...

...

In Real-Time Bidding (RTB), the exchange's numbers calculations are used for billing, and absent something going significantly wrong, . Typically, the exchange's numbers should billing calculations will be slightly higher than the bidder's numberscalculations, which leads to differences between predicted and actual spending. On some exchange exchanges, discrepancies can be as low as 2% of spend, while on others (especially for video -orientedimpressions), the difference can be as high as 10%.

FreeWheel accounts for these discrepancies automatically in our its bidding algorithm to make the exchange's billing numbers and the platformFreeWheel's numbers closer than they would be otherwisecalculations as close as possible.

How it Works

Here's an a spend and billing example without a discrepancy allowance (zeroing out the platformin both examples, FreeWheel's fee is ignored for simplicity's sake):

• Customer A customer spends $1,000 on the exchange, according to FreeWheel's data
• Exchange The exchange sends the customer a bill at the end of the month for $1020

The problem that arises from this flow of data is that As a result of this discrepancy, there can be a significant and unexpected reduction in margin at the time of reconciliation. InsteadSo, the platform FreeWheel automatically applies a small discrepancy allowance when bidding:

• Customer "The customer spends " $1,000 on MoPub according to FreeWheel's datathe exchange, but the platform FreeWheel was bid- reducing bids by 2% so they only really spent $980 on the Exchange exchange (according to FreeWheel's data)
• Exchange The exchange sends Customer the customer a bill at the end of the month for roughly $1000 , since their numbers are generally 2% higher.

With the discrepancy allowance applied, the net difference between the original spend and the

...

exchange's bill is closer to zero.