The forecasting tool provides customers with a projected number of available auctions for purchase based on matching targeting and creative parameters. This allows media buyers to better understand if a line item has the adequate scale of supply to achieve its goals.
How it Works
To access Availability Forecasting:
Navigate to the Edit Line Item screen and click on the Analytics tab on the far right:
To interact with this feature, select a Sampling Date Range from the left-hand side. This controls the sampled data used to run the Forecast over. A customer can set the range up to 30 days ago through yesterday.
This selection is crucial to retrieving an accurate forecast as Beeswax uses historical data to populate results. As changes may be made to the overall traffic being sent to a customer’s bidder, the user must be careful to choose a date range that considers any added or removed QPS filters which significantly alter traffic makeup. Otherwise, they run the risk of using an unrepresentative sample.
If the user has no preference, we recommend using the default value (last 7 days).
Once a date range is set, hit Run Forecast. The expected wait time is roughly between 5-30 seconds, and closer to the latter if the line item’s targeting expression is especially complicated. It’s important to note that Forecasts are run based on the last saved changes to the line item. Unsaved changes will not be accounted for.
These tiles display our forecasted projections to the user, providing a daily estimate and a lifetime estimate. In addition, the tiles display a percent comparison against the last forecast run for the line item so that a user can compare between two different targeting keys / adding a new creative type / trying a sampling date range that had a different filtering template, etc. The comparison will only persist within the same page view in the UI.
On the right-hand side, the forecasting tool displays a time-series bar chart. This shows the projected number of available auctions for the next 7 days. The data is derived by displaying the “latest” observed value for that day of the week in the sample date range.
Parallel to the x-axis there is both a line representing the daily average (the same value populated in the left side, but shown in relation to the forthcoming 7 days) as well as a dotted line representing the daily average for the previous Forecast. These can be used to compare how a Forecast differs compared to the previous run.
Unlike Metamarkets, Availability Forecasting supports segments, with a few nuances outlined below.
Segments can be forecasted with the include, exclude, require logic expressed in the targeting setup. The following rules apply to whether a segment is eligible for Forecasting:
- 1st Party Segments – The segment must have existed in the customer’s bidder for the sampling date range provided. The segment did not previously need to be targeted by any line item.
- 3rd Party Segments – The segment must have been shared with and targeted by a line item within the customer’s bidder during the sampling date range.
- While Segments and segment booleans are generally supported, Segment Recency and Segment Extension are not currently supported.
The following targeting and creative features are not accounted for by our Forecasting service:
- Lat/Long Lists
- Lists <2 Hours Old - It will take up to two hours after creation or changes made for lists to be up-to-date for use in Forecasting.
- Segment Recency & Segment Extension
- Some Creative Attributes including MIME Type, Video Bit Rate & Expandable Directions
- IP Range Functionality
Q: Can I access Availability Forecasting via the API?
A: No, this is a UI only feature.
Q: Can I forecast on the Campaign LeveI?
A: No, forecasting is only supported on the line item level.
Q: Does Availability Forecasting project line item delivery?
A: Not at this time. Delivery Forecasting will be considered in future updates to the tool.
Q: How does this feature differ from Metamarkets?
A: This feature supports availability forecasts based on audience targeting. For details on segment support see the relevant section above.
Q: How does Beeswax calculate the forecasted available auctions?
A: We take a sample of the auctions being routed to a customer’s bidder and log them to our Forecasting Database. When a user generates a Forecast, we then calculate the average daily number of auctions matching the line item targeting keys and select creative properties.
To calculate the lifetime approximation of available auctions we take the number of days the line item is intended to run and multiply the daily average by that number. If no end date is set, we will display a weekly estimate instead.