The Forecast Accuracy Report
The report shows the accuracy of the cost and revenue models by day for specified portfolios. By default, it includes the daily predicted and actual revenue, cost, and clicks — and the accuracy of the forecasts — for each portfolio. It includes data from campaigns that are currently mapped to the portfolios.
You can view data for the previous 18 months.
-
This report provides the same data as the portfolio-level Model Accuracy Report except that you can run it across multiple portfolios and you can change the attribution rule. You can also run and schedule the report using custom parameters, and you can use it to create spreadsheet feeds.
-
The best practice is to view the Forecast Accuracy Report for at least the last seven days because, regardless of the portfolio’s spend strategy, most portfolios see an inherent day-of-the-week trend. The optimization capability takes this trend into consideration and allocates spend accordingly.
-
For cost forecasts, a deviation of 10% in the last seven days is considered acceptable, so actual spend that is between 90% and 110% of the forecasted spend is fine. For revenue forecasts, a 15% deviation in the last seven days is considered acceptable, so actual revenue that is between 85% and 115% of the forecasted spend is fine. Forecasts with higher deviations should be investigated.
-
When keywords in the portfolio are associated with bid shift constraints, the portfolio over- or under-spends by the total amount caused by the bid shift. As a result, the predicted cost columns deviate from the target spend by the increased or decreased spend.
Available columns
The following are the columns that are available for each report. The default columns are automatically included by default. You can add the available custom columns from the Columns section of the report settings.