Questions:
Why was the forecast wrong?
Why is the top range of my forecast higher for a smaller audience than a larger audience, when all other inputs are the same?
Why don't ranges tighten as targeting becomes narrower (for example, with a smaller audience)?
I imported an audience, but it's returning a forecast of zero. Why?
Why does the reach look larger than my country's population?
What are the dates for?
Why are the results in a range?
** Forecast cannot use audiences with data segments younger than 3 days
I built a tactic based on my forecast, but it did not spend in the range the forecast indicated. Why not?
It is impossible to deliver a 100% accurate forecast. Although historical data is the best proxy we have for predicting future events—and although we use seven days of data to forecast—supply and demand, market dynamics, seasonality, and other factors are always changing.
Forecasting uses ranges to help convey the uncertainty that comes with projecting future performance based on historical events. Ranges are aimed to help you understand confidence around forecasts and more accurately portray what you can expect to see a tactic's spend.
Think of these ranges like weather forecasts. Meteorologists rely on robust algorithms and a number of inputs (like wind patterns, historical weather patterns, and so on). Although a weather report may tell you to expect four to five inches of snow and you end up with three inches of snow, it still benefits you to know that you should be prepared to expect a moderate amount of snow so you can respond accordingly.
Why is the top range of my forecast higher for a smaller audience than a larger audience, when all other inputs are the same?
If you use a smaller audience to forecast, the Forecasting tool will generate a wider range to account for error rates.
The lower end of your forecast should be lower than the lower bound of a larger audience.
Why don't ranges tighten as targeting becomes narrower (for example, with a smaller audience)?
If you use a smaller audience or other restrictive dimensions to forecast, the Forecasting tool will generate a wider range to account for error rates.
Very small dimensions cause more uncertainty in forecasts, and providing a wider range indicates that there is uncertainty. We do not feel it would be accurate to show smaller numbers in a tighter range due to this uncertainty.
I imported an audience, but it's returning a forecast of zero. Why?
Please allow one to two days for data to populate from a newly imported audience. Until this happens, we will not have enough data to produce a forecast.
Why does the reach look larger than my country's population?
Forecasted reach is intended to be a directional number to the number of unique people that you can reach. We may provide counts that are greater than the population for a given market because person counts are based on person IDs that group together IDs belonging to the same user. Because of this grouping, it is possible that a person can have multiple person IDs that the Forecasting tool would provide counts on.
The reason for this grouping is the fact that the way in which people use devices can be very different. For example, we see some people use devices only at work and never at home. In this situation, we would have one person ID grouping the IDs they use at home and a different person ID for the IDs they use while at work.
In cases where the forecasting reach is greater than a population, the forecasts are set to Target Everyone. Target Everyone includes non-cookied users, and these are all counted as individual users. In fact, setting your audience to Target Everyone almost guarantees your forecasted reach will be reported high.
What are the dates for?
All outputs are for the date range that you selected. Use the dropdown menu to the right of your forecast tiles to adjust the projection length of your forecast.
Keep in mind that the further out you forecast, the less accurate your forecast will likely be due to fluctuations in avails, win rates, and limitations imposed by your specified frequency caps. In other words, forecasts cannot be exact due to variations in inventory supply and advertiser demand across the platform.
Why are the results in a range?
The Forecasting tool's outputs are conveyed in ranges, displaying varying degrees of confidence for different scenarios. Ranges are intended to help you understand confidence around forecasts and more accurately portray what you can expect to see an ad group spend.
Because forecasts are based on seven days of historical data, we cannot predict the future with perfection. As a result, a range is a more accurate depiction of what an ad group might spend.