Traffic shaping has been used within the ad tech industry for a while, particularly by supply-side platforms (SSPs) and demand-side platforms (DSPs) filtering out available programmatic ad inventory that isn’t quite up to snuff before passing them along to media buyers and advertisers.
Recently, however, a few publishers have started taking traffic shaping into their own hands for a variety of reasons, but with the ultimate goal of only putting their most premium ad inventory up for sale.
“If we are sending texts that don’t get answered up to the buy side, we need to curb that behavior… so that we can be better partners to those working to represent our inventory for us. Ultimately, if I send more bid requests to an SSP partner that can be won, I become a more profitable publisher partner to my SSP because they no longer have to filter out as many unwitting or unsellable requests from me,” said Justin Wohl, CRO of Salon.com, TVTropes.org and Snopes.com.
WTF is traffic shaping?
Traffic shaping can be boiled down to publishers being more thoughtful about what ad requests are sent to intermediaries like SSPs. But it’s a rather “squishy” term, according to John Shankman, founder and CEO of Hashtag Labs, a company that provides ad ops solutions to publishers as well as tools that enable publishers to undergo traffic shaping.
For the majority of time that traffic shaping has existed, it’s been left up to SSPs and DSPs to handle, which they did by filtering ad requests based on categories — like geographies, browsers, devices, etc. — that advertisers ask for specifically. Now, publishers are either building the capabilities in-house or using tools provided by companies like Hashtag Labs to start filtering and reducing the number of ad requests they send out in the first place, to unclog the programmatic pipeline, so to speak.
“The publisher has always been told to send out as many ad requests as you can,” said Shankman. “And really, traffic shaping is the opposite of that in the sense of, let’s think about where we send out ad requests and what we send them out for.” He added that this is an evolution of the industry where publishers are finally able to have some access to those controls.
What does traffic shaping look like?
For a publisher, traffic shaping can come in a few different forms.
- Requests by bidder
- Publishers can add parameters to the prebid code used to describe their ad units to SSPs to determine how and where their ad requests get sent out. This method allows a publisher to filter its ad requests by geographic location, device type and browser, which are lower hanging fruits, but they can go deeper into things like connection speed.
- No ads if idle
- Using their first-party data, publishers can determine when their users are active on their site, and if they measure inactivity for a prolonged amount of time, they can turn of ads on the page. This ensures that ads aren’t getting sold that will ultimately not be viewed by a user.
- Throttling/Smarter requests
- This is a form of performance-based traffic shaping where a publisher can pause ad requests that go to specific SSPs if they’re under-performing or continually sending bids that are below the publishers’ price floor for a prolonged period of time. After a certain period of time (which the publisher can experiment with), the SSP gets turned back on to see if their bids improve.
Hashtag Labs currently offers its clients two ways to implement traffic shaping: requests by bidder and throttling. Unwind Media, which publishes gaming sites Solitaired.com and Solitairebliss.com, has tested and implemented one of those strategies (by another name called “smarter requests”) as well as another strategy called “no ads if idle.”
How do publishers choose which form to use?
Emry Downinghall, svp of programmatic revenue and strategy at Unwind Media, said that so far the tests for no ads if idle and smarter requests have yielded positive results, without having a negative impact to revenue or performance.
No ads if idle
The sites that Unwind Media operate are casual gaming sites like Solitaire and FreeCell, which by nature gives the company a good amount of data into when site visitors are active or inactive on a page. If a user has been inactive for more than one minute, Downinghall said they stopped serving them ads.
“We saw a 20% decline in desktop ad requests overall … And we saw a 12% increase in overall viewability for desktop,” Downinghall said. “A 12% increase in the viewability is meaningful and that was a great sign that can be linked to performance improvement, but I didn’t go into this test thinking we’re going to improve viewability. It was just an outcome.”
Overall, Unwind Media did serve fewer ads as a result, because if you’re sending fewer ad requests to SSPs, it’s natural that you’ll end up selling fewer ads. However, those ads that were dropped were already being sold “below the price floors that we were setting,” so it didn’t impact revenue, Dowinghall added.
As a warning to publishers, Dowinghall said he could see this test having a more negative impact on revenue depending on the site, but it’s all a function of A/B testing to figure out which factors and variables make the most sense.
Unwind Media sells its ad units by individual user and session. For each unit, his team set up two parameters to determine whether or not it would continue to send ad requests to its SSPs. The first parameter was a specific period of time (in his case 3 minutes) and the second parameter was how many times an SSP responded to an ad request with a price below the company’s price floor (in his case 3 times).
“If an SSP fails to respond three times in a row over three minutes, we stopped calling that SSP for that user,” said Downinghall, adding that after a period of time (which his team has experimented with in various testing phases), they will return to said SSP and start sending bids again in the event their responses improved.
Requests by bidder
Shankman has not yet published the findings for what results this traffic shaping will yield but said that on average, he’s measuring similar results to what Downinghall has reported.
Is it only used in open marketplace inventory?
No. According to Shankman, “being thoughtful about when your ad requests go out is a very broad statement … and I do think it can really apply to anything that’s done programmatically.” For example, publishers can say they only want to send Safari requests to private programmatic marketplaces.
Why should publishers do traffic shaping?
According to the executives at publishers and ad tech firms interviewed for this story, traffic shaping can help publishers to increase CPMs and lower carbon emissions, among other benefits.
- Revenue optimization
- Win favor with SSPs
- Improved user experience
Because traffic shaping is arguably demand path optimization, it not only decreases the amount of computational power that is used and thus reduces carbon emissions within a publisher’s scope three output. But it also reduces the amount of work that SSPs need to do, in theory, which bodes well for publishers in that relationships dynamic.
What’s more, Downinghall said that fewer ads running on a site improves UX. “Our product team loves the idea of more efficient ad requests, because as the user goes deeper in the session, if we’re turning off bidders, the load on the page is lessened.”
Will this improve revenue immediately?
Ultimately, Downinghall, Shankman and Wohl believe that traffic shaping on the part of publishers will have a positive impact on programmatic advertising revenues, but all cautioned that this really isn’t a method for short-term revenue gains. Rather, it’s a way to initiate longer term efficiencies within the programmatic bid stream that should lead publishers to a higher win rate and higher CPMs.
“I never thought of implementing these things as [a way] to increase CPMs immediately or in the test cell. I always focused on just breaking even,” said Downinghall, adding that over time, improving the efficiency of what bids are sent to SSPs and then downstream to DSPs, will “increase the profitability of a publisher’s profile” to those parties. As long as it hasn’t had a negative impact on revenue – which Downinghall claims has not been the case – he considers these tests to have been a success.
Wohl on the other hand said that his brands have experienced a short-term gross revenue hit of “a few percentage points” in the months of April and May when the initial traffic shaping tests were first underway. But as soon as September, he said he expects these efforts to “pay dividends to us” to make traffic shaping well worth it from a revenue gain perspective.
“If we continue to prune in this direction and figure out what bid requests are not worth sending or not worth selling as a publisher, we can incrementally charge more for the ones that are worth doing,” said Wohl. “There is inherently a margin that publishers give up if we’re selling a really good opportunity, at the same starting price, as what we should know is a pretty bad opportunity.
More in Media
Adalytics Research asks, ‘Are YouTube advertisers inadvertently harvesting data from millions of children?’
Publishers’ Q2 earnings reveal digital advertising is still in a tight spot, but digital subscriptions are picking up steam.
Experts reflect how the failures of social media and online advertising can help the industry improve the next era of innovation.
Ad position: web_bfu