Guardian News and Media has launched a publisher trading desk to buy inventory from other online publishers, including rival news sites, to allow marketers to target Guardian readers with relevant adverts when they leave theguardian.com.
The trading desk, called Response +, is being run in partnership with demand-side platform (DSP) MediaMath, which provides the technology linking ad buyers with sites selling inventory.
It's a sensible next step for a company that has embraced automated display advertising methods and several other UK publishers have similar schemes.
But is this the only way a publisher of the Guardian's size can sustain itself on online advertising alone?
"We're not just competing with the New York Times or the Daily Mail," Guardian News and Media revenue director Tim Gentry tells us.
"We are competing with the AOLs, Yahoo News and the broader set. We want as much scale as we can gain without sacrificing one iota of quality."
This trend is news publishing's response to two characteristics of the online economy: to make significant money, you need scale and data.
Victims of their own success
Newspapers can now reach far more people than ever before via the web. With 40 million unique users in June, the Guardian is the third biggest news website in the world after MailOnline and the NYT.
Yet the players making the real money from online ads are many times bigger. Both Yahoo and Google reached close to 200 million unique users in the US alone in August.
The new trading desk is run by the Guardian's Response team, which has specialised in selling standard online ad formats through automated channels such as real-time bidding exchanges and private marketeplaces.
The Response team is primarily paid by ad buyers for user responses paid on a CPM basis, but what persuades ad buyers to keep spending with the team, and spend more, is user responses – trial signups, email details, online purchases and other measurable interactions – which advertisers crave as proof their message isn't being ignored.
No matter how carefully you target your ads, only a tiny percentage of impressions served are going to produce these responses. To get more responses, the Guardian needs to serve lots more ads, which is also one of the motivations behind its international expansion.
"[The team] has had a very good year, but they are a victim of their own success," says Gentry. "They've grown revenues at 50 percent over the last year, but we're running out of capacity to deliver the campaigns. Most of their campaigns are measured of on the basis of very hard metrics. The advertisers we are building up are very loyal and want to stay [with us]."
By extending their reach on to other (carefully vetted) sites the Guardian can increase the number of responses each campaign generates. These ads are being bought and served in real time, as the automated systems pick out each desired user when they are being served an ad.
Gentry says he expects revenues from the Response team to be generating "multiple millions" in revenue over the next year.
The Guardian handles 25 billion "data events", pieces of information on who is coming to their sites and what they do there, every month. This is what allows the Guardian sell an ad on another site for more than it pays for that space. It provides the advertiser with its own data on who is looking at that ad.
Gentry says the data isn't enough to give publishers an edge over the truly huge companies such as Facebook or Google, but it does mean they can should be more effective than smaller ad networks and other middle-sized players.
"The best scale players are in as strong a position, because they are able to capture data from the whole universe to capture data which is as rich."
"Where publishers have unique advantage is over the medium scale networks, whose data relationships are based on the partners and their advertiser's sites, rather than over the owned operated."
As usual in the online advertising game, it comes down to a combination of data and scale. The Guardian is looking outside its own sites for scale in order to make better use of its data.