Editor’s note: In this article, Cameron Church from Watching That talks about Fill Rate Calculations in Advertising (their importance and the correct way to calculate them), followed by an interview by Ben Morrell, Contributing Author at OTTVerse.com.
Table of Contents
Why is it critical for your business that you can calculate your video fill rate?
Every second of everyday media and entertainment companies around the globe produce and publish videos that deliver audiences at a scale that they can sell to advertisers, generating billions of dollars in revenue across the industry every year…
The core product they offer advertisers are known as ad slots, and these make ad breaks that are exposed to every viewer. They come in 3 flavors: Pre-roll (before the content starts), Mid-roll (a break in the content), and Post-roll (after the content finishes).
Knowing how well these ad slots are being sold and delivered is the difference between business success and failure.
One measure of this critical performance indicator is called the Fill Rate: how many times, over a time interval, are the ad slots filled with ads. 100% would mean every ad slot has had an ad, whereby 20% would mean only 2 out of 10 ad slots have had ads.
By observing the dynamics of their fill rate, operational teams can learn what conditions of their setup lead to the best outcomes (most number of ads sold for the highest price).
Calculating the Fill Rates for Video Advertising
If you think about it, calculating the Fill Rate should not be complicated. We defined it in a blog post on WatchingThat.com as
Fill Rate = impressions rendered / ad requests sent by the client where a play request is present.
As the formula shows, you need two pieces of information –
- The number of times the video player requested an ad (the Ad Request).
- The number of times the ad was successfully rendered in response to the request (the Ad Impression).
You plug these two pieces of information into the formula, and you have the Ad Fill Rate.
Well, unfortunately, in the real world, what we just explained is easier said than done.
The trick to getting an accurate fill rate measurement is determining the right data points for both the Ad Request and the Ad Impression.
The problem is that these data points do not come from the same system in the stack. The minute you have to connect two different data sets, you enter a whole world of hurt filled with discrepancies, lag, and blind spots.
However, it’s not impossible (and platforms like Watching That make it turn-key simple), so once you have the measurement bit sorted out, you can now calculate this mission-critical KPI and start understanding how well you’re succeeding in generating ad revenue and how much more you can unlock.
Common Problems in Fill Rate Calculation
Before we can talk about common pitfalls when it comes to calculating your fill rate we need to take a step back and look at the video view’s lifecycle and how the video ad sequence fits into that puzzle.
A video view is a compound of several different parts, as shown in this diagram
- Environment (the code that makes the page, the device used, the network that delivers the video streams);
- Consumption (how much the viewer watches, does the viewer skip through the content, is the volume changed);
- Monetization (does the viewer pay for the view, is advertising used, does the viewer leave before the ad finishes, who served the ad, how much did money did it make your business).
The problem arises from the fact that almost every video delivery pipeline component comes from a different system or vendors – right from ingesting the video, compression, packaging, DRM, origin servers, CDN, playback, apps, ad tracking, ad intelligence, ad insertion, etc. Each of these provides its own LOCAL dataset, representing that component’s specific activity and participation in the overall flow.
The AD REQUEST data point needed for a Fill Rate calculation originates from the Environment part of the video view (from code on the page or in the app that sends the request to the primary ad server).
The AD IMPRESSION data point comes right at the end of the monetization flow. Typically, the ad server has handed off to a 3rd party programmatic partner, which can then hand off several times.
Because the two required data points are separated by several degrees they are impossible to correlate and therefore teams that follow this method need to keep in mind the error margin will be significant.
To combat this error margin, teams sometimes just rely on closest approximations available from the ad server – after all it is central to all ad requests.
The problem here is that the ad server only sees the requests it receives – not what is sent, so it will not know about any issue like ad blocking, network dropouts, etc. It is also not the final delivery system, so we cannot know what happens after the ad request is handed off to another system to finish the ad delivery.
Here teams need to know that their Ad Request count will be lower than the actual sent ad requests, and their Ad Impression count will be inflated from what is actually delivered by 3rd party ad partners. The net result is an inflated fill rate calculation (so they think they are doing better than they actually are).
What are good benchmarks for Fill Rates?
While there is no definitive answer to what a good fill-rate is, we have a useful rule-of-thumb here at WatchingThat.com.
- 0% is terrible and means that the ads are not being shown.
- 25% is a sign that editorial output is out of balance with commercial demand.
- 40% probably means that you have a lot of inventory (high supply) met with low demand.
- 50% is right, and you should be aiming above that.
- 60% is where you want to be if you are optimizing your output.
- 80% ambitious but reserved for specialized inventory groups (specific markets, content verticals campaigns)
- 100% fill rate is perfect and, generally, unattainable.
Interview with Cameron Church, CEO of Watching That by Ben Morrell
Ben Morrell: In this open-ended series, I will speak to key industry players, customers, and experts who have an influence on the outstandingly diverse Asia-Pacific region. It seeks to find out how technology innovation solves key business problems as well as try to get answers to the deeper questions media and entertainment companies are searching for as they launch and expand their own services.
I’d be keen to hear from anyone directly who’s interested in having their say on a wide variety of topics. You can reach me on Linkedin.
Ben Morrell: After one has a fill-rate analysis and the relevant data, what steps can be taken to improve the fill-rate?
Cameron Church: Two immediate findings will come from an accurate fill rate analysis:
- Understanding where you’re not filling and beginning on the journey to understand why not. This leads, ultimately, to an action plan to fix any issues and inefficiencies that turn these unfilled ad slots into impression;
- Understanding where you are filling ads and what conditions of your setup leads to that successful outcome. The journey from here leads to control over these conditions so you can do more of what it takes to lead to these outcomes.
Tackling the unfilled portion of your ad slots / inventory stats with understanding why they are failing – you do this by looking at the various error codes used by the systems to describe why the ad sequence has failed. These sign posts then direct you and the team towards what part of the process is failing and that leads to the ability to fix and repair.
Ben Morrell: I would imagine it’s even more important to analyze the data continuously, not just on a one-off basis. Can you comment on how you have seen the ad-market change during Covid, how businesses have had to adapt and what the future holds for ad-funded content as the marketing dollars rebuild?
Cameron Church: We live in an ever-changing, highly dynamic world, and, as such, teams need to adopt ways of working that champion continuously observing the outcomes of their business activities and processes. Stamping out one spark before it becomes a flame doesn’t mean that the risk of fire is eliminated. Or, more optimistically, just because you missed catching that wave doesn’t mean there is a whole new set right behind it. Video teams need to be constantly on the lookout.
I believe that many teams are already adopting this way of working and because of it the impact of COVID wasn’t as severe as predicted. There are other macro reasons while the media industry was able to better sustain the blind siding of the pandemic, but teams that were already agile and had built observability into their systems and workflows were very able to react quickly and effectively to the trading turbulence that was encountered over the last 12 months.
For a more in-depth discussion on the topic of Fill Rates, please check take a look at the following blogposts on WatchingThat.com