Performance Reporting¶
Measuring the uplift associated with using LiteSwitch requires understanding the difference it makes to the composition of traffic sent to Buyers and then measuring the performance of that traffic over a baseline. LiteSwitch recommendations affect the density of the high-quality traffic being sent to the Buyer, which results in changes to the following metrics:
The number of requests sent to the Buyer
The fill rate
The impressions count
The media spend
LiteSwitch can provide regular reports on these metrics via emails, please contact your account manager to set it up.
Measuring Uplift¶
In order to calculate the effectiveness of LiteSwitch, you can use the metrics provided by
the Recommendations Response. You should track separately the trading traffic versus the holdout
traffic, which is a random sampling. The holdout traffic is identified using the
"learning":1 field in the LiteSwitch recommendation response.
Metric |
Holdout Group |
Optimized Group |
|---|---|---|
Recommendation requests |
100,000 |
1,000,000 |
Buyer Requests sent |
100,000 |
200,000 |
Buyer Yes-bids |
10,000 |
70,000 |
Impressions |
1,000 |
9,500 |
Media Spend |
$1 |
$9.5 |
In this example comparison, LiteSwitch yields the following improvements
73% traffic compression (300K requests sent out of 1,100K potential ones)
40% bid rate increase (70K/200K vs 10K/100K)
35% win rate increase (9,5K/70K vs 1K/10K)
5% decrease of the media spend compared to the absolute potential ($9.5/1M requests vs $1/100K requests).