overview
what's new
articles
Synamedia Iris forecasting is designed to help customers predict the availability of ad opportunities over a given period in the future. Forecasting uses historical data and trends to allow customers to plan and optimize available ad space and make informed decisions about order acceptance, pricing, targeting, and campaign delivery. The service is available with API only.
Synamedia Iris supports forecasting for all Streaming Types but capabilities vary depending on your selection. Two types of forecasts are supported:
All inventory forecast: When enabled, you may specify dates, segments or, for OTT streaming types, Key-Value Pairs (KVPs) as inputs and Synamedia Iris will return the expected number of upcoming ad opportunities that meet your criteria. The number of ad opportunities returned will be the predicted total of all matching inventory, as this forecast does not remove impressions that have already been booked from the total provided. It is not possible to specify more than one streaming type on a forecast.
Campaign forecast: Campaign forecast is available for Broadcast and OTT streaming types. This forecast is designed to help determine if sufficient inventory is available for a Line Item and enforces the same constraints as a Line Item booking. You must specify the streaming type, Segments (multiple segments are not currently supported by the forecaster), and the campaign's start and end times. In addition, a date range is mandatory and the campaign forecaster supports Line Items with a maximum of 120 days duration from the day the forecast is run. This forecast returns the total number of available impressions, the number of already allocated impressions, and a summary of competing campaigns. If insufficient available inventory is predicted, customers will then be able to make business decisions to amend the Line Item, competing campaigns or source additional compatible inventory.
The Line Item forecast will show impressions that are expected to be delivered. This may be fewer than the total you are booking for the campaign. The forecaster will also show impressions that could be used by the Line Item being forecast but are under commitment to other campaigns.
In these circumstances, you will need to make a business decision to adjust another Line Item to free up the inventory or source additional inventory. Finally, the system will also show the number of impressions that cannot be achieved because there is insufficient availability in your system. In this case you should source additional inventory or reduce the size of the booking.
The Line Item forecast cannot predict for some available targeting options, such as device or geo targeting. If incompatible elements are targeted in the Line Item, then the system will be unable to provide a forecast. Synamedia is working on adding additional capabilities to the forecaster. The ‘all inventory’ forecast is able to provide a prediction using Key-Value pairs, but this is not available at the Line-Item level.
You can contact Synamedia Iris if you wish your instance to be configured for Iris forecasting. Once configured, the system requires ‘production level’ data of three months (for each streaming Type that you wish to enable) before it can start to generate forecasts. After three months of data has been collected, the forecasting API will be available to start to run queries. Note that the streaming types can be configured to start the training phase at different times.
For Broadcast, Synamedia Iris requires both the National Panel Data and As-Run Logs to be able to generate the forecast.
Once Synamedia Iris has the initial historical data that can be used as the foundation for the forecast, then any new segments or KVPs will only need four weeks’ worth of data before they will be able to return usable forecasts.
The accuracy and reliability of any forecast depends heavily on the quantity of the data used to generate it. For Synamedia Iris forecasting, the more historical data available and the more comprehensive and detailed that data is, the better the forecast is likely to be. Without enough data or incomplete data, the forecast will be less reliable.
Therefore, it is recommended that you do not forecast for a future time that is greater than the amount of data used to generate the forecasts (that is, with three months’ worth of data the forecast should only be used to forecast an upcoming three months). As Synamedia Iris has more historical data then seasonal trends, and other inventory variations, can be better applied to the predictions.
Similarly, significant changes to the number of ad requests made to Synamedia Iris, or a change to the National Panel data provided for Broadcast inventory, such as a sudden increase in volume from the introduction of a new channel or content, cannot be reflected in the forecast until the Iris forecaster has sufficient data to recognize that the volume change is permanent.
