Data collection
This section covers how the data in Rich is collected and calculated.
The Rich agent works with sampling, and will collect data for a subset of the views.
Rich was designed to provide quality data with high confidence and a low margin of error. Various factors will impact the quality of data, such as; The size of the campaign, The number of data points collected and Script access.
Why campaign size matters
Campaign size is important for the accuracy of the data.
We designed Rich to provide reports with high confidence and a low margin of error. Some events are relatively rare occurrences in a campaign. For instance, campaign engagements are commonly in the single percents. To get a good accuracy we need about 100,000 exposures.
How much is enough data?
How much data is needed varies for each metric.
Clicks and exposures are always collected by Rich, and will be accurate as soon as you see data in your reports.
The less common a data point is, the more data the campaign as a whole needs to have. As a rule of thumb, to be accurate any data point needs about 1000 data points collected.
For instance, Rich will not show data about a small blog if it has collected less than a thousand data points for it.
However, at the same time, the blog exposures will be used in reports about viewer engagement for the ad format on the blog. Rich may have collected more than a thousand data points about ad format engagement, while having too few exposures for the blog.
Why script access is important
Script access is important for data quality.
Some metrics can only be collected if the ad is allowed script access to the site. It’s required for the ad to "see" outside of the flash container, and access browser data. When script access is blocked, data for some metrics can’t be collected. The result is a lower data quality for those metrics.
You can see the script access percentage in the Diagnostics report of each campaign.
Metrics that are always collected:
- Exposure
- Click
- Engagement
- Time of day
- Ad size
Metrics that require script access (examples):
- Visibility
- Site
- Visible Reach
- Placement,
What happens when Script Access is too low?
Data will be N/A and/or estimate notifications will be shown in the campaign reports when script access is low.
We calculate how much data we need to be confident enough to display a metric. Rich will not show metrics if it doesn’t have enough data. They will be replaced with “N/A” (not available).
Estimate notifications are shown when there is enough data, but script access for the campaign total is low. The estimate calculation looks at the known data, and then assumes it’s representative for the whole campaign. Doing this we extrapolate an estimation for the metric.
The data is usually accurate (statistically significant), but the error of margin is likely higher than the rest of the data.