analysis can then be extended to other metrics. Our team is ready to discover your expectations. The forecast provided to you today has been built in silos based on historical data from your account. Prediction performance was appreciated for its ability to predict your account history. The algorithm developed has been particularly developed to offer a good prediction of exceptional sequences such as the main moments of market activity (Black Monday, Sales,...). These major market upsurge movements are one of the characteristics of the web. Once you have access this analysis, you will notice that it is retroactive, so that you can see how good would have been over past periods.
The Baseline shows how it would have evolved without the intervention of external factors, based on last 4 weeks activity. (Hidden bydefault, click on the legend to activate it)
The trend illustrates global tendency on the (non-predicted) period. (Hidden bydefault, click on the legend to activate it)
This metric help you to see what will be your expected minimum income for the day in the worst case scenario.
This metric show you what will be your average basket at the end of the day.
On the lower part of the report, you will find insights that help you contextualize the prediction, including a series of metrics on the current period (excluding predicted days) that help you compare the difference in performance between the observed period and last year : Average daily conversions/revenue, average basket by week, revenue vs last year.
