What is the difference between "Confirmed" and "Estimated" metrics?
Ever wondered what the difference is between “Confirmed” and “Estimated” numbers in your reporting? It all comes down to how we clean and organize your data.
Noisy vs Household IPs
Many IP addresses are what we call “noisy,” filled with many devices connected to the same IP address. Many of these are cellphone towers, or university or office networks where many devices are connected to the same WiFi with the same university/corporate IP address.
The non-noisy IPs are what we refer to as "Household" or "clean" networks. These are more precise than a noisy/non-household IP because a smaller, more consistent group of users and devices tend to consistently connect to the network.
Attributing "Confirmed" Numbers
In order for Chartable to provide an accurate measurement, we work with a device graph provider. The device graph provider matches multiple devices (tablet, phone, desktop) to one user, then maps the user to a household IP address while steering clear of the noisy IPs where it can’t clearly categorize the devices together. The household IP address is where your “confirmed” numbers come from: the cleanest, highest-fidelity data we have to confirm attribution.
Generating "Estimated" Numbers
Since every campaign will have impressions from both household and noisy IPs, we use the behavior seen from the pool of household IPs to extrapolate performance across the entire ad run, including cellular and corporate IP addresses. That’s how we generate the “Estimated” numbers, which essentially represent "Confirmed" data + our estimate of how the noisy IPs probably performed.
This is based on two main factors:
- The ratio of noisy to household impressions in your campaign performance,
- and the performance from the "Confirmed" group.