Confused about server-to-server? It might not be relevant for you anyway.


Confused about server-to-server? It might not be relevant for you anyway.

Purch’s announcement in December about their big, bold move to server-to-server has led to a flurry of contradictory articles on the topic. Some experts passionately advocate that this is the future of programmatic advertising. Others question the sustainability of what they argue is a non-transparent quick-fix.

Where does this leave publishers? Well, pretty confused.  

Let’s take a step back. No need to suffer any FOMO attacks just yet! In fact, this isn’t really a new topic, but an extension of the ongoing debate about header bidding and its advantages and inefficiencies. Let’s get to the bottom of what the fuss is about, what you really need to know about server-to-server, and whether it’s relevant for your website monetization anyway.

The basics: header bidding and server-to-server

The principles behind header bidding and server-to-server are the same. They both offer unified auctions, meaning that all demand partners bid at once, in contrast to a standard waterfall. By increasing the competition among the demand partners, publishers incentivize these demand partners to bid higher to win. Done properly, this should translate into higher revenue for the publisher.

There’s one key technical difference. Header bidding auctions take place in the publisher’s browser, while server-to-server bidding occurs on a third party server. That’s a simple difference with a significant impact.

Auctions on the publisher’s browser slow down the page load time. This is the main problem with header bidding: it is a synchronous process which causes page latency. And that’s been keeping publishers up at night. A long page load time negatively affects the user experience, and is increasingly being punished in search results.

Cut to the server-to-server solution. Moving the bidding process away from the publisher’s own browser reduces latency, speeding up the page loading time.

Pros and cons of server-to-server

Server-to-server answers the biggest concern that publishers have with header bidding. It moves the bidding to a server, provided either by an independent ad tech vendor or by one of the large tech companies like Amazon. The bidding is processed faster. Free from concerns about page latency, publishers can now include more demand sources in the auction.

Publishers can sleep sound again – right?

Not quite. The move server-to-server solves one problem at the expense of creating another. When the bidding occurs on the publisher’s browser, the process is relatively transparent for all the parties involved. In stark contrast, the bidding on a server is hidden. Reporting is limited.

This doesn’t sit well with many in the industry who are demanding greater transparency, fraud prevention, and accurate data reporting. It’s very possible that companies providing server-to-server solutions along with their own demand will gain an unfair advantage. To the detriment of the publisher, they could easily use the data to take the last bid and reduce their advertising costs.

Should publishers create their own server-to-server solution?

In theory, if publishers have their own server on which to host the bidding of all their demand partners, this would largely solve the problem. Purch reported that their page latency reduced by half a second while revenue remained constant after they introduced their server-to-server solution. For publishers with immense resources at their fingertips, the road has now been paved and the opportunity awaits.

In reality though, smaller and mid-tier publishers, as well as many larger ones, don’t have the resources required for this. It’s not a feasible solution. Unlike Purch, the vast majority of publishers don’t have large in-house engineering teams nor the scale required to persuade partners to build their end of the connection.

Here’s the challenge for the ad tech industry

It falls to each publisher to know their own priorities when it comes to data collection, page latency, and revenue optimization. Some may be tempted to experiment with server-to-server solutions from large providers like Amazon, and probably in the future, Google too. We’re all for experimentation and seeing what works best. But before you dive in too deep, you need to ask yourself if you really believe that this imperfect solution is the long-term future of programmatic advertising?  

The fuss about server-to-server is less about server-to-server per se, so much as it is the excitement about solving a key challenge associated with header bidding. It’s the excitement about the quest for great alternatives. This quest is ongoing.

Ad tech partners should take inspiration from the various solutions that do exist, and continue in their mission to develop even better ones. We must remain committed to the principles of transparency, improving user experience, and enlarging the pie to increase the value of online advertising for the publishers and brands. Sacrificing one value for another just isn’t good enough.