Introduction to Floor Price Optimisation.

One of the most important optimisation methods for Publishers using Google Ad Exchange is Floor Price Optimisation through Google Ad Exchange Pricing Rules. This is where publishers can set a minimum floor price for their inventory below which advertisers won’t be able to bid. So pricing rules play an important role in influencing ad revenue and ad CPM of publishers.

Publishers can set pricing rules through their Open Auction Pricing Rule section in inventory tab of their Google Ad Exchange console. There are many targeting criteria available in a pricing rule such as, one can specify ad unit, sizes, countries, key-values, device categories etc. Also, there are 3 branding types where one can set floor prices. Namely:

1. Branded: Buyers see complete URL of publisher’s page where their ad is going to be displayed.

2. Semi-Transparent: Here, buyers will only see the top level domain of the publisher’s website.
Or Publishers can also specify custom URL that needs to be sent to Buyers. Also, this will only be available for Desktop and Mobile web inventories. Mobile App inventory always transacts as Branded.

3. Anonymous: Buyers will be having no information where their ad is going to be displayed.

If you’re setting pricing rules, it is always good to keep it as granular as possible. You can set the pricing rule targeting specific dimensions such as ad unit, inventory sizes, key-values, devices, etc. or you can try different combinations of dimensions such as ad unit – county, ad unit – inventory size, etc. For example, inventory size of 300×250 would give better CPM and revenue than the size 210×229, so keeping 300×250 a little higher floor price than 210×229 may yield a better revenue for the publisher. Similarly for a particular inventory, we observed that 30% of the traffic is from USA and 60% is from India. We have decided to set different floor price for USA and India and set a separate pricing rule. We could see 45% upside in revenue and CPM.

You can also set pricing rules according to the market behaviour. Usually advertisers are more likely to bid high CPM on festival seasons like Diwali or Christmas. Also, companies like Flipkart or Amazon bid a very high CPM when there’s a big upcoming sale. In that case, publishers have the flexibility to target selected buyers or advertisers and set a separate floor for them.

Biggest challenge in setting up of pricing rule is determining what should be the right floor to set. Because it is very dynamic and yesterday’s floor may not hold good for today or may not be the best possible floor! Again, right floor is different for different dimensions such as ad units, inventory sizes, location of the ad etc. Also for the same inventory, it is different at different time intervals such as weekdays, weekends, beginning of the month, end of the month, festival season etc. So it becomes extremely important to understand your inventory to arrive at the right floor price.

Some of the key metrics to look for when analysing pricing rules are Fill Rate (Coverage), Ad CPM and Ad Request CPM. One can expect a drop in fill rate when one keeps a high floor and vice-versa. The trick is to balance these two. As mentioned in my earlier blog posts here, Ad Request CPM would be the ideal metric that captures the balance between Ad CPM and Fill rate.

One can play around with Branded and Anonymous floor prices. We have tried setting only branded and blocking anonymous completely. There was mixed response. For some ad units, we were able to see good incremental revenue and CPM, and for some, although there is increase in CPM, fill rates have dropped dramatically resulting in overall drop in Ad request CPM (and revenue). Usually blocking anonymous works when there is a good demand for that inventory otherwise your fill rates would become too low to see any upside. We have also tried keeping our branded and anonymous floors $0.00 and the result wasn’t that pleasant. Fill rate was close to 100% and CPM dropped dramatically hence reducing the revenue.

Monitoring and Analysing all these on a day-to-day basis would be tedious and eat up a lot of Publisher’s time. Hence to make things easier and to get the best possible yield for Digital Publishers, at Tercept, we are working to fully automate the entire process with the help of A/B Testing.


Vinay B Rao
Senior Business Analyst, Tercept

  • Share:

Leave a Comment