Campaign weight and pacing

When the system knows how many impressions actually are available for the campaign, the next step is to calculate how quickly it needs to deliver the booked impressions.

The procedure for doing this differs a bit depending on the campaign priority level and settings. In the explanation below, the following terms will be used:

gross available – available impressions when targeting parameters and already reserved inventory are not included in the calculations

net available – what is “really” available for the campaign, after taking the targeting, daily capping into account and campaigns with higher priority into consideration

remaining - available impressions from the day to consider (inclusive) until the campaign end date (inclusive), for all content units that the campaign is booked to

  • Exclusive campaigns (priority 0) have a customer defined weight, xx%. Each day, the campaign reserves xx% of all available inventory.
  • Campaigns with the setting “deliver as fast as possible” always reserve all the impressions it can.

For all other campaigns, the following happens:

  • For each day the campaign runs:
    • Calculate gross optimum today, the optimal amount of impressions to deliver based on remaining gross avail, gross available today and remaining to deliver.
    • Calculate net optimum today, the optimal amount of impressions to deliver based on remaining net avail, net avail today and remaining to deliver.
    • Adjust the optimums today according to the daily capping.
    • The highest of gross optimum today and net optimum today will be used to calculate the weight. The weight that day will be optimum today / net available today.

Why is the highest of net optimum today and gross optimum today used? You can see it like this. Gross optimum today is the best to use if we want even delivery related to the “cu imp curve”. Net optimum today is the best to use if we want even delivery related to the actual available impressions. The thing is we want both. Consider the following examples:

Example 1 (why we need net optimum today if that is largest):

Two campaigns on a content unit with 200 impressions per day during 2 weeks. Campaign 1 is exclusive 100% and booked week 2. Campaign 2 is paced, booked to 1400 impressions week 1 and 2.

If we would use gross optimum today campaign 2 would expect delivery on the red dotted line but that is not possible due to campaign 1, thus net optimum today is better in this case. Now we will get weight 100% w1 and 0% w2 for campaign 2 instead of 50% both weeks (which would cause under delivery).

Example 2 (why we need gross optimum today if that is largest):

Two campaigns on a content unit with 200 impressions per day during 2 weeks. Campaign 1 is exclusive 50% and booked week 1. Campaign 2 is paced, booked to 1400 impressions week 1 and 2.

If we would use net optimum today only, campaign 2 would expect delivery on the red dotted line but that is not exactly what we want. In this case it is better if the campaign is delivered relative to the gross avail along the green dotted line. Therefore gross optimum today is bigger and better in this case. Now we will get weight 100% w1 and 50% w2 for campaign 2, instead of 67% both weeks (which would cause uneven delivery).

The drawback with this method is that if there is more inventory in the near future than in the far future, it will forecast too high weight in the beginning, in practice this is not a big problem since most campaigns are booked in the near future.

Example 3 (potential problem, similar to example 1):

Two campaigns on a content unit with 200 impressions per day during 2 weeks. Campaign 1 is exclusive 50% and booked week 2. Campaign 2 is paced, booked to 1400 impressions week 1 and 2.

Here campaign 2 will get weights along the green dotted line. Intuitively the red dotted line would be better but that won't be the result with this algorithm.

Cxense © 2012 | Support