In this article, we will cover how to place ad units on pages, different techniques, and a few other issues to consider that you can use to optimise your advertising revenue generation. We will touch upon Google’s Core Web Vitals briefly when it is relevant to the actual topic, but we will not cover them extensively, although they contain important metrics to consider while developing the overall website. The focus here will be specifically on the ad units rather than search engine optimisation and other similarly important topics.
Basic Ad Unit Placements on Page and Ad Coverage
Ad unit location on a page is naturally one of the most important topics here. Fortunately, gone are the days - mostly! - where ad units were inserted at the bottom of long news website pages without any lazy loading. These so-called “trash ad units” that nobody saw and unsuspecting and non-optimising (programmatic) buyers ended up buying.
Another similar positive change within the last five or a few more years has been that the leaderboard (the big top banner ad unit) has generally moved down from the very top of the page to at least under the navigation. This gives it better viewability as users will not immediately scroll past it on their screens.
These are a few examples of some basic issues to consider when inserting ad units on a page. Below we will go through more technical, many already known, and much-used, ways to optimise your ad units.
Also a word of caution for reducing your on-page optimisation to simply putting one more new ad unit on the page. Many SSPs impose restrictions on how much ad coverage they consider acceptable on a page. The usual upper limit to follow in this regard is around 30% ads and 70% actual content.
Nobody benefits from a situation where a user only sees ads on their screen - at least in the long term. Too many ads decrease the user experience leading to less service usage and a decrease in impressions per ad unit. They also increase the page loading times, which negatively affects the website’s Largest Contentful Paint (LCP) Web Vitals score and thereby its search engine rankings. Too many ad units also make the page less interesting to buyers because the person viewing the page will less likely to pay attention to the buyer’s ad.
So now that we got that out of the way let us see how we can optimise the ad units that we can have to get the most revenue out of them.
Viewability and Beyond
When considering where and how to place your ad units you should make sure that at least around 70% of the ad impressions in them will fulfill the definition of a viewable ad. In short, this means that at least 50% of the ad should be shown on the user’s screen for at least one second for display ads and two seconds for video ads. Although this may sound like a fairly low bar it is something that most ads in ad units in the wild could not achieve even a few years ago. This has fortunately changed now but other requirements have emerged or are emerging.
Although viewability is an important metric it is not enough anymore. The significance of other metrics is increasing in importance with increasing amounts of algorithmic bidding. Having a viewable ad unit is not enough if nobody happens to look at it when buyers bid towards clicks or conversions. Some agencies, such as Dentsu, have even come up with new terms for this. Dentsu talks about aCPM meaning attention cost per mille.
Some traditional display formats, such as skyscrapers at the side of the screen, have really suffered in these measurements. Even if they might be technically viewable on screen nobody just actually looks at them. This in turn increases the so-called aCPM price for those units since many of the impressions will be counted as not having gotten user attention.
Looking at clicks alone cannot address this either. Many especially in the buy-side of the industry are well aware of the miss-clicks with some ad units where you accidentally press an ad when trying to scroll down or some other similar situation. Even without algorithms helping them in the buying process, it is fairly easy to calculate averages of those types of clicks when comparing the buying platform data to website analytics data. An easy way for a publisher to analyse the same phenomenon is to run its own house advertising and similarly compare the numbers.
So with the increasing sophistication of algorithmic buying an ad unit prone to this will get caught. Especially so when buyers use demand-side platforms such as Google’s Display & Video 360, which collects platform-wide (if and when given permission by the agency account owners) bidding data to feed its buying algorithms. Although in this case, the ad unit in question would not face punishments it also does not get the benefits, if buyers optimise towards conversions rather than clicks.
Heatmaps can be one extra tool to give you insights to determine where an ad unit might get a user’s attention and thus make the ad unit more interesting to buyers and buying algorithms. Basic heatmapping services are not foolproof but they offer more understanding when used in conjunction with other ad unit metrics. When doing their research Dentsu has apparently gone even one step further with this to proper eye-tracking at least in limited fashion.
Standard Popular Ad Unit Sizes and Accepted Ad Sizes
This should be fairly obvious and we have previously covered much of it in other blog post but it is always worth repeating. If you use ad units that buyers are not interested in or deviate from standard sizes you will probably lose money in programmatic sales. For deals and direct sales, it might mean that the advertiser would have to create custom creatives only for your ad units, which costs time and money.
There are exceptions even to this rule of course. Some non-standard high-impact ad units may work great in direct sales and bring in high CPM sales. But when considering these you should estimate whether you can bring enough demand to them as they will most likely take quality space from the standard ad units. A middle road here can be if the high-impact ad unit is capable of also accepting standard sizes when not being sold out through direct campaigns.
In open RTB advertisers might not even look at which websites have which ad unit sizes and rather target their ads based on other criteria, such as data or contextual targeting, and broad include/exclude lists (formerly called black/white lists). So they might simply ignore your ad units in that setting without even being fully aware of it because your ad units deviate from the standard ad sizes that they use.
Check from regional IABs and exchanges, such as Google, which ad sizes are most popular and use those. In addition, we recommend accepting smaller ads than the maximum size of the ad unit into your ad units in most cases. This will have to be configured to work. Though, some caution should be taken when doing this. One decision that you will have to make is whether the size of the ad unit will adapt to the size of the ad and there is no one correct ainswer to this for all cases.
We recently wrote that you might want to consider keeping the ad unit size the same even when selling ads that are smaller than the full size of the ad unit. This would prevent the ‘jumpiness’ of page content, which annoys users and can affect negatively even to your search engine optimisation by increasing the Cumulative Layout Shift from Web Vitals. On the other hand, sometimes a smaller otherwise acceptable ad in a larger ad unit might look a bit out of place on a page when the ad unit will not adapt to the size of the ad.
Implementing lazy load
Viewabilities have been greatly improved with lazy loading ad units This means a technique where ad units only load when they come to the user's view on-screen. This could even make the ad units at the bottom of the page, which we already called “trash ad units” a viable option since they would only load up if a user actually scrolls all the way to the bottom of the page. The inventory for these (amount of times the ad units would load up and thus become possible to sell an impression to), however, would probably be quite limited, especially in longer pages.
Some publishers are also running tests to see whether they can improve viewability even more by triggering the ad unit loading with lazy loading slightly before the ad unit comes into view. This gives the programmatic auction process that tiny bit more time to complete, which in turn could give the ads better viewability as will be ready or almost ready when the ad unit comes into the scrolling user’s view.
It is worth noting that if you plan on using this technique make sure that despite the name ‘lazy loading’ the ad units load as quickly as possible when users scroll to them. We have seen some unfortunate cases where lazy loading is actually lazy, or slow to be exact, which completely defeats the point of using this technique. In these unfortunate cases, the ad unit has loaded so slowly after being triggered by lazy loading that users have already scrolled past them by the time they manage to complete an auction process and display an ad.
Sticky Ad Units
So-called sticky ad units are another common technique used nowadays to improve ad unit viewability. This means fixing the ad unit in a certain part of the user view even while the user scrolls up or down the web page either for the full length of the page or between certain points of the page.
Skyscraper ad unit implementations at the side of the screen are common ways of using this technique. Another one is having a billboard ad unit at the top of the screen travel a small distance with the user as they scroll down.
It is important to tell the developers with these partially sticky ad units to actually close them when the ad unit disappears from the user's view. Only hiding the ad unit behind other page content without actually technically closing it is definitely not enough in these cases. In fact, by only hiding the ad unit you will actively risk the whole ad setup because SSPs will still consider the hidden ad unit as viewable, which means that they will receive and transfer deceptive information from your page. SSPs may even block a website if they get caught doing something like this because it is not in their interests either to supply false information to buyers.
Ad Refresh
Ad refresh is a technique where the ad unit will be resold after a certain period of time. This works for ad units, which have a long viewable time on screen. Most often you see this in services such as in-browser games, other small applications, or dashboards. The obvious benefit from ad refresh for publishers is getting more impressions out of one-page load since more impressions will be shown to users who stay on the page for longer durations.
Some care should be taken when using this technique. The already previously mentioned use of fixed-sized ad units when allowing the sale of smaller ads could be especially important here. You can probably imagine how annoying it could be to the users if the page layout would change with every refreshed ad.
You should also make sure that refreshing the ad unit will not hurt the performance of the impressions sold to it. This could happen especially if the ad refresh is done too aggressively.
Ad refresh should not be triggered when no use of the page is detected and doing so is either against the policies of many exchanges or at least frowned upon in ones where it is not explicitly disallowed, since it indicates that the user is not actually looking at the page and thus you would be selling ads that nobody will end up seeing. So while the ads would be technically “viewable” since they are in the user viewport for long enough, nobody would actually see them.
The best practice is to wait at least 30 to 60 seconds before doing a refresh and to keep the maximum number of reloads within sensible limits per page load because at some point the user will become “blind” to the new ads.
Limit the number of SSPs
Limiting the number of SSPs to only those that bring in extra revenue is a good practice in general but also worth mentioning as the last topic with on-page ad unit optimisation. Fewer SSPs most likely means faster auction processes and better viewability. Another way to look at it is through ‘green coding’ practices and addressing the CO2 emissions in ad tech. Less auction processes to conduct in SSPs the less processing power is required to deliver an ad into an ad unit.
This does not mean that we would recommend limiting the use of SSPs to only one or two. We would rather recommend analysing and testing which ones bring added revenue and removing the ones that do not contribute extra. The best set of SSPs and partners for your ad units can also vary between different geographic or other segments. This means that, for instance, a website that operates in two countries could conduct an auction with a different set of SSPs for the same ad unit for users coming from those countries if it is found to bring in the most revenue.
Conclusion
This is by far not the complete set of issues and techniques that you can use to optimise your ad units but it is a collection of issues and techniques to consider. We recommend actively monitoring, analysing, and testing your ad setup to see how you can get the most out. To help you do that we will naturally recommend our Relevant Yield set of services that help you unify your reporting, manage your setup, conduct tests in it, and monitor your setup’s performance with sophisticated reports and alarms.
