A BLOG POST BY CO-FOUNDER CHLOE GRUTCHFIELD
A couple of weeks ago, we hosted a breakfast (well, more like a therapy session) with some of our publisher friends to talk about a few of their most challenging frustrations, in the bid to brainstorm ideas for improvement - not only for them as businesses, but for the end user (and the industry as a whole).
The hottest topic by a mile was the viewability and brand safety paradox; how both are having a huge impact on:
the volume of impressions publishers can monetise
the performance of the website (from a latency perspective)
In parallel, we spoke to leading industry publication, Digiday who were keen to understand our findings on viewability and brand safety - and how they are both impacting the performance of websites. So we shared a few stats from our insights tool DIAGNOSE to tell the publisher story.
The reaction from the Digiday article was huge; industry spokesmen from all different regions, backgrounds and businesses were all singing from the same tune on these hot topics. So, we wanted to share a few more thoughts and ideas on how we can collectively solve this problem.
Creatives are getting complex!
Over the last three months, we’ve analysed a number of ad tags (the piece of code that will trigger every piece of content related to the ad) to understand how ads perform on a blank HTML vs. on a live website. An ad may look great and load fast when being previewed on its own but websites are complex with an average of 400 HTTP requests/responses from dozens of different vendors all loading content, ads and track audiences and performance on a website.
And wow! Ad tags have become big heavy beasts too: the ad content itself, a brand safety tag, a viewability tag, DMP tracking, impression tracking, 3rd party footfall measurement - it’s all happening in unison. I’m not exaggerating. In some extreme cases, we’ve seen 30 different tags associated to an ad and a five seconds lag on a live website between the first tag related to the creative and the advertiser ad server tag recording the impression.
Those five seconds mean A LOT of discrepancy that ultimately punish the publisher.
The viewability paradox
The objective of viewability is to assess whether an impression was in view or not. The paradox lies in the fact that the viewability (and brand safety) tags that load before/after/as the ad is loading are putting pressures on websites, making ads in general less likely to be viewed. On a live website where loads of tags are firing at the same time (asynchronously) or not, what looks like a 200ms lag on a blank HTML can be five seconds on a live website!
What does DIAGNOSE say?
Here at RedBud we look at data all the time. The insights from our DIAGNOSE tool are helping our clients improve their ad tech stack to maximise their revenue. The stats that follow are based on a one month sample across 20 of the top 100 UK publishers and across multiple types of pages; Home, Category and Article.
On average, we found that 10% of a website’s weight - which includes visible content like ads and editorial content as well as invisible content like tracking pixels - is caused by viewability and brand safety HTTP requests/responses. In some extreme cases, it can go up to 21%.
When looking at the total duration of every single HTTP request/response on a web page, whether it's loading visible content (such as editorial content or ads) or non visible content (tracking pixel, brand safety tags etc.), brand safety, viewability and verification requests/responses summed together represent an average duration of five seconds, sometimes more. Most tags are firing asynchronously though - so when adjusting that figure and looking at the total contribution of those vendors on the total load of a website (from first request to last request, before the ad refreshes), we're looking at an average of 1 second - in some instances much, much more.
Granted, in many cases those technologies' requests/responses are happening once the editorial content has loaded. But they are competing with other content/tags that need to load (the impression tracking pixels of the ads those vendors are measuring against, other ads, insights & analytics tracking etc.).
Their bloated duration and weight is contributing to poorer viewability scores for publishers (we think the viewability stats shared with the buy side may be underestimated), higher discrepancies between buy/sell side and across systems and lost revenue for publishers.
One of our key values at RedBud and one that Rhys, Fi and I have been living and breathing for the past three years is solutions, not problems.
We’re clearly not suggesting to get rid of viewability all together. It has become a common currency in the market and key for advertisers and agencies to determine where the eyeballs are and how to improve their creative and targeting strategies. But, it needs a bit of rethinking because currently it’s clearly not working well for publishers.
Here are a two initial ideas we’ve had to start ironing out the viewability and brand safety issues discussed in today's post:
Limit the pressure on websites: Can we look at reducing the pressure on websites by firing viewability and brand safety tags across a sample of impressions instead of all impressions?
Adjust the viewability score: Can the latency impact of complex creative be part of the viewability rate calculation i.e. the viewability stat is adjusted to take into account the impact of the tags associated to creatives so publishers are not penalised by heavy creative
We’d love to get vendor’s feedback on our suggestions above, so get in touch here to let us know your thoughts.