The Myths Surrounding Viewable Impressions

I have a natural inclination toward pessimism – it’s in my DNA, please don’t judge me. As I began to hear about viewable impressions, my mind started to crank out hypotheticals on how it would “really” work, as opposed to the utopian descriptions and overviews written in trade publications or presented in discussion panels by thought leaders. I understand the optimism people have preceding a new technological product rollout, but I also know that new isn’t always better.

My lens is that of a web publisher – I have experience as Senior Inventory Analyst and Senior Account Manager, which entails analyzing countless excel spreadsheets to optimize performance based on multiple data points. I look at CPMs, Impressions, and performance metrics all day long, and then I find trends in that data, and find ways to make campaigns or revenue opportunities perform better.

So here it is, my hypothetical and completely antagonistic viewpoint: I believe that, as a publisher, viewable impressions will not make your inventory more valuable. As an advertiser, it will not reduce costs or improve campaign performance. 

If a Tree Falls in a Forest...

For those that are unfamiliar with the term, according to Principle #1 of the (3MS) as outlined through the IAB, ANA, and 4A’s Guiding Principles of Digital Measurement, a viewable impression is when a digital ad impression is counted only when an ad is in focus or “seen” on the web page. The proposed threshold is currently 50% of pixels in view for one second within the viewable area of a visitor’s browser window. If the browser does not see the minimum threshold required, it will not be counted as viewable. 

Advertisers do not want to pay for ads that a visitor never sees. However, as publisher sites work now, all ads are served immediately once the page loads, the impressions are counted, and all the costs are applied. This happens regardless where ads are located on a page, or whether the browser window actually is in view of that ad. Viewability will essentially enable advertisers to eliminate the cost for any unseen ads, which is a legitimate argument and sounds good… in theory. 

However, it can be proven mathematically that it might be in the advertiser’s best interest to keep the current cost structure. This obviously depends on the number of ad placements currently sold on a publisher website, so please bear with me for a few more moments.

Publishers justify the current ad serving model by charging a fractional cost for any ad that is “below the fold” on their website. Along with supply and demand, the risk factor is already baked into the reduced cost of the media, anticipating that the ad may not be seen by a visitor and may not perform well. 

With the model in place for viewability, a publisher will “serve” fewer ads, inventory will diminish and so will revenue. To offset this loss, the costs for all viewable ads will increase in relation to the displaced revenue, driving up CPMs across the site. 

In addition, any ad visible will be considered the same value as a premium placement as it is guaranteed to be displayed within the user’s window. The reduced cost to offset risk no longer applies. This will also reduce the number of impressions than an advertiser can deliver for the same budget.

Furthermore, if a publisher doesn’t offer an advertiser enough scale during the proposal, that publisher may not be included on the media plan when the buyer makes their decision. 

Case Study No. 1: The Publisher

A publisher currently serves 10 ad calls on every page of their website, driving 100 million page views, with 1 billion ads served monthly. 

Four ads are above the fold (assuming 100% viewable), and are considered premium at $3.00 CPM each, earning $12.00 for each page view.

Six ads are below the fold, and are at a reduced cost of $0.75 CPM each, earning $4.50 for each page view.

With viewability, it’s projected that 75% of the below the fold placements will not be seen. This will eliminate revenue for 450 million ad calls at a value of $337,500 dollars per month. (Note – the average measured viewability range is 27% for publisher placements, 19% for ad network placements.)

This $337,500 must be made up by the publisher to retain profitability, so that lost revenue is evenly distributed to 100% of the viewable placements, all considered “premium” as they are viewable and each ad unit now earns a CPM of $4.13.

So publisher eCPMs have increased and the available inventory has decreased. Now let’s take a look at the impact on performance:

Case Study No. 2: The Campaign 

Using the same example above, an advertiser has a campaign objective to acquire new customers. Their CPA goal is $300, and they have been purchasing inventory with a publisher at an average eCPM of $2.00. This serves approximately 5.55 million impressions above the fold ($3 CPM), and 4.44 million below the fold ($0.75 CPM), totaling 10 million impressions.

If the publisher at optimal performance is able to convert 100 users for every 10 million impressions served, this calculates to a CPA of $200, which is well within the client goals.

With viewability, that same publisher in the example above has been forced to increase the CPM to $4.13 to account for lost revenue due to the reduction of inventory.

For that same $20,000 budget at $4.13 CPM, only 4.84 million total impressions can be served, which will only convert 48 users, thus driving the CPA up to a whopping $413 dollars. 

Conversions would have to almost double compared to the original campaign, while also serving around 700,000 less impressions (4.84 million compared to the original 5.55 million 100% viewable, above-the-fold impressions). So, the original above-the-fold priced impressions would still outperform the new campaign by converting 55 users, compared to 48.

This $413 CPA is too far outside the threshold of the client goals, and the advertiser will more than likely cancel the campaign with the publisher. As a result, the publisher is no longer on the proven performers list and will never receive another media dollar from that advertiser again.

Basically, the mathematics in this example proves that viewability is bad for both the publisher and advertiser.

When It Rains, It Pours

Another argument that advertisers have for the application of viewability is the false attribution of view-through conversions. Some web publishers overload their web pages with below the fold advertisements to serve more impressions and to capture conversions because it is the last ad call “seen” before a visitor converts a transaction on an advertiser’s website. 

False view-through attribution muddies the ability to track where an actual conversion came from, and can result in an advertiser spending money with the wrong publisher website. Therefore, if an ad is only served if it is actually seen by a web visitor, it will prevent view-through conversions from being attributed to the wrong publisher site.

Now I’m no genius, but if I were a publisher that purposely overloaded below the fold advertisements to attain more view-through conversions or had no qualms about selling non-performing inventory, I might come up with a few workarounds to ensure all ads are seen on every page.

I imagine it’s possible to create a website where an auto-scroll could be enabled, forcing the visitor to watch as the browser window scrolls from top to bottom and views the entire web page therefore calling all the ads before the user can interact with the site. 

Also, a minimized view of the web site could be loaded, somewhat similar to microfiche, which would show every ad call on the page, and then the page could “zoom in” again to normal size. Either method will counter the purpose of enacting viewable impressions in the first place. Problem solved.

For legitimate web publishers, they may be forced to top-load more ad calls above the fold. Making websites more cluttered, which may also affect user experience, by forcing them to view more ads at the top of the page, and scroll further down for relevant content.

So, now I’ll ask you, as a publisher, do you really believe viewable impressions will make your inventory more valuable? As an advertiser, do you really believe viewable impressions will reduce costs or improve campaign performance?

*Reference: (http://www.iab.net/media/file/VIEWABLE-IMPRESSION-ADVISORYfinal.pdf)

**Average measured viewability range is 27% for publisher placements, 19% for ad network placements.

Related Event: 
OPS Markets


Jeffrey Mayer is the Senior Inventory Analyst at WhitePages.com. He also performs additional duties as Senior Account Manager, the Manager for Display and Mobile Networks, and qualifies new ad tech partnership opportunities for the Ad Sales and Operations Department. Jeff is also not quite sure what job title he should use on his business card.

He was among the first to pass the IAB Digital Media Sales Certification examination, and has more than nine years of digital experience. In his spare time, he enjoys seeing live bands, brainstorming, building websites, and is also the webmaster and an improvisational performer at Gotham City Improv.


Comments

Thanks for taking the time to write this Jeffrey, it's great to spark discussion on this complex but critical issue for the industry.
----------------------------
Knowing exactly what you are buying, can never be a bad thing

I think the most important result of the industry moving to a standard whereby advertisers are only paying for ads that are viewed, is that it will create a more level, transparent playing field that reduces questionable tactics by even large publishers with well known brands - and the point made about increasing advertiser trust.
----------------------------
In-view is mainly about CPM

Also important to note is that In-view guarantee is also more critical for CPM brand buys than CPC or CPA campaigns that generally will have lower eCPMs and automated optimization advantages - and of course have to be as a result of a view (if we don't include ad fraud in the equation). So I think the calculations you make should at least be based on CPM buys only.
----------------------------
We need to go for brand $

A viewable, non clicked, non acquisition impression still has value to advertisers for branding and driving the purchase funnel. TV, print, radio has had this model for over 50 years and if the online industry can help advertisers gain more trust that their ads were actually viewed, we can help retain more of the revenue that these declining media types enjoyed.
----------------------------
Publishers all have different pricing/process models

Your statement that publishers you have worked for reduce the price of below the fold placements, isn't necessarily the case for all publishers. Eg some publishers may sell a campaign as a sponsorship of 10m impressions run of channel, but delivered at lower priority, and most will deliver below the fold completely un-viewed.
----------------------------
Publishers can't commit fraud or they will go out of business

Publishers who try to trick the system and advertisers will be black-listed, as this is just blatant fraud. The argument to not take steps to improve transparency because some publishers will work the system, is like saying not to have a police organization because people are going to commit crime.
----------------------------
Publishers are not markets in themselves

If publishers who had poor in-view rates start to lose revenue on their non-viewed ads, they can't just simply increase the rates of their in-view ads, thats not how markets work. In reality - the ads they are selling that currently aren't viewed, are worthless . They are currently simply, selling something that has no value. In any other industry that would be considered borderline fraud. Previously publishers had the excuse that the technology wasn't available to avoid this, or they weren't aware, or everyone else is doing it too - but they can't say that anymore.
----------------------------
If we want in-view to be taken seriously, it needs to mean more

Lastly, the bar currently set for determining a valid in-view ad is so low (50% of ad viewed for 1 second) - if the industry wants this in-view initiative to be taken seriously, this bar needs to be raised considerably. I'm not sure I would want to pay that much more for 1m impressions of ads that weren't seen at all vs 1m impressions of ads that were 50% seen for 1 second. But that's a hurdle for another day! One step at a time.

I agree with James' point that ultimately advertisers and agencies will make using this kind of tech a condition of their investment.

I would, however, challenge the maths within this example.

I cannot agree with the $4.13 CPM figure

Before any viewability/visibility tech the publisher is earning $1.65m from the page:
400m ad calls from ABF placements @ $3 = $1.2m
600m ad calls from BTF placements @ $0.75 = $0.45m

As Jeffrey says viewability/visibility reduces the ability for the publisher to monetise the 450m ad calls. This leaves 550m ad calls that are visible/viewable and that can be sold.

If the page was previously generating $1.65m then each ad impression needs to be sold @ $3 to generate this revenue:
550m * $3 CPM = $1.65m

This is because, as the author says, the price has already been baked into the difference between the ABF and BTF pricing.

So for the advertiser example the impressions bought changes from 10m to 6.66m due to the difference in CPM.

But when the advertiser was serving 10m impressions they may have been recording 100 conversions, only 66 of these were valid and from visible/viewable impressions.

When the advertiser is delivering 6.66m impressions these are all visible and the 66 conversions they record are all valid.

Nicely written Jeff, but I'm with Stephan i.e. this industry development is for the better. But I would say that - given my buy-side history in the industry (10 years with digital media agencies), not to mention nearly two years repping AdXpose (now part of comScore) in Europe which afforded me an early education on the impact of "viewability" in online display. That said, despite my admittedly one-sided view, this dev is doubtless good for both sides of the equation.

Regarding your questions:
- Do you (publisher) really believe viewable impressions will make your inventory more valuable?

Mine is a seasoned buyer's perspective, but I think of the following in response to your question:
a.) Although many publishers have indeed believed that all display ad impressions purchased were "viewable" i.e. between the upper and lower folds to some degree, many will indeed pay more for guaranteed OTS (opportunity to see); presumably as measured by a third party like comScore, etc. Why? It’s partly down to short memories and perception continuing to be as robust as reality, but isn’t it just a matter of pricing borne out of the laws of supply and demand?
b.) This industry is evolving and growing at a phenomenal clip. I recommend publishers take a long-term view. In the short term i.e. now, fast innovation - faster than your competition - is the way to get through the early pain associated with standardisation of "viewability" in online display advertising.

- Do you (advertiser) really believe viewable impressions will reduce costs or improve campaign performance?

From publisher to publisher, damn right viewable impressions will reduce costs. I regularly saw some premium publishers with InView rates between 5% and 50% (during my Invizua/AdXpose days). To be clear, those pubs were serving between 50% and 95% of imps out of view. Now, to be fair, in most cases those pubs were releasing "remnant" inventory to SSPs and ad nets which were fetching eye-wateringly low CPMs. Either way, at volume, when we remove out of view imps, we remove ad spend. That cost may make its way into viewable imps, but - certainly for DR advertisers and their agencies - only if they perform.

Regarding performance, yes, I'd like to think that given marketers' growing sophistication around the use of 1st party, *2nd party and 3rd party data to drive increasingly precise ad targeting, not to mention measurement practices borne out of the application of in-ad JavaScript tags (capturing enormous impression-level consumption and interaction detail), 1st party ad serving, device recognition/fingerprinting services (yes, this, too; an inevitability) and clever uses of systems like **BrightTag (http://bit.ly/H5arud) will absolutely lead to improved campaign performance.

Lee importantly calls out the importance of varying counting methodologies (core drivers of "performance") - and many a marketer will most definitely want to treat ad viewability differently. Some will want 100% InView rates + 1, 2, 3 or more seconds of InView time, while others will be happy with 50% InView at 1 second. Some advertisers may require an InView %, plus guaranteed engagement time (also measureable).

Thinking more broadly, the various LUMAscapes (http://bit.ly/AmQCh8) should act as inspiration for publishers and advertisers alike – the opportunities presented by each of the businesses behind each logo are enormous (and many logos are missing). Not only will the industry get through the “ad viewability” conundrum, but we’ll get through it stronger and better than ever before.

*My definition of 2nd party data is drawn from Xaxis's Mark Grether who spoke at www.d3con.org last week. 2nd party data being that information gathered across clients, but not client specific, for application across Xaxis client (Group M agency) campaigns. Here's a related article: http://bit.ly/YQioHT

**Full disclosure: I run BrightTag's EMEA business.

My three pence from London.

James

Jeffrey, this is a useful analysis, but I think you miss the point slightly.

The whole point of 3MS is to attract more advertising dollars to the online category by improving the confidence marketers have in what they are buying. When confronted with the option of being able to buy viewable v non-viewable impressions, 100% of marketers will say they only want to buy viewable ads. And are they willing to pay more for it? No, because they have been assuming for years that their ads were, in fact, viewable.

The implications of this for individual publishers varies depending on your current % of viewable impressions, your ability to influence that % positively through smart site design and content layout (not gimmicks like the ones you give as examples), and how comprehensively the marketers who buy your inventory have migrated to viewable as a standard. In some cases viewability could increase CPMs (if supply is restricted), but mostly, I suspect, it will only increase sell-through rates.

I truly believe that a whole-scale migration to viewable as a standard will be good for industry as a whole as it will improve advertiser confidence and will force more of the ad revenue to premium content publishers who create viable opportunities for brands to connect with audiences on their sites.

The cat is out of the bag! Publishers need to figure out how they are going to respond instead of trying to argue it away.

Paid Member

Hi Stephan,

I completely understand that buying confidence is one of the reasons why viewability is popular among media buyers, but this article is an example of why it may not be a viable solution for either the advertiser or the publisher based on cost.

My article is not based on a "gimmick" publisher site. In fact, it's based on a premium publisher, and I believe I was very conservative in the number of non-viewable ads vs. viewable ads, and the CPMs used in the example. If you take a look at some top publisher websites, please let me know how many ads you see loaded in view immediately vs. the number of below the fold placements (or out of view ads) at risk. For example, I visited the home page of a very popular news site today where only one ad loaded in view, and seven were below the fold. Based on my math above, the numbers would be much worse for the advertiser in both eCPM and performance.

If you think this is all the fault of poor site design or greedy publishers, you're off-base. There is limited space within the viewable window of a browser to display ads, and as a publisher we have to consider user experience for content and ad placement. Maybe in the future we'll only allow 25x25 sized ads on our pages instead of the IAB rising stars like 300x1200 Portraits and 970x250 Billboards. Or, another alternative would be to build a top heavy site where its overloaded with ads until you scroll down for relevant content. As I recall that type of design died out back in the early 2000's along with the dancing baby gif. At least it did with the websites I frequent.

If you believe a publisher will accept the loss of ad revenue without applying it elsewhere, you are also mistaken. Publishers depend on the ad dollars to remain in business. There is no chance that they would cut out such a large percentage of revenue without making it up where they can. As an advertiser, you will be paying more per ad placement, and you won't have a choice.

There was never any assumption for years that you were buying viewable ads, everyone knew that below the fold placements were not always seen, that is precisely why they are so inexpensive.

If you believe this will increase sell-through, you are incorrect. For one, most publishers use networks to fill 100% of their inventory, so the only change would be the ratio between the ad delivery of direct sold vs. network/RTB ads delivered. You are not increasing sales on the site and no additional revenue is earned.

As a publisher, my response to you and to viewability is that needs to be vetted, tested, and proven before it can be implemented.

In the end, I may be wrong and it may work, and it may be good for the industry, and when that time comes, I'll write a new article saying so, but I'm not jumping on any bandwagon yet.

Paid Member

Great article Jeffrey. You didn't get into the issues on around how viewable impressions will be counted or whose technology will be used. Also, what about all those impressions that for any number of reasons, can't be counted as in or out of view?

This really turns into a guessing game on the seller said and, like you stated, means that rates will go up or sites will start getting clever about how they get their ads counted as in view.

Thanks.

I think your arguments are good, but I don't believe the math is quite that bad for the advertiser.

The 100 conversions for 100M impressions includes 4.44 million below the fold impressions that only have a 25% view rate.

So the conversion rate would seem to be 100 for every 6.66 million VIEWED ads.

6.66 million ads at the new rate would be $27,505.80 - $275 CPA, more than it used to be, but inside the guidelines.

But building that viewable ad infrastructure is going to cost a fair amount of money in new ad systems and integration that the publisher is going to have to recoup in higher CPM.

Paid Member

Hi Jonathan,

Thanks for the comment & I recalculated the math on my end as well. I sent in a revision for the article based on the new numbers which are more accurate. I had to change the case study goal from $300 to $250 CPA to allow the article to sustain its validity since it was just an example to address the negative impact of performance.

With the adjusted 100 conversions for 6.6 million viewable impressions, that backs out to 72 conversions for 4.8 million impressions in the new campaign. ($277 CPA).

Jeff

Rocket Fuel