The prognosis for programmatic video looks pretty bright: eMarketer estimates that ad spend will hit $5.37 billion in 2016, which is more than 50% of total predicted digital video ad spend ($9.59 billion). However, the current situation is a bit cloudier: issues with latency and standards are making premium publishers wary of the channel, especially when their pre-roll inventory is already limited.
To improve delivery and efficiency, tech providers like VertaMedia have developed algorithms to avoid dreaded VPAID errors and improve fill rates. Advertising Director Alex Volker took me on a trip through the weeds of programmatic video, while also discussing the potential for outstream and what makes him giddy about the latest VAST update.
Gavin Dunaway: Issues with VPAID errors have long been a thorn in the side of publishers jumping into programmatic video waters. How does VertaMedia address this issue?
Alex Volker: As long as we are living in the real world and have to play by market rules, we have use cases when a VPAID error can appear, though we are doing our best to decrease latency even in the case of errors. We apply a wide range of pre-targeting rules to reject unmatched traffic at the XML-request level – this logic includes our automatic waterfall, which analyzes a wide variety of traffic characteristics and eventual fill rates from demand partners to create a traffic portrait.
Another common reason for ad cancelation is real-time third-party decision-making logic (e.g., viewability analysis, brand safety). Our brand safety function has its own quarantine module based on various scoring systems and proprietary logic, so if we find traffic somewhat suspicious, we won’t serve the VPAID.
On the client side, we are doing our best to speed up the waterfall by applying third-party decision-making in an asynchronous fashion. This means that partner tags will start to initiate regardless of our decision-making flow, while all the third-party logic will be launched simultaneously.
GD: In building your efficiency forecast, what variables (for example, fill rates, time to response) tend to be most important, or does it really differ from campaign to campaign? Have you had any surprising findings or seen any interesting trends?
AV: Response time is definitely the most critical factor because high latency kills the opportunity itself. Besides response time as a key variable in building waterfalls, we`re always looking for alternative ways to compensate for latency and increase fill rates.
A 100% fill rate could easily be achieved if users spent enough time on a page and the process of ad uploading would not block the target video content delivery, but this is not the case today. The reasons these issues are not solved yet are diverse – for example, regular players are programmed to pause the main video while searching for an ad, which makes the corresponding rate of ad rejection rise.
We are developing several products to solve these issues, including our in-house video player and the plugin for video.js to move forward the process of ad preloading in the background and therefore not blocking the main video content. This model is expected to be integrated within our outstream solution.
However, this is rather a workaround than the true solution. We expect that VAST 4.0 will solve some of the issues. We already have some interesting ideas on different partner integration models to speed up processes based on new opportunities within the protocol.
GD: If VPAID is updated, what should be the central fixes?
AV: We would consider providing more non-blocking time for VPAID execution in the integration toolset. We prepare our partner tags up to the “adsReady” state in background. Though, to make this behavior common practice, each unit should know how much time it has to prepare. This would be a really handy part of the protocol.
GD: Some video content providers (besides YouTube) actually avoid adopting VPAID because they feel it gives too much control of the player to the buy side. Do you think this is a mistake, or is it the right move for certain publishers?
AV: As long as we think that full player control is strictly needed for creating real interactive ads, we are concerned with VPAID ads having full access to browser context. There should be a separate bridge for scoring and viewability analysis systems that require this level of access, while the ad itself should be limited to player environment.
GD: What do you like and dislike about the proposed VAST 4.0 update?
AV: Although VAST 4.0 brings a lot of enhancements to the standard and we are planning to integrate in the upcoming month, we are slightly skeptical about the buzz around it. It will definitely make things easier with post-impression analysis, though it doesn’t answer the question of how to work with pre-impression APIs based on real-time scoring.
GD: Could you explain what that means and why it is an issue?
GD: What are your thoughts on the potential for outstream video? Do advertisers see the value?
AV: The outstream video ad format is actively gaining popularity – about 80% of our clients have plans or have already allocated budgets for buying outstream video inventory. Outstream offers our clients access to complementary inventory and ad placements, which could be highly effective for certain types of brand awareness campaigns. In addition, this ad format has demonstrated high performance in decision-making logic based on viewability metrics.
We are launching an outstream module as a part of our Videe.tv product and as soon as we get feedback and usage statistics, we will most likely provide an integration for our SSP clients.
GD: Private marketplaces in video have proved pretty successful, but there are a lot of questions about the viability of the open programmatic marketplace – what does this space need to see additional trading?
AV: Indeed, private marketplaces have proved highly efficient and here at VertaMedia; we see real value in this highly promising niche of our business. In particular, we are pushing forward with PMP development and strive to arm our clients with more options than standard PMPs provide.
In addition to ad serving technology and a pool of advertisers, VertaMedia provides SSP/Videe.TV clients with access to a library of premium video content delivered via our streaming servers. That way, we have control over not only ad placement but also video content metrics on the publisher side. This service is particularly advantageous for advertisers, who can target both video placement and content before an ad us served. It gives more opportunities for advertisers and provides a higher level of brand safety. We call this ePMP, or extended PMP.
There are still a lot of issues surrounding the open marketplace: transparency, brand safety, the quality of inventory available, etc. At the same time, the open marketplace allows access to a huge amount of inventory, which is a key aspect of our business. However, we are working hard to reduce these pains with our own in-house technology.
GD: Is mobile programmatic video finally becoming a hot space? What will it take for this channel to mature? What about advertising on connected TVs and OTT devices – is interest in this area ramping up?
AV: Mobile programmatic is growing, seeing more and more budget. Despite the fact that our technology is concentrated on desktop screens predominantly, we can’t ignore rising mobile programmatic trends. This segment is highly dynamic, raising many issues alongside its evolution such as infrastructure development. However, we expect them to be solved by the end of the year.
Video content distribution through connected TVs is actively growing and we don’t stand aside there either. Our platform is already fully compatible with connected TV devices, though we still put a lot of effort into the development of functionality.
GD: It’s early in the year, but what do you think will be the biggest trend in programmatic video this year?
AV: I believe that the primary attention will be focused on VAST 4.0 and Header Bidding. Many expect that these technologies will solve a number of issues that grow essentially.
GD: Are you already embracing header bidding for video? Is the chief concern latency?
AV: Not yet, we are still in the middle of implementation and plan to launch around March. Publishers demand for this functionality is growing, so we are doing our best to provide it ASAP. From the market perspective, we really like the idea of header bidding as it will decrease latency, and does allow for some real-time decisioning logic in browser context, though we would like it to be a part of VAST protocol.