1. What is it?
The forecast provides projection of 18 months and for weekly subscribers also 26 weeks in addition with weekly visibility starting from the latest period of GfK Market Intelligence data - also referred as Point of Sale (POS).
The Forecast insights show sales units and revenue at market and segment level (total and values side by side e.g. different inch sizes, split by channels (e.g. internet and physical store sales, where available). The Forecasts are updated weekly and/or monthly as new data becomes available, automatically capturing the latest economic trends and adjustments.
2. What can I do with it?
Use it to plan stock ordering and allocation decisions, manage inventory, maximize sales and manage your sales Targets.
Industry leaders use Predict Forecasting because it provides an objective baseline based on the most accurate data in the market, free from human bias whilst incorporating fast-changing economic realities.
It is backed up by advanced and robust data science techniques, with extensive testing and measurement of accuracy.
Industry clients use our forecasts to plan for brand-specific promotions, events, and to measure the impact of extraneous or exogenous events.
Click the download button in the top right to easily export Forecasting values to an Excel compatible file.
3. How can you use it?
There are two different data visualization options in the Forecasting page.
Time Series to see segment value e.g. screen sizes side by side to identify future importance (over or underperformance), with the x axis being time (weeks or months) and the y axis the size of the market (Units or Revenue).
Year over Year to spot seasonal patterns and trends for a longer time frame with the month or week as x axis and with years as Y axis.
Both Time Series and Year over Year offer the following insights:
- Market Size and Changes at the top of the chart.
- A black vertical line which represents the latest current available data (left side including current month or week) and the forecasted data starts on the right hand of this line.
- Hover over functionality to see the values displayed in the chart.
- Legend easily deselect a value in the legend by clicking on it and select by.
- Download Functionality with Excel (.xlsx) and Power Point (.pptx) Export option.
- The Accuracy of our Forecast at the top right corner of the graph, see more on Accuracy here.
Interactive fields in orange text colour to customize your analyses:
- Channels, including Panelmarket and Total Market (for monthly) depending per country
- Selected Key Segments, including Total Market
- KPIs selection: Units or Revenue
- Time Frame: Monthly with a Forecast for 18 months or weekly for 26 weeks
- Displayed as: Actuals or Year to data, which is the cumulation of the actuals since the beginning of the calendar year (first week or January)
- Changes in actual or percentage
This is a Time Series Example to see segment values future performance side by side:
This is a Year Over Year Example to identify seasonality patterns:
4. How is it calculated?
The model: Deep Learning Machine Learning Model
The forecasting output is produced using deep learning that learns from up to 10 years of historical MI / POS data. The data are consumed by brands, product groups, segments on a global level. This type of model can capture highly complex correlations, a significant improvement on traditional statistical methods. This is enhanced with market expert input for special events.
These types of models have been proved the gold standard in financial market forecasting, since they can handle and even learn from a wide range of time series data.
Input data
The model combines point of sale data with external indicators to form intelligent predictions whether a market is in future decline or growth:
- Multiple GfK MI / POS metrics (including sales volume, revenue, and store information, segment level). The Model can identify Trends, Global Correlations and Seasonal vs non Seasonal patterns
- Data on important events (Black Friday, Prime Day, Soccer events, local and regional holidays by country like Christmas or Lunar New Year etc.)
- External financial indexes, Markets Trends and Volatility
We built the model to incorporate additional data sources as they become relevant.
You can find an overview and more details in below screenshot:
