Dynamic Floors: Our Real Life Results

Honestly, if we assume that you should optimize your ad revenues by innovating and refreshing your approaches… with actual data on internet, are dynamics floors the first solution you are thinking about ?

Well, we've decided to shed light on the subject for you !

In the ever-evolving landscape of programmatic advertising, finding the right starting price for your ad inventory is akin to setting the stage for a successful auction.

The Auction Dilemma: Finding the Sweet Spot

Imagine auctioning a water bottle. Set the starting price too high, and there is no taker; set it too low, and risk underselling. The same principle applies in programmatic advertising, albeit on a colossal scale.

Enter Dynamic Floors—a game-changer in maximizing the value of every ad impression while maintaining a healthy fill rate.

How Dynamic Floors Work: A Deep Dive

🔄 Fill Rate Calculation: Dynamically adjusts the minimum auction price based on the previous day's fill rate, representing the proportion of occupied ad slots.

🚫 Exclusion of Non-Optimizable Ads: Excludes certain ads from the model to calculate a fill rate specific to optimizable inventory.

✨ Multiplication of Base Floor: Establishes a "base floor" as the starting price, multiplied by a value determined by the previous day's fill rate.

📈 Variable Management: Adjusts variables based on demand and fill rate, gradually increasing the floor in high fill rate scenarios.

🛡️ Safeguards: Implements safety mechanisms to prevent excessive adjustments, maintaining a balance between fill rate and higher bids. If the publisher needs it, the minimum safeguards can be it’s own floor.

📈 Presented below is a graph depicting the impact of Dynamic Floors from October 15th to November 19th in 2023. The data exclusively focuses on display inventory, with the RPM normalized between 0 and 1.

The issue with A/B testing dynamic floors, is that, in the case where it's a random 50/50 split on the exact same inventory, there might be a huge bias as bidders won't be able to properly understand which part of the inventory behaves differently, hence potentially adapting their bidding behavior on the overall inventory instead of adapting it only on the 50% with dynamic floors. And if they do that, comparing A and B could become useless.

So, here we chose 2 different inventories but that had very similar proprieties within the same Publishers, all the while being different enough so the effects of dynamic floors won't change the bidder's behavior on the inventory which doesn't have dynamic floors. The inventory with dynamic floors is shown here in blue, and the inventory without dynamic floors implemented is in red.

Notably, a noticeable stability of the RPM and its upward trend in revenue are observed when Dynamic Floors are active. This underscores their effectiveness in increasing the RPM and maintaining it by mitigating events that would otherwise impact it.

Dynamic floors show a strong upward trend (in blue) compared to our test inventory

🕰️ An interesting graph is the level of adjustment per hour of the day on average:

Our floor multiplier is evolving by a wide range throughout the day

Here, you can see how our algorithms dynamically adapt to the hourly bidder demand. At night, bidders are not very active, hence a lower multiplier. Throughout the morning, multipliers progressively increase towards a plateau that’s maintained during the day, and decreases during the evening, reponding to a surge in inventory available at peak audience. If demand stays high but supply dramatically spikes, multipliers need to be adjusted accordingly.

Result: Smarter Bids and Maximized Profitability

By employing dynamic floors, Hubvisor strategically refines advertising bids, favoring high-demand slots while enhancing the overall quality of ad inventory. Think of it as a dam—raising the floor on highly demanded slots redirects lower bids to less filled slots.

Unlock Quick Wins for 2024: Dynamic Floors by Hubvisor

This free optimization module, available to all Hubvisor suite users, has already demonstrated its efficacy. It maintains the fill rate while boosting the value per ad unit by an impressive 15 to 25% on average.

Take advantage of this strategic tool to elevate your programmatic advertising game in the year ahead. For a personalized demonstration or any inquiries, reach out to us at hello@hubvisor.io.

Here's to a profitable and dynamic 2024!

Best Regards,

Hubvisor's Team