As a key channel partner of Rubicon Australia – our team was invited to attend their regional summit on Thursday 24th August 2017. For publishers, Rubicon are one of the big 3 SSPs/exchanges that dominate the Australian market – so getting an insight into what they’re prioritising gives us a good idea of what’s happening in this market. The 3 key take-aways for publishers from this event were – Prebid is dominating the Header Bidding world and will only get stronger. Rubicon are testing first price auctions – which means advertisers will pay their top bid, not the second price in the auction. This could mean better CPMs for publishers. And Rubicon are trying to make buying easier and more effective with their purchase of NToggle and better filtering for advertisers. Which again could mean better CPMs where you have valuable audiences.
Unusually for this type of event – Rubicon picked an out of CBD location for the summit. Zest at the Spit, Balmoral was a sick choice for a day like Thursday:
Most of the key buy and sell side players in the region were attending – eBay, Spotify, Domain, Cadreon, Publicis, IAB Australia and of course Publift! Rubicon CEO Michael Barrett (formerly of SSP AdMeld which was bought by Google in 2011) kicked things off with an interesting keynote. He highlighted how quickly the digital display industry was evolving right now and discussed how the number of players in their space wasn’t sustainable. As the barriers to entry have been relatively low up until now we’ve seen lots of SSPs/exchanges/networks spring up. This has fragmented the market and made things more complicated for buyers and sellers. Rubicon however, has the luxury of having lots of capital in the bank and is investing this into being ahead of the curve in the Header Bidding and PMP space. So, they are predicting that the market is going to consolidate with ad technology providers like Rubicon reducing to just a few key players. Those key players will strengthen in-region relationships, improve their tech and aggressively reduce their revenue shares to drive other players out of the market. This sounds like good news for publishers – less complicated integrations, better tech and better revenue shares.
The middle portion of the summit addressed advertisers concerns around brand safety and scale. Header Bidding has introduced some complications as well as opportunities for buyers. Transparency into where they are buying has become more difficult with added technology layers and more user based targeting – which worries them when it comes to their ads appearing next to non-brand safe content (think the recent issues with ads appearing next to ISIS videos on YouTube). The volume of impressions available to buyers now, because header bidding also causes them some challenges – they simply cannot process the number of ad requests available so just shut digital campaigns off when they hit a certain volume each day. That’s bad news for advertisers and publishers. Rubicon expects to address this issue with their purchase of NToggle which makes filtering audiences much easier for buyers. On the brand safety front – although Header Bidding has made it harder for advertisers to see where they are buying all their ad impressions, Rubicon is investing more into a safe platform than any other SSP/echange. Only working with the top tier publishers in each country being a key component. IAB’s ads.txt movement should also bring a lot more transparency for buyers and hopefully premium rates for good, brand-safe sites.
Rubicon, Publicis, Integral Ad Science, IAB Australia and REA Group on the Brand Safety panel.
Brand safety for publishers was also addressed. Compared to other SSPs and networks we generally find Rubicon’s creatives very good. We’ve only had a few instances of nasty pop ups or bad ads and they have typically been resolved very quickly by the team at Rubicon. But it was good to hear about the technology they are using to make sure only top-quality advertisers can use their platform. Lots of filtering and automated detection systems to prevent spammy ads coming through.
Spotify provided a really cool distraction from all the display ad talk with their experience of programmatic audio. Yes, it does exist! 10% of Spotify’s AU users are free users – they don’t pay the premium license fee. This means they get ads between the songs and limited search/listen functionality on the app. You probably thought those ads were just the same as the ones you hear on the radio, right? Well – Spotify can pass a lot more information to the advertiser than radio stations can so their audio ads can be a lot more targeted. For example – most people sign into Spotify using their Facebook login. That means demographic data, interests, friends, etc. Add this to mobile app data like usage and location – then you have a pretty useful mix of targeting for advertisers. It’s not there yet – but those free users might be getting invited into the Nando’s they are walking past with a 30% discount on lunch in the very near future. The ad might even hear their name and be offered food that they probably like or have had before. Creepy? Maybe a bit – but very effective and targeted advertising. Programmatic audio is actually Rubicon’s fastest growing product segment in the APAC market.
Getting back to the display stuff that’s relevant to us and regional publishers – the afternoon of the summit focused on header bidding, PMPs and Data.
On the Header Bidding front (is it always a capital H and B? I never know but let’s go with it), Rohan from the Rubicon team gave a high-level view of what Header Bidding is and why it has spread so rapidly. He probably did the best job I’ve seen of explaining with a visual why Header Bidding is so effective – he has kindly shared those images with us below:
So, the image above shows what used to happen in most ad servers. The blue squares are impressions that were allocated to direct orders. The grey impressions went to remnant orders. If we look at the spread of the most valuable impressions we see a pattern like this:
Those red squares are the most valuable impressions on your site. You can see they are spread throughout the whole space. Your direct orders will take all the impressions on the left-hand side (typically the first impressions your site generates each day). There may have been some valuable impressions on the left side that SSPs/exchanges would have paid a very high price for – but they never got the chance to compete for these impressions. The ad server had already allocated those first 30 impressions to direct orders. So, when we add in those valuable impressions SSPs/exchanges bought we get an impression distribution like this:
The direct orders took their 30 impressions. SSPs/exchanges bought some valuable impressions (red boxes) and the rest went to remnant orders. You probably made OK money from your impressions, but you didn’t make as much as you could have. What’s the solution?
You need to mix the 2 models so that you still deliver the high paying direct campaign impressions, but you also sell more valuable impressions to the SSPs/exchanges. That blended model looks like this:
Great – more money for the same impressions. Of course it makes sense for publishers. But why are we only doing this now? Well before Header Bidding it was impossible for SSPs/exchanges to identify the most valuable impressions (red squares) until the ad request was allocated to them by the ad server. It was kind of a lucky dip – where sometimes they would get a great impression (red square) but most of the time they would get a rubbish impression that didn’t generate much revenue (grey square). But with Header Bidding – we allow the SSP/exchange to see the ad request before it reaches the ad server. They can then send a bid to the ad server with the ad request. This extra information means the ad server can now allocate the best impressions to the highest paying order. If an SSP/exchange has a higher bid than a direct campaign – it wins the impression. This is represented by the red squares that are in the blue area. The direct orders still delivered their impression goals – just later in the day.
Rubicon has taken a bit of a shift in their direction with Header Bidding. We partnered with them last year to get the first Australian and New Zealand publishers live with Rubicon FastLane (their Header Bidding product). And the initial results were great. All the publishers saw at least a 30% increase in programmatic earnings/eCPM. But the implementation was difficult and some programmatic deals didn’t work too well with the FastLane zones. This made the product hard to scale with other publishers that didn’t work with an ad tech partner like us or didn’t have the in-house expertise to do the implementation. Rubicon then decided to partner with other ad technologies like AppNexus and Index Exchange to develop the open source Prebid solution. This is a smart move – pooling resources towards a universally accessible product which should be good for everyone (apart from Google – see our blog post on Header Bidding twisting Google’s arm). As the CEO said in the keynote – Rubicon might’ve been a bit slow to the Header Bidding party, but they are now investing heavily to make sure they are market leaders going forwards. Our early testing of Prebid does indeed support this claim – we see them responding to ad requests quicker than nearly all other SSPs/exchanges.
On the PMP side – Rubicon are already leaders in this field, but they are continuing to develop new products and improve the existing ones. They key part here is really the integration of NToggle which will hopefully mean advertisers can much more effectively target the audiences they want to reach even with the large volume of impressions Header Bidding is giving Rubicon access to. When the advertisers can target the exact audience they want – they will pay premiums to reach these users through PMP deals. Rubicon are also making the deal proposal process easier for sellers/publishers – by giving them tools to contact buyers directly and setup deals. Previously the Rubicon Storefront was the only way to do this and you had to wait for buyers to find your inventory and approach you. Soon you will be able to approach them, without having to have a previous relationship in place. Again – this sounds like good news for publishers. Making it easier to attract those high CPM PMP deals.
First Price Auctions
Rubicon are testing first price auctions – this means in some auctions they will take the advertisers top bid as the winning bid. Previously the winner only paid the price of the second bid in the impression auction. Rubicon will only do this when an advertiser has chosen to allow first priced auctions. For the advertiser this means they can bid more aggressively when impressions come from their valuable target audience. And for less valuable impressions they can just use second priced auctions. For example – Mercedes might have 2 audience segments identified to target. People likely to buy a Mercedes in the next 6 months and future customers that might be interested in the next 10 years. They are obviously willing to pay a lot more to show ads to the first segment as they are near the end of the purchase funnel, so Mercedes might use a first price auction with much more aggressive bids for segment 1. And for the second segment – they can target more generically with less action orientated ads which put Mercedes in the thought process when that user is thinking about buying a new car and has the finances to be able to. What does this mean for publishers? If you have valuable audiences with buying intent on your site – you’re probably going to get higher CPMs from the first price auctions as advertisers fight more aggressively to show an ad to them. If you have less valuable audiences on your site, you will probably continue to see the same CPMs that you have previously.
For Data – I wasn’t at that round table but I think they were talking about first and third-party data. Specifically, about what advertisers could do with NToggle’s capabilities.
Simone (Rubicon Country Manager for Australia & New Zealand) wrapped up the summit with a quick overview of the key discussions. All in all – a very good ad tech event. We learned that most of the major SSPs/exchanges are heavily investing in and backing Prebid. We began to understand how Header Bidding is affecting buyers. And we heard about some interesting new technology like programmatic audio. A good way to spend a sunny Thursday. The drinks were good, especially with the views over the harbour. Unfortunately, as the designated driver I could only have Coke Zero so didn’t stay for long.