The Need for Speed: How Video Load Time Affects Ad Delivery
June 16, 2017

Shlomi Gian
PacketZoom

Share this

Ask any mobile app developer, and they'll tell you that one of the greatest challenges in monetizing their apps through video ads isn't finding the right demand or knowing when to run the videos; it's figuring out how to present video ads without slowing down their apps.
 
I'm sure every developer at some point has done a Google search for something like, "Which ad networks are optimized for mobile apps?" or "Which mobile ad network is the fastest?"
 
If the primary revenue you get from your app is ad generated, speed is hardly inconsequential.
 

The Developer Challenge

 
One company explained their ad situation like this: After a user completes a level, the user either goes into their store and browses around for a few minutes or goes to the next level within a few seconds. The developer's goal is to show a video ad right before the next level starts, but they've found themselves in a pickle. Most developers will cache their videos a few minutes before showing them, and the expiration is around 15 minutes. If the developer starts the video download as soon as the user finishes the current level, and that user then browses the store for a few minutes, the ad could expire before it can be shown, i.e., right before the next level.
 
On the flipside, if the developer starts the video download shortly after the user finishes the current level, and that user skips the store to go immediately to the next level, then the video has not downloaded yet and cannot be shown to the user before the next level. In this case, it would be ideal for the video to be delivered consistently within a set amount of time. But since ad retrieving timeout is around 300-400 milliseconds, even when you cache the video, slow ad retrieval leads directly to missing revenue.
 
Net net, while video pre-caching is a valid technique, it comes with a cost, and often times expired content would still result in real-time video downloads that usually are not fast enough.
 

Compounding the Challenge: Painful Disconnects

 
Connection drops are another issue that affect video ads. Disconnects happen all the time, especially when users are on the move, such as when they're commuting or walking around a busy city and their connections must transfer to a different network type. If a disconnect happens during an API call to an ad network or while downloading a large video asset, then that ad will fail to load.

When disconnects happen, the end user typically doesn't realize that the fault lies with the networks. Instead, they usually assign blame to the developer and the app itself. Many users will get so frustrated with a dropped connection that they won't even put in the brief time it takes to reload an app.
 
Some networks are faster and more reliable than others. Verizon and T-Mobile top the US for speed, and lag slightly behind AT&T in disconnects. The problem of disconnects is particularly acute in developing countries that don't have reliable cellular and WiFi networks. But it can infest developed countries as well. According to our benchmarks from February of this year, apps users in Russia, Indonesia, Germany and Brazil all suffer from more than 10% disconnects on average, with disconnect rates as high as 30 percent on 2G networks.
 

Examples of Various Ad Download Times

 
The charts below show ads within a popular news app. The top chart shows video ads delivered by SuperSonic and the bottom chart shows those delivered by Vungle. The x axis shows the amount of time a video took to download in milliseconds; the y axis shows the percentage of the total transfers that finished within that bucket of milliseconds. What we can see overall is that ads aren't delivered within a consistent amount of time; they're usually delivered anywhere from half a second to two seconds.





 
What this says is that if you use a mediation ad network, you should expect your ads to load anywhere from less than one second up to four seconds or longer. That doesn't help you to engineer your app if it has time constraints like the example above.

So what do you do? Well, if you choose to just use one ad network, you should be able to engineer your app/game to that specific ad network based on how long that specific network loads ads. Or, you could continue to use an ad mediator with the understanding that some percent of your users will not see the ad due to it loading too fast or too slow.
 
Bottom line is, attention shifts when something takes longer to load, and time is money. So keep this in mind with your in-app video ads.

Shlomi Gian is CEO of PacketZoom
Share this

The Latest

October 04, 2024

In Part 1 of this two-part series, I defined multi-CDN and explored how and why this approach is used by streaming services, e-commerce platforms, gaming companies and global enterprises for fast and reliable content delivery ... Now, in Part 2 of the series, I'll explore one of the biggest challenges of multi-CDN: observability.

October 03, 2024

CDNs consist of geographically distributed data centers with servers that cache and serve content close to end users to reduce latency and improve load times. Each data center is strategically placed so that digital signals can rapidly travel from one "point of presence" to the next, getting the digital signal to the viewer as fast as possible ... Multi-CDN refers to the strategy of utilizing multiple CDNs to deliver digital content across the internet ...

October 02, 2024

We surveyed IT professionals on their attitudes and practices regarding using Generative AI with databases. We asked how they are layering the technology in with their systems, where it's working the best for them, and what their concerns are ...

October 01, 2024

40% of generative AI (GenAI) solutions will be multimodal (text, image, audio and video) by 2027, up from 1% in 2023, according to Gartner ...

September 30, 2024

Today's digital business landscape evolves rapidly ... Among the areas primed for innovation, the long-standing ticket-based IT support model stands out as particularly outdated. Emerging as a game-changer, the concept of the "ticketless enterprise" promises to shift IT management from a reactive stance to a proactive approach ...

September 27, 2024

In MEAN TIME TO INSIGHT Episode 10, Shamus McGillicuddy, VP of Research, Network Infrastructure and Operations, at EMA discusses Generative AI ...

September 26, 2024

By 2026, 30% of enterprises will automate more than half of their network activities, an increase from under 10% in mid-2023, according to Gartner ...

September 25, 2024

A recent report by Enterprise Management Associates (EMA) reveals that nearly 95% of organizations use a combination of do-it-yourself (DIY) and vendor solutions for network automation, yet only 28% believe they have successfully implemented their automation strategy. Why is this mixed approach so popular if many engineers feel that their overall program is not successful? ...

September 24, 2024

As AI improves and strengthens various product innovations and technology functions, it's also influencing and infiltrating the observability space ... Observability helps translate technical stability into customer satisfaction and business success and AI amplifies this by driving continuous improvement at scale ...

September 23, 2024

Technical debt is a pressing issue for many organizations, stifling innovation and leading to costly inefficiencies ... Despite these challenges, 90% of IT leaders are planning to boost their spending on emerging technologies like AI in 2025 ... As budget season approaches, it's important for IT leaders to address technical debt to ensure that their 2025 budgets are allocated effectively and support successful technology adoption ...