Why QPS optimization matters
The programmatic industry has achieved remarkable growth and scale over the past decade. The open internet is massive, with consumers increasingly connected to screens in their pockets, on their desks, and on their walls at home.
SSPs, on behalf of media owners, are responsible for absorbing the entirety of the open internet, processing hundreds of billions of requests daily. That scale only continues to increase, elevating the importance of maintaining efficiency and delivering value.
To better manage this continued growth, DSPs set limits on the QPS, or queries per second they receive, which helps manage efficiency, sustainability, and cloud computing costs. But if QPS is not effectively optimized for the dynamic conditions in today’s programmatic market, it can result in missed opportunities for DSPs and marketers.
Let’s walk through QPS optimization and how it helps maximize value in the programmatic ecosystem.
What are QPS limits?
Starting with the basics, what is QPS? QPS stands for “queries per second.” In programmatic, that’s specifically how many bid requests SSPs send to a DSP within a given second.
Think about all of the activities that happen on the internet every second. Universally, let’s estimate that there are about 10 million QPS. That’s way too large for any one DSP to listen to, especially when you consider that they work with multiple SSPs.
So, DSPs set QPS limits, which cap the total amount of available inventory they see from the open internet at any given second. Of that universal 10 million QPS on the internet, each DSP will set a much lower limit to match their business needs, say, around 1 to 3 million QPS. Anything beyond that, they won’t see.
Ultimately, DSPs want to receive the most relevant opportunities with the least amount of waste so they can deliver the most value to their buyers as efficiently as possible.
How do DSPs set QPS limits?
How a DSP decides to set its QPS limits is complex and nuanced depending on its specific business. Let’s look at a few common factors for consideration.
First, there are infrastructure limitations. Capabilities and scale vary by DSP, and limits are used as a barrier to ensure a DSP doesn’t receive more QPS than it can handle.
Second, DSPs work with many SSPs, each of which brings different variations of supply, value, efficiency levels, and performance. DSPs will give different SSPs varying QPS limits based on these factors, with the best performing and most efficient SSPs typically receiving higher limits.
The importance of dynamic QPS optimization
With the rise of new channels like streaming TV introducing different traffic patterns, the factors used to set and determine QPS limits—and how those limits are used—need to evolve. The importance of dynamic QPS optimization has never been higher.
Let’s look at an example in streaming TV that demonstrates this—in this case a highly anticipated live sports event.
You can see there’s a drastic spike in bid requests to fill all of this broadcaster’s ad pods during halftime coverage when ads run, and then a drastic drop in requests as the game resumes without ad breaks. Imagine a steady line at the lower tenth percentile representing a DSP’s QPS limit. With this cap, the DSP would never see over 90% of the bid requests during halftime.
They would essentially be limited in how many impression opportunities they see, reducing the number of households and viewers their buyers can reach during this peak period. As halftime ends, you’ll see they have more than enough tolerance within their QPS limit to ingest all the bid requests, but it’s too late. The DSP would have missed out on the opportunity to evaluate all of the bid requests for this broadcast.
This is where the importance of QPS optimization comes into play.
By being able to effectively optimize QPS and filter requests that are of the highest value and quality, a DSP can ingest and evaluate a larger proportion of relevant bid requests, ultimately making better decisions and use of the QPS limits in place.
How to improve efficiency and drive higher ROI for marketers
As our ecosystem intensifies its focus on supply path optimization, where are there opportunities to improve efficiency and drive higher ROI for marketers?
Overall, the industry should look to improve signal visibility and accuracy, while making content object metadata—such as genre or language—available and transparent at scale. This will help provide better inputs for machine learning models to curate a large volume of traffic, improving the ability to optimize QPS.
The more transparency there is into a given opportunity, the better buyers, DSPs, and SSPs can understand its value and effectively optimize. This means DSPs should be transparent with SSPs on what metadata is important to them. And media owners should make signals transparent and visible in bid requests to help SSPs and DSPs properly evaluate them and determine their value.
We also need to rethink how QPS limits are managed, as their static nature today is not ideal for the massive programmatic bursts during live events in streaming TV.
In the future, QPS limits are likely to become more dynamic in nature.
SSPs have a crucial role to play here. Because of their relationship with media owners, they’re typically tasked with seeing the entirety of what the open internet has available. And because QPS limits for DSPs are set within SSPs, they’re in a position to ensure DSPs have access to the open internet, without having to absorb the full cost of doing so, and while also not being disadvantaged when it comes to reaching audiences that are most appropriate for their buyers.
Effective QPS optimization makes for a more efficient and sustainable programmatic ecosystem, which ultimately will generate more value for everyone involved.
Learn more about how our solutions for DSPs deliver higher efficiency and value.
