Lazy loading isn’t something new. Publishers — big or small are quite familiar with the technique. In fact, Google Chrome has browser-level support that’ll help publishers to lazy load ads and images — just by using “loading” as an attribute and many other browsers support the same.
So, it’s fair to assume you have implemented it in one way or another in the past. But what’s surprising is, even with its wide-spread adoption and familiarity, we often see that many generalize the impact of lazy loading ads and oversee the nuances involved in it.
Yes, lazy loading ads can increase viewability, improve page speed, and typically results in a reduced number of ad impressions (as you are only requesting and delivering the ads when they are near the user’s viewport).
But that’s not the end of the story. It’s not as simple as it seems — when you slice the results based on geographies and devices. In this post, we’ll walk you through the impact and derive inferences that can help you come up with a better lazy loading strategy.
We believe once you reach the end of the article, you’ll take a methodical approach towards implementing lazy loading — rather than just implementing it across the board.
Sidenote: If you’re planning to lazy load ads to improve user experience, then go right ahead. This study is aimed at the publishers who’re implementing lazy loading to increase eCPM and thus, overall revenue.
Let’s get started.
First, we’ll slice the data and look at the impact of lazy loading on eCPM and viewability across two sets of geographies:
(a) Target Geographies:
As the name implies, target geographies are the countries that you target — with your content and attract the majority of your users from. If you’re a US news publisher, then you’ll cover news related to the US and get most of your hits from the States. So, U.S is your target geo.
Here’s the impact of lazy loading on target geographies.
As you can see, eCPM jumps by an impressive ~25% and ad viewability by 8.35%.
(b) Non-Target Geographies:
Non-target geographies are the exact opposite of target geographies. That is, countries you aren’t targeting with your content and tend to get less traffic from. If we use the previous example, except the U.S, any other countries can be non-target geographies (say, the UK).
In non-target geos, eCPM increased just by ~5% but ad viewability jumped by more than 12%.
It’s in contrast to the target geos. To put this into perspective, for a 1% increase in ad viewability, eCPM jumped by 3.1% in target geo and just by 0.4% in non-target geo.
The next question is why? Well, we need to address this in two parts:
- Why viewability increase is higher in non-target geo and
- Why eCPM jump is higher in target geo.
– Viewability: The impact of lazy loading highly depends on the user behavior on the site. The number of users from your target geo tends to be far more than that of non-target geos. So, this means, your chances of having microsessions (user sessions lasting less than 15s) increase dramatically in target geo. Having more microsessions means more bounce rate, lesser time spent on the pages (skimming), and this leads to a lesser impact of lazy loading on ad viewability.
To put it simply, if you are a US publisher, you attract almost all of your traffic from the US and lesser traffic from other countries. So, more traffic means more chances of readers just skimming the content and bouncing off which decreases the effectiveness of lazy loading.
In non-target geo, the effectiveness of lazy loading tends to be better as you’ll have a lesser number of microsessions.
– eCPM: Now, eCPM. Why would eCPM increase a lot on target geographies when the viewability doesn’t?
It’s because of three factors: CTR, ID’ing users, and target audience.
Users from target geo are likely to be the right target audience and buyers will have a better chance at ID’ing the users as they would’ve been bidding on the other sites the same users visit.
Most importantly, CTR is higher in target geo. You can pull the CTR of the users from non-target geos and compare it with the CTR from the target geo users, you’ll be able to derive the conclusion yourself.
If your viewability is already above average, start by implementing lazy loading ads on target geographies. Because even for a slight jump in viewability, you can get comparatively better results in terms of eCPM and revenue.
Now, we’re going to look at the impact of lazy loading on eCPM and viewability across mobile and desktop.
Lazy loading ads had a better impact on mobile ad impressions — eCPM increased substantially (~30%) and viewability by 7.96%.
On the other hand, lazy loading ads on Desktop improved ad viewability but didn’t push the eCPM numbers up.
In fact, it went down a bit. It’s because of the unexpected increase in traffic on Desktop — resulting in more impressions and lesser eCPM. But the question remains — why the eCPM didn’t increase even when the viewability did?
There are three interrelated reasons here:
1. Viewability Bucket:
Desktop viewability was at ~83% before lazy loading and so it’s safe to say that the viewability was pretty good even before the experiment.
After lazy loading, it hit ~89%. While there’s an increase, it isn’t going to change the bid as per our expectations. DSPs can bid based on viewability but they don’t offer granular buckets. 10%+ increments are offered and that probably explains why there’s no increase in the number of bids (and eCPM).
Source: Google DV360.
Because if your viewability goes from 83% to 89%, you’re still under 80% or greater. You aren’t going to get new bids from the buyers. The next bucket starts at 90% or greater.
As the viewability bucket goes higher, the number of advertisers bidding on, will greatly reduce. Because most in the open web run ads for branding purposes and reach is paramount. Having a higher viewability as a condition to bid can cut their reach and so advertisers are very unlikely to go over 90%.
So, even if you increase the bids from 85% to 95%, we can’t expect a drastic increase. In this case, that’s why we didn’t see any uplift.
On the contrary, if you increase it from below 50% to above 50%, it can have a great impact on eCPM (and CTR). For example, we’ve seen eCPM increase by around ~200% when our product AXT increased the viewability from ‘30% or greater’ to ‘60% or greater’ range.
3. Bid Prices
On top of this, note that the desktop bids are typically higher than that of mobile. Now, considering that the advertisers are already bidding higher on desktop and you didn’t increase the viewability enough to jump to a higher viewability bucket, it makes sense to not expect a noticeable increase in eCPM. Even if there’s an increase, it won’t be as high as you see on mobile.
Before going forward with the implementation, take a look at the desktop viewability and bid prices. If your viewability is at a great range — say over 80% and bid prices are in line with your expectations (equal to or above industry average), then it’s better to skip Desktop and focus only on Mobile. For a 1% increase in ad viewability, we are seeing 3.75% jump in eCPM on Mobile.
What about Advertisers?
Wait, so far, we’re pretty much focused on publishers and suggesting them to implement lazy loading strategically but what about advertisers?
From an advertiser’s perspective, it’s great to have an uplift in viewability even if it’s marginal right? No. CTR pretty much remains the same once your viewability is over 80%. So, let’s say you implement lazy loading on Desktop and get viewability from 83% to 88%, then it’s not going to have any impact on the bid prices and CTR as well.
If you think from the user’s perspective, it makes sense. If it’s at 85%, you’ve pretty much seen the ad. Just increasing it by a few more pixels is unlikely to have an impact.
We’ve seen publishers turning it on and off every now and then to ensure they are using lazy loading to improve viewability and also not losing out on a lot of revenue. Rather than implementing it across the board and facing the dilemma, publishers can run lazy loading strategically based on the discussed considerations. From viewability buckets to target geographies to devices, you can use your own data to come up with a set of rules that’ll help you reap the benefits of lazy loading without losing out on revenue.
Madhavan Rathinam, VP of Ad Operations & Customer Success, Automatad
About: Automatad, Inc. is a digital media products company that provides a suite of programmatic monetization solutions that drives efficiency and superior monetization at scale. Using its platform, digital publishers can create, monetize, and optimize for the best ad experience.