The State of Server-Side Ad Insertion 2019

Ad blockers can detect ad calls made by the client to a known ad-serving URL. If ads are stitched on to the stream before the content reaches the client, then the ad blocker has less chance of detecting the ad call and can’t block the response, in theory.

Vendors are scrambling to architect signaling and beaconing appropriately to get at proper attributions. If you don’t have systems that meet that scale, you run into issues where you don’t fully monetize your audience on behalf of the publisher, so that’s one issue at the foundational level that needs to be solved, says Friend.

“The challenges we see are that [while] an ad blocker may not block the playback of the ad, it (can) block the ad impression audit log,” says Oreper. In this worst-case scenario, the analytic system says ‘Hey, I just played an ad,’ the end user sat through an ad, but the service provider is never going to get paid for that impression because it was blocked by the ad blocker.

Scale as Fraud

“If you have an audience at scale you need to have complementary infrastructure,” says Friend. “What that means is you have to have CPUs behind all this stuff, which is different from providing the content stream, because the content stream is one-to-many. Dynamic ad insertion is one-to-one.”

“Let’s just use a single game as an example, where the game is way better than anybody anticipated and therefore the size of our audience has grown larger than what we sold against—we rely on the programmatic markets to help us f ill this space,” says Friend. “The issue is, however, that to a lot of the programmatic algorithms, when they see sudden scale, it looks like fraud to them. So suddenly at scale they say ‘Hey all this traffic is from a single source, it’s probably bad. So just ignore it.’ That’s one of the issues that they need to address.”

Friend also uses a hybrid/linear model which can toggle a single ad position (or however many you determine) to be SSAI or linear pass-through. “If your audience scales larger than the inventory sold, the ad unit will switch to linear pass-through so you can get the attribution for the ad without serving all your marketers’ paid digital inventory in a single event,” says Friend. Translation: He uses and gets attribution from running broadcast ads on digital when his digital inventory has been used up.

SCTE Confusion

SCTE markers indicate cue points for ad breaks in live environments. “Customers just need to be aware of whether they’re outputting SCTE 35 or SCTE 104 and which of those (potentially both) is accepted by their SSAI vendor,” says Broadstone. The SCTE marker embedded as part of the stream includes information about when an ad break starts and how long it is. Traditional broadcast has two types of SCTE markers—104 for baseband (HD-SDI) and 35 for IP video.

“Our systems are driven by SCTE for the cue-in, cue-out, so we spend a lot of time and energy on this, both with internal systems for monitoring and quality control,” says Magnusson. “We also work with upstream partners to ensure that the signaling we get is accurate and conforms to specification as well as contractual agreements for both sides.”

Cloudy With a 100% Chance of Ads

The Weather Group owns The Weather Channel, and has created an app called Local Now that provides hyper-local weather, traffic, news, and sports, running SSAI. “Now step two is to do server-side ad insertion where we don’t have control of the player—that’s basically with our distribution partners or the virtual MVPDs [vMVPDs],” says Victor Ortiz, senior technical product manager, The Weather Channel. The company wants to use SSAI to monetize across its distribution partners.

The challenge: Sending out tracking information in an SSAI environment to his advertisers about viewers, so that he can pitch and price the inventory without wasting impressions. Distributors typically don’t share the audience information, which he needs to target advertiser segments.

Economics

“YouTube and Facebook together account for just over a third of global in-stream video advertising revenue, meaning that 60%+ of global in-stream video advertising revenue is accounted for by other platforms [including owned and operated and OTT-delivered vMVPD content],” according to Matthew Bailey, senior analyst, Media and Entertainment, Ovum.

“We’ve seen customers double revenue by recovering lost revenue due to ad blockers using SSAI,” says Broadstone. “With the stats that our customers have shared with us, they’re seeing something like 40% to 50% more ad impressions after they enable server-side ad insertion compared to client side and that’s because it defeats ad blockers in most cases,” says Oreper.

IAB Tech Lab Weighs In

The IAB Tech Lab’s recent Open Measurement SDK is also pushing more standardization on measurement, where the SDK can be used to replace the need for one from each vendor in the ad-tech stack. “In the context of SSAI, we have added some updates in Video Ad Serving Template (VAST) 4.1 that will help with transparency around requests and with tracking signals. For example, the ‘SERVERSIDE’ macro indicates whether the signal is coming from the tracking call, or directly [from] the server, or by the server on behalf of the client,” says Amit Shetty, senior director, Video & Audio Products, IAB Tech Lab.

“Another example is definition of the HTTP headers that should be included by ad stitching providers in SSAI scenarios to avoid being considered fraudulent traffic, and to provide more context of the client.”

The VAST (video ad serving template) 4.1 standards released in November 2018 (go2sm.com/vast41) recommend the following: “The separation of the media file from executable code, the use of a mezzanine file to help with transcoding (and support more devices/ environments), the use of a ‘UniversalAdID’ to identify assets, direction regarding tracking, standardized ad requests, etc.,” Shetty says.

Why SSAI Math Is Hard to Do

Let’s now look at vendors’ opinions about reaching scale. “I would say anecdotally from my experience with customers we see maybe 30 or 40 percent of the time server-side ad insertion is used as opposed to client side. It is definitely growing,” says Oreper.

“Any virtual MVPD that you can name (like Sling and Hulu), they’re all server-side ad insertion,” says Allen Klosowski, SVP, Advanced Solutions Group, SpotX. “All the major broadcasters are pretty much using server-side ad insertion at this point, particularly for long-form content.”

“We just pulled a report, and about 52% of the ads served to our premium OTT clients (broadcasters, MVPDs, etc.) is enabled using SSAI,” he says. “It’s a hard estimate, because unfortunately my top line reporting mixes in short-form as well, like the pre-roll videos on desktop sites ... and I think that skews it downward. I’d still say almost every conversation we have about long-form content (live, linear, or VOD) includes a discussion of the SSAI platform and integration path.”

“For Verizon Digital Media Services, the vast majority of ad-supported content we deliver is with SSAI, and we do it successfully at scale. Exact percentages of server-side vs. client-side ad insertion rates are complex to calculate, as it’s possible to perform client-side insertion in parallel to SSAI,” says Sidi. “[The technology] has been architected to handle broadcast viewership numbers online, scaling to support tens of thousands session starts per second and millions of concurrent viewers.”

“We’ve scaled for events forecast for audiences over 5 million peak concurrent users, where every user is getting a unique set of ad decisions, unique ads and unique tracking,” says Sewell. This includes handling highly dynamic break patterns and a combination of first-party and programmatically sold advertising. “The architecture we have can scale to audiences in the tens of millions.”

Other Innovations

SSAI is being used at events like Formula 1 to replace content signage. Viewers at the event will see one type of graphics, broadcast viewers a second version, and online viewers a third creative, says Steve Miller-Jones, VP of product strategy, Limelight Networks. “Other data, whether it’s statistics in a data feed, advertising, or sponsorship can be swapped out so different viewers see different things.”

“The lesson that the industry has learned from [the virtual MVPD] is that if you are a major cable company you can do broadcast quality delivery of content using IP in a personalized manner,” says Klosowski. “There’s no reason that you can’t actually do that inside of your cable environment using identical technologies, and you can drastically bring down your costs.”

Traditional cable companies are all looking at how to move to HLS and DASH for their content delivery. “The future is coming very, very quickly, because we’re already seeing companies make these types of investments,” says Klosowski. The result would be the broadcast environment following what’s been happening in digital. Being able to deliver SSAI on digital and also deliver into a linear feed truly would be SSAI at scale.

[This article appears in the March 2019 issue of Streaming Media Magazine as "The State of Server-Side Ad Insertion."]

Get instant access to our 2019 Sourcebook. Register for free to download the entire issue right now!