Commentary
Unique Visitors Come In Two Shapes And Sizes
- by Jodi McDermott , March 20, 2009
Unique visitors come in two shapes and sizes -- the Web Analytics Association (WAA) version and the Interactive Advertising Bureau (IAB) version. For more than 10 years, the term unique visitors has been known as the count of unique cookies (de-duplicated over the period of analysis) to a Web site. As most sites do not require authentication or login, this unit of measurement (although not foolproof by any means) has been the de facto standard that most Web analytics vendors have used when configuring the unique visitors metric within their software.
In December 2006, the WAA published a standard for unique visitors that aimed to establish the definition of this metric around the use of counting unique cookies, with the option to use authenticated users if available. The goal of the standard was to educate the Web analyst to the most commonly used definition and to encourage vendors to openly document any variances from the standard, given that data collection and processing techniques may vary from vendor to vendor.
On February 23, 2009 the IAB published a set of Audience Reach Measurement Guidelines that specifically address the counting of unique users (also referred to as unique visitors in the document) as reported in panel-based measurement tools such as comScore and Nielsen, but they go on to say that census-based measurement tools should comply as well. The guidelines require that the measurement organization "must utilize in its identification and attribution processes underlying data that is, at least in a reasonable proportion, attributed directly to a person." Now keep in mind that the IAB's focus is to provide standards that are specifically applicable to the buying and selling of advertising, but its definition leaves one to wonder how census measurement tools (aka Web analytics vendors) will ever comply with this standard. Will they start to collect personally identifiable information? Will their upcoming product pipelines include the development of "algorithms and other data adjustment procedures" to meet the guideline? Could this guideline ever even be applicable to them -- or will it only ever apply to the audience measurement firms?
Bravo to the IAB for forcing the issue with audience measurement companies to standardize the way that they report uniques, but from a Web analyst's perspective -- and as a member of the WAA Standards committee -- I wish they would have not allowed the term "unique visitors" to be redefined in such a way as to allow for multiple definitions in the space. Web analysts and media planners today have a hard enough time trying to figure out which data source to use and which standard to apply when performing their job -- but that issue is now compounded even more by multiple definitions of unique visitors. In defense of the IAB, its membership is comprised of some heavy-hitter companies who are not about to change that "tab" in their reporting UI that says "Unique Visitors" on it. But in defense of WAA individual and company members, which include vendors such as Omniture and WebTrends (who were both listed as "Project Participants" on the IAB document, interestingly enough), neither are we. The term will live on in both places.
So where do we go from here? It is critical for those of you who use both Web analytics data AND use audience measurement data to know when to apply each standard.
Advertising: When buying or selling advertising and reporting the unique visitors number, you must use the IAB standard as your definition. This is
most applicable when referring to your audience measurement data that is derived from panels that consist of identifiable people.
Web Site Analysis: When analyzing your
Web site for any other purpose, the WAA standard is sufficient and is most commonly what you will find as the technical calculation in your Web analytics tool.
For those of you wondering, "Aren't you guys working together on standards?" Yes, we are talking, but we aren't approving each other's standards, only reviewing and raising the issues to each other and to our memberships when we agree or "agree to disagree." Our organizations differ greatly in composition and purpose and we aim to increase education and communication in the space as best we can, one metric at a time.
I understand the concern, but the long-term answer is clear: we must switch to counting people, not cookies. Cookies don't view ads or buy things, people do. And reporting cookies as unique viewer count is dishonest exaggeration, plain and simple. Why? Because most buyers don't understand the difference and most publishers conveniently don't explain it.
It's unfortunate that the IAB didn't work with the WAA on these specifications, but they didn't cause the problem, they were trying to address the problem of too many vendors using the term unique visitors using entirely different methodologies, with the end result being that end users would get wildly different numbers for what appeared to them to be the same metric. I think some of the IAB definitions are somewhat arbitrary, but they are moving in the right direction to focus on transparency of methods and clear terms to identify different ways of measuring uniques, and in the advertising space this can relate directly to the cost or effectiveness of advertising campaigns.
John - I agree that the IAB didn't cause the problem and in fact or actively working to solve an issue that has caused so much discrepancy between the audience measurement companies. To your point (and Steve's), there definitely is a need for a standardized way to count people. The main concern here is that not all systems have a way to count people. Given that Web Analytics vendors do not commonly collect personally identifiable information, they have defaulted to counting cookies. Both metrics should be available to analysts. They just should not be called the exact same thing.
Jodi
I think I commented quite a while back that no one would dispute the fact that counting bodies instead of computer is the right way to go for media plan/buy. However, I said out loud too that unless the post campaign metrics is audience based (in other words, unless the success of the campaign is measured via bodies), it does not make too much sense to base off the entire plan on audience measurement only. One is only as good as what one is being measured against. And "unfortunately" most post campaign diagnostic is based upon adserver reports, not ComScore or Nielsen (I do not want to get into the issue why ComScore/Nielsen cannot adequately address the post campaign metrics issue here).
There are technologies out there that can add another dimension to tracking which completely bypass issues associated with cookies. While they are certainly not applicable to all internet traffic, they do provide good subsets of data that can used for predictive analysis. They can also provide new targeting capabilities ensuring delivery of media to specific individuals that meet predetermined criteria. Best part is, no user names or passwords are required.
The technology can be applied to access points open to the public for internet consumers on the go. With 10+M WiFi enabled device hitting the market every year, I expect that it could online ad community with something to consider.
Until people can agree on a standard, there can be no standard. Agreeing to disagree doesn't help.
This is a pretty easily solvable problem on the surface. Like Jodi says, we just need to use different terms for the different things we measure. The IAB spec relabels uniques based on how they were measured, introducing terms like "Unique Users" vs. "Unique Cookies" (as well as a couple of other terms). I think they took the right approach by trying to differentiate the measures, but I take issue with the approach because it doesn't necessarily help marketers understand the distinction other than through implementation. I wrote about it right after the IAB Forum on Audience Measurement last December when the spec was released (http://blog.adometry.com/?p=11).
A very interesting post Jodi. And I must say that Steve Crozier has hit the nail on the head - most buyers don't understand the difference and not only do most publishers conveniently not explain it, they actively leverage it.
Due to the rapid growth of the online world it is littered with terminology that is at cross purposes. Unique Visitors is just one of them. What web analytics (cookie-based) terms "Unique Visitors" is actually "Unique Visitations". The "Visitor" is the person, the "Visitation" is the action. Cookie-based measures are visitations.
The same goes for the term "Unique Browsers". In a cookie-based world this is really "Unique Browses" - the action, and not "Browsers" - the person.
Another example is the use of the word "Unique Searches" - the action. As a planner I am after "Unique Searchers" - the person.
Cookie-based metrics are good for establishing the quantum, but fall short on measuring the people behind that quantum. First, they have to be washed to remove, bots, spiders, crawlers, internal traffic etc. They generally also need a geographic filter to focus on domestic users rather than global users.
Cookie-based measures really struggle with the 'unique' part of the equation due to the now well-documented issues with cookie-deletion. Frankly, for any time period of longer than a day cookie-based measures should never use the term 'unique'.
This is not to say that panel-based measures are the way to go. They manage to understate the quantum of traffic largely because they struggle to measure at-work and public-place access. However, in terms of audience measurement a panel generally understates the unique audience by less than cooki-based measures overstate the unique audience.
The IAB is correct in ascertaining that "unique visitors" needs to be people based. Just because the term has been used by 'web analytics' for a decade is not a good enough justification to continue using a misnomer. Let's get it right NOW - before any more water flows under the digital bridge.
And while we are on the subject of misnomers, 'web analytics' is a misnomer in itself. A publisher may adopt a web analytics vendor such as those that Jodi names - but they are in effect only providing analysis on the publishers sites ... not the Web. They have no 'knowledge' outside of that slim vertical world of the publisher. Isn't it about time that we renamed them "site analytics"?
John Grono
GAP Research
Sydney Australia
Jodi,
I'm going to have to agree with Steve, John, Chen, and John here ... your assertion that the WAA's definition should be preferred over the IAB's or that somehow there is room for two definitions is wrong. More importantly I think this kind of thinking does our entire industry a disservice.
I expanded on these thoughts and commented on how I think the IAB's definition of "Unique Users" should be adopted in my blog:
http://blog.webanalyticsdemystified.com/weblog/2009/03/unique-visitors-only-come-in-one-size.html
Eric T. Peterson
Web Analytics Demystified, Inc.
http://www.webanalyticsdemystified.com
Unique Visitors, how you count them might become obsolete as a KPI for a site as Online Video takes over the web. What will really count is the number of Video Views, rather than Unique Visits. You Tube may well have 100 Million Uniques, but with hundreds of millions of Video's, what I want to know as an advertiser is how many times my (viral) video has been seen. Combine that with the fact that this is a Lean Forward media (versus Lean back for display advertising), the importance of number of Video Views will increase dramatically. As the web evolves towards Video and Search will become Video Search (the reason Google acquired You Tube I believe), Unique Visits will become a thing of the past!
Jodi:
As I tweeted to you, I wanted to comment here but was on my iPhone only and these big fat thumbs do not lend themselves to facile commenting on that device. The sentence I just typed would have taken me 15 minutes on the iPhone.
The IAB and the WAA have two very different focuses. Specifically the IAB is mandated to focus on advertising. I think it is fair to say that their standards should be viewed in the context of planning, buying and selling advertisng only.
The WAA is charterd to focus on "understanding and optimizing Web usage." I don't know if there is an ad sales component in there at all.
These two diffeent mandates suggest diferent applications of data.
When the IAB standards talk about how to define reach, they very specifically refer to an ad sales context. Thus, I do not believe the "measurement organizations" in question are in fact Web Analytics providers; because WA data is designed for a company to use internally, not externally in the sales process. I believe it is the publisher who chooses to take it's "internal data" into the sales process that these standards are targeted to. Not Omniture, for example; but the Omniture publisher who wishes to use Omniture repoted data in the sales process.
An additional relevant window comes from the fact that George Ivie of MRC was the primary author of this document; the MRC currently audits syndicated audience measurement companies like Nielsen Online and comScore, and also publishers like Forbes, Disney and CNET, who presumably use site-centric data in their sales presentations. But they do not audit, say, Omniture or Webtrends.
Strong opinions from all angles here. The WAA and IAB are both trying to support different constituencies in their approach to standards and guidelines. The issue that I am trying to shed light on here is when you have sites who are analyzing their own clickstreams, selling and buying advertising - that is where things get gray. I am not proposing one definition over the other (and I am not saying that the WAA is correct either), but what I am saying is that you have a lot of people (analysts and their execs) in the industry that struggle with the terminology to begin with. In discussing this topic with the IAB (because we did have a good dialogue with them about it), they agreed that we have some difficult struggles here to overcome.
@joshchasin, how do you think that the publishers who want to use web analytics data in the metrics that they report externally will adjust their numbers to meet the guideline? My guess is that it won't be possible for quite some time on a standardized, mass scale.
Jodi