Should we start thinking about ‘link viewability’?

  • 0
  • October 8, 2014
Stephen Kenwright

Stephen Kenwright

Director of Search

Statistically you’re more likely to survive a plane crash than click on a banner ad.

‘Banner blindness’ is partially to blame – TechCrunch claims that 63% of people who see ads will ignore them. But how many people are actually seeing the banner ads at all?

Viewability is a word that appears in relation to banner advertising with increasing frequency. According to Digiday:

Viewability is an online advertising metric that aims to track only impressions that can actually be seen by users.

Advertising that can be seen by users should be standard. For banner ads, though, this is becoming a source of controversy. CPM is the most widely used metric for gauging banner ad spend, but to understand Cost Per Mille – or the cost per thousand impressions – marketers need to understand what we’re getting when we pay for an impression…and to do that, we need to understand what an impression means.

According to the IAB:

A measurement of responses from a Web server to a page request from the user browser.

The idea that registering 1 impression means 1 person has seen your ad simply isn’t true. It means one page request has been made, and a person may or may not be on the end of that.

The impression that I get

According to MediaPost’s Reid Tatoris, banners fail about 15% of the time. The wrong ad is loaded, or the right ad fails to load, but as Tatoris says “most ad networks will report, and try to manage, these numbers”.

Bots account for 61.5% of traffic on the internet, though IAB standards say that this should be filtered out when you’re measuring impressions.

Worse still, around 25% of the entire ad market is fraudulent. There’s no human or computer error here – just big money available for people who want to steal your ad spend on purpose.

Seth Godin talks about this in his book Purple Cow

[Prior to the dot-com bubble] banner ads were selling for a CPM of $100. (CPM is the cost per thousand ad impressions.) That means you’d pay $100 for a thousand banners.

What advertisers who measured (the vast minority) soon realized was that every time they bought a thousand banners, they got exactly zero clicks. The banners had a hit rate of less than .000001 percent.

Today, you can buy banner ads for less than a dollar a thousand. A 99 percent drop.  I did a deal with one site in which I bought 300 million banner ads for a total cost of $600. The funny thing is that I lost money on that deal.

300 million people saw Seth’s ad and he still lost money. Must just be a badly designed banner, right?

Actually, Seth paid for 300 million impressions, and if Tartoris is to be believed, this could mean that Seth’s ad was actually only seen 24 million times.

It could be that when you’re paying for 1,000 impressions, only 80 people are seeing your ad.

It’s worth mentioning that in the above example, Seth Godin was buying banner advertising in 2002 – there was significantly less bot traffic online.

But you’re not buying banner advertising in 2002. You’re buying it now.

Paying for banners that aren’t seen is a waste of money regardless of whether you’re Seth Godin or not.

Why should your link acquisition efforts be any different?

So should we track impressions when reporting on links?

Let’s get the caveats out of the way first:

  1. ‘Link viewability’ is not an IAB sanctioned metric – nor should it be. But how many people are likely to see the link you’re acquiring is something you should bear in mind
  2. Placing links where we’re able to track impressions is manipulative – you shouldn’t be placing links in locations you have control over and you certainly shouldn’t be buying them
  3. An impression is not a sale

…but this does blur the lines.

Recommendation engines such as Taboola and Outbrain – and now Facebook’s Atlas – create links that are for users and not for Google. We can track impressions from these tools (because we still use that to measure success), but it’s fine – these links are for users and not for Google.

This implies that there’s such a thing as links that are for Google and not for users – this absolutely should not be the case in 2014.

It’s likely that in future Google will place less trust in links on sites that do not receive traffic (for anyone interested in exploring this further Marcus Tandler is the man). If a link is treated as an example of one site vouching for another’s credibility – but the linking site has no credibility itself – that makes complete sense. If TechCrunch is ‘vouching’ for an author’s credibility on a subject with a link, but so is generic-tech-blog.info, Google is likely to trust one more than the other (no prizes for guessing which). If nobody ever sees the link on some guy’s tech blog there is no reason for Google to transfer any authority from that site.

We report with this in mind when we’re running online PR campaigns for our clients, providing

  • A seeding list with outreach targets – social metrics can help justify why we’re talking to specific bloggers, because ‘authority’ is more than just PageRank
  • Campaign reports containing referral traffic to the assets we’re using
  • Conversions we can directly attribute to the campaigns we’re running

Suggesting that you should build links as if Google didn’t exist is naïve. But building links that nobody will see is just a waste of time.

Free of charge. Unsubscribe anytime.