A Unique Visitor is NOT a Person

A few weeks ago, Brian Clifton wrote a great blog post about how measuring unique visitors is meanlingless.

I wholeheartedly agree. Especially in higher education, the recipient of an analytics report rarely knows what the term *unique *really means in this context or the problems with measuring unique visitors.

“Unique visitors” is misleading. Why? According to Brian’s post:

Firstly, cookies get lost, blocked and deleted. Research has shown that after a period of four weeks, nearly one third of tracking cookies are missing, which means the visitor will be incorrectly considered a new unique visitor should they return to the same website.

The longer the time period, the greater the chance of this happening, which makes comparing year-on-year data invalid for example. In addition, browsers make it very easy these days for cookies to be removed – see the new ‘incognito’ features of the latest Firefox, Chrome and Internet Explorer browsers.

However, the biggest issue for counting uniques faced by both on and off-site web analytics tools is how many devices people use to access the web.

The problem I see with unique visitors is the name. Most (and by most I mean the recipients of your report!) *assume* unique visitor means a person. Unique implies … well … that it’s just that … unique – one – a person.

Even the Web Analytics Association definition of unique visitors says it’s an “inferred” person.

The number of inferred individual people (filtered for spiders and robots), within a designated reporting timeframe, with activity consisting of one or more visits to a site. Each individual is counted only once in the unique visitor measure for the reporting period.

Using the term “inferred” makes the statement, I guess, more true, but it’s misleading nonetheless. The document does go on to note the problems with measuring unique visitors, so that’s good.

The only way to true measure of “unique visitors” is by using authentication, but the majority of our sites and pages don’t do this.

In the past, when I’ve been asked for unique visitor rates because we want to measure how many *people* have come to the website, I’ve pushed back.

The argument for using unique visitors has been that we want to create conversion rates for enrollments, which are “people” so using “unique visitors” would be apples to apples.

That’s not true, though. That *assumes* it’s apples to apples, but it isn’t.

Use visits instead. It’s a more accurate number and more importantly doesn’t “infer” that it’s something it’s not.

Last week while taking a training session at Omniture Summit, I was shocked that our instructor said that unique visitors meant people. He preceded the statement with something like, “yeah, yeah, there are a ton of things that can skew unique visitors, but …” and then proceeded to say that they are essentially people.

I almost flipped my lid.

I know. I take this stuff too seriously.

This is big, though. Remember who the recipients of your reports are. Management? Leadership? Will they assume unique visitors is people? Probably. That’s not only misleading, it’s irresponsible on our part.

If you absolutely must use unique visitors, make a big, red, flashing label on your report that spells out why that term does not mean people.

Better yet, just use visits instead.

8 Responses to “A Unique Visitor is NOT a Person”

  1. Christina says:

    ah geesh, this analytics thing is really difficult to unravel. I really appreciate your very direct and simple posts like this one. I can digest one itty bit of info at a time. And for the record, damn! What’s the point of even showing/listing/tracking unique visitors then? Thanks for teaching me something new =)

  2. S.Hamel says:

    Hi Shelby,
    I agree, taken literally, unique visitors is not a unique account of unique people. Two sides of the coin:
    1) Web analytics is in the realm of statistics. We’re talking about a “population”, “segments”, “margin of error” and “significance”. So, even if x% of people delete their cookies regularly and y% others use multiple computers or browsers (thus, screwing up the “unique” metric) , does it make your unique count irrelevant and inaccurate beyond the threshold error limit you would like to get? Probably not. Would it significantly alter the trends? Certainly not. Web analytics is excellent to asses the behavior of a population segment, not to account for 1-1 relationship with what should be handled by core corporate systems (student enrollment database in your case). So in this respect, I agree “unique visitors” is not a strict 1-1 relation with “people”, but “unique visitor” is a statistically valid representation of “people”.
    2) By using “visit” in your reports, you assume that every visit is an opportunity to convert (or enroll in a class), which is rarely the case. Especially if you don’t filter bounces and people who didn’t come with a “conversion” goal in mind (like just browsing for course info). Furthermore, by focusing on visits, you loose track of the sales cycle that certainly involves searching for a good school, a good class, reading info, maybe subscribing to a newsletter or participating in an online survey/quiz, etc. You don’t look at the larger picture of all the smaller activities that led to your ultimate conversion goal.

    My 2 cents :)
    Stéphane

  3. Ted S says:

    Wonderful post and something that needs to be drilled in more and more until we all change our use of the word. I’ve certainly had many chats with management and coworkers about how we get more “people” converting because a report showed a certain CVR for uniques and honestly there’s been many instances when I’ve fallen into the trap myself as it’s so easy to write off the issue as just being for a “small part” of uniques.

    But even forgetting cookie deletion and blocking, the issue of multiple access points alone erodes the metric. For most sites people are already using multiple access points — home PCs, work machines, mobile devices and that’s only going to go up.

    While it’s not 90% of visits I think it does have a major impact. In web analytics we often measure for goals of 2-4% thus if just 10% of people clear cookies over a month span and 10% use multiple browsers your 3% conversion rate is really 3.75% of distinct people – a significant difference.

    Of course trying to explain the fault in the metric always raises the question “well just how wrong is the number” and “how do we get a better picture”. I’m not sure there’s an answer yet to either question but seeing as how the issue is only going to grow it’s one that does have to be addressed… Visits don’t look at enough time span and uniques don’t account for the true nature of an individual anymore.

  4. @Christina – Yes, analytics can get complicated, but I think Stéphane brings up a good point. If numbers are statistically off, say 10%, then they will be across the board. This is why trending is so important. Just doing away with the metric isn’t the answer, though. Knowing the limitations of the metric is.

    @Stéphane – I think you hit the nail on the head when you said, “Web analytics is excellent to assess the behavior of a population segment, not to account for 1-1 relationship with what should be handled by core corporate systems.”

    Your insight into using visits is excellent. Point definitely taken.

    We are often pressured to come up with a 1-1 relation with web traffic and, unless your site forces authentication, that’s virtually impossible. Therein lies the problem.

    @Ted S – Brian hits home your point about access points on his blog. He has a specific scenario that is very common.

    Thanks all for your comments. I appreciate it.

    Don’t get me wrong. Those that know me know I am a web analytics evangelist. Especially in higher education, it is so needed and so under-utilized. We need to use the tool for what it *can* do, though, not try to force it to be used for things that can be so misleading.

    Wouldn’t it be great if all sites forced people to log in? I kid. I kid. ;)

  5. Nice post Shelby and thanks for the reference.

    I am surprised at the Omniture trainer’ response, as Matt Belkin has blogged along the same lines as my post some years ago. I guess trainers are feature experts/sales evangelists i.e. don’t consult directly with clients and therefore lack that crucial experience you have.

    Anyhow, I see my friend Stephane has responded with some good points – he is always around the same hang-outs as me these days ;) Point 2 is valid, but I do strongly believe EVERY visit is an opportunity to convert.

    Of course, when reporting to senior management, removing bounces (single page visits or visits less than X seconds) is key to communicating your ‘effective conversion rate’. That is, “removing the noise, our conversion rate is…”

    My point is that the conversion goal for each visit, same visitor is not going to be the same i.e. a conversion for a ‘info researcher’ can be the download of a PDF prospectus, whereas when the visitor is in ‘sales’ mode the conversion is the enrolment.

    The key is that visitors have different mind sets and therefore different conversion goals during the sales life cycle. Its the opposite to the Persuasion Architecture (or persona) approach pioneered by Future Now. But that’s for another post/article…

    Brian

  6. [...] up is showmeanalytics.com. You might know my opinion toward unique visitors and Angie’s last couple posts have been around this topic. Very interesting. I love the [...]

  7. Marcelo says:

    Hi, great post, but does anyone have any reference (book, paper, study) about this particular subject?

    Thanks
    Marcelo

  8. @Marcelo – thanks so much for the comment. Off the top of my head, I can’t think of a book or white paper that deals specifically with this topic.

    That being said, Brian Clifton, who commented earlier in this post, has a fantastic white paper about web analytics in general and also wrote a great post on this subject.

    Hope that helps.

    Shelby