What’s a Price Floor and Why Is It so Difficult to Optimize?
A price floor is a minimum threshold below which an advertising partner cannot serve a campaign. The lowest winning bid has to be equal to or higher than the floor price.
As a publisher, optimizing and setting price floors is very challenging because they can be very different for particular audiences and dimensions, for example:
Contextual — domain and subdomain, article, ad-unit size, position, etc..
Demographic and behavioral — device, user country, browser type.
Non-behavioral — session thickness, time of the day/week.
Being a price-floor specialist takes a very good sampling methodology and the ability to observe and enhance performance over time.
Google Ad Exchange vs. Google AdSense
Previously, getting direct access to AdX was reserved only for very large publishers. Little – to medium-sized publishers had no choice but to use AdSense, as getting approval for AdX was challenging. Recently, this has started to change and Google has made AdX more accessible. Now, U.S. publishers using as little as 5 million pageviews per months have the ability to get approved for an AdX chair through their Google Ad Manager.
4 Floor Pricing Models and Optimization Methods
There are four main optimization methods that publishers can choose from when it comes to open-auction pricing rules.
Zero price floor
The publisher does not have any flooring pricing collection and leaves everything up to the Google AdX algorithm. This can work well if there is a lot of demand for particular impressions; in such a scenario, the bidders optimize each other because of additional demand pressure.
Problem 1: There are periods when there are fewer bidders, so during such periods there is very little demand pressure, and using price floors will deliver greater revenue.
Problem 2: You allow Google optimize automatically. Google’s algorithm is designed to be fair to both advertisers and publishers–it doesn’t prefer publishers. The majority of Google’s money comes from advertisers, so the scale is often tipped in their favor.
The publisher has a single, flat floor price that they leave and do not touch. This can work very well when the publisher has a really strong direct-sales team and would like to make sure they don’t sell their inventory below the minimum price.
Problem: Programmatic has distinct demand based on parts of the day, week, month, and year. It’s difficult to know what the best price floor should be without repainting the inventory and observing performance, so a fixed price floor generally doesn’t deliver optimal revenue.
In this model, the publisher strives to optimize coverage and revenue to avoid having 100% coverage so that they don’t sell their inventory below the optimal market price. Publishers can choose to change their floor pricing once per day, or multiple times every day.
Problem 1: It’s tough to have a fantastic methodology which maximizes fill rate while ensuring optimal price floors. In many cases, publishers are just guessing what the price floor should be and don’t have a clear sense of how it impacts revenue.
Problem two : It’s very time consuming and can take several hours each day depending on how you are performing price-floor analysis. This leaves less time for other high-value tasks like focusing on deals, formats, etc..
A relatively new concept for setting AdX prices where fill rates and pricing strategies are adjusted automatically to achieve the maximum revenue and not go below a certain eCPM level. It functions as a soft price floor, allowing Google to buy inventory which it wouldn’t be able to buy normally with a hard price floor.
Problem: It’s relatively new so we do not know the specific impact this has, but from our experience, together with Target CPM let you still have to adjust price floors to achieve optimal monetization. Based on our analysis, this procedure delivers only around 2% additional uplift from traditional pricing methods.
The Human Element of AdX Optimization is Not Efficient
For those optimizing their price floors manually, the human element is one of the factors that makes it really difficult.
An individual return analyst generally works 9-5, five days weekly for a total of 40 hours. Demand can change drastically for different times of day and on different days of the week and price floors should be adjusted to account for these changes in demand.
Needless to say, not all hours of the day are equal and most people sleep through the night, so there’s far less website traffic through this time, but there is still a big gap in coverage if you’re optimizing manually.
The Solution to Price-Floor Optimization Challenges: Artificial Intelligence
During our own Ad Manager alone, we serve 10 billion impressions monthly (most of which come from Google AdX) and manage advertising operations for many client accounts that have their own AdX and Ad Manager stacks.
Our return analysts were spending too much of the time on guide price-floor optimization, and while the results were better than zero price floor, fixed price floor, and goal CPM optimization methods, we knew there was a better way which was both more efficient and could deliver greater revenue.
We (AdopX) began by assembling a group of top data scientists, software architects, and product managers. Together, we set out on a path to locate a solution superior to all other AdX optimization methods which could automatically optimize price floors, and have since made several big breakthroughs in this field. To summarize:
Our solution is designed specifically with the publisher in mind and we do our best to maximize the amount of money advertisers pay publishers. We essentially tip the scale in the publishers’ favor.
We capture and analyze your RTB data and employ machine learning and neural networks for predicting and adjusting pricing strategies in real time.
Our ML algorithm analyzes millions of combinations of measurements (contextual, demographic and behavioral, and non-behavioral) in search of uniform breaks in pricing strategies.
Thanks to our algorithm, automatic sampling, and historical data, we could choose the optimal floor price for any given time and demand.
Unlike other similar solutions on the market, our platform is able to optimize price floors as many times per day as it deems necessary, so it could be once daily or dozens of times per day.
To be able to analyze and examine new AI algorithms, we built a framework allowing for an algorithm fight–that is, a competition between algorithms, where the algorithm which delivers more revenue wins.
The bottom line is that you are able to get approximately 15–40% more revenue compared to any other price-floor optimization method. For a publisher that is making $20,000 per month, this is anywhere from $3,000 to $8,000 a month of additional earnings, while a large publisher with $500,000 per month in AdX revenue stands to earn additional $75,000 to $200,000 per month.
Our solution frees your analysts’ time to concentrate on other activities such as creative optimization while the algorithm maximizes price floors automatically.
Last but not least, we also provide complete transparency because we use your Google Ad Manager and Ad Exchange accounts, meaning the monetization results are at any time in Ad Manager’s query tool.