{{ searchResult.published_at | date:'d MMMM yyyy' }}

Loading ...
Loading ...

Enter a search term such as “mobile analytics” or browse our content using the filters above.

No_results

That’s not only a poor Scrabble score but we also couldn’t find any results matching “”.
Check your spelling or try broadening your search.

Logo_distressed

Sorry about this, there is a problem with our search at the moment.
Please try again later.

Unique users or unique visitor numbers are often seen in press reports, internal dashboards and case studies, and are used as the basis for making decisions.

But there is a problem with ‘uniques’; they are not what they seem, and while this may be known by those on the inside, business decision makers are being fed information that is really very misleading.

Exploring the issue

Have a look at the table below from Google Ad planner (January 2011). This table shows the number of unique visitors for Facebook.com in the USA calculated in two different ways, and the difference is startling.

Unique visitors (estimated cookies) is worked out using cookies and then equating one cookie to each 'separate' visitor. The unique visitors (users) number is Google’s estimate of the actual number of people who visited the site, and Google does not disclose its calculation method. 

To cross-check the estimate of people we can look at comScore, which uses a panel based methodology to estimate audience, and in January they published figures estimating 153m people visited Facebook.com in the month.

OK, not an exact match by any means, but the comScore estimate is close to the Google estimate.

Why is there such a gap?

What this data shows is that there is an enormous gulf between the two numbers, and what it means is that each person has more than four cookies per month.

The reason for the gap is that the assumed one cookie = one person model is broken, and I don’t think that really surprises anybody these days.

However, it is the size of the fracture which is impressive. Unique visitors (based on cookies) overestimates the real audience by 350%.

Those four cookies come from people using multiple devices, multiple browsers or deleting cookies.  As people start to access the web from mobile devices in ever greater numbers this trend is set to accelerate.  

The key thing to remember is cookies track browsers, and browsers are not people!

Does this matter?

Yes! If you use ‘uniques’ as a measure you can make the wrong decisions. I have had a debate with a senior business leader who was being told that his marketing team had saturated the existing market because the number of ‘uniques’ was almost equal to the known market size.

When the number of uniques was divided by four it told a very different story and the strategy was put under much closer scrutiny.

So what are you using unique visitor numbers for? How many of your tools rely on cookie based tracking? What do your business decision makers think a Unique user is?

I believe it is time to put the unique visitor metric behind us and to rethink the way we measure and analyse online data. What do you think?

Matthew Tod

Published 31 March, 2011 by Matthew Tod

Matthew Tod is CEO at Logan Tod & Co and a contributor to Econsultancy. 

4 more posts from this author

Comments (15)

Comment
No-profile-pic
Save or Cancel
Avatar-blank-50x50

JO

Another problem with the cookie discrepancy comes from AOL users who use the service's browser to access sites. AOL can serve up a different IP address for each pageload!

over 5 years ago

Chris Moffatt

Chris Moffatt, Online Strategist and Project Manager at The Other Media

Interesting points. Within e-commerce, unique visitors has always struck me as a more useful metric than absolute visitors as - particularly with high price product - visitors tend to return to the site more than once when they're considering a purchase. So in calculations of conversion, it should give a more accurate picture of what's going on.

A "joined up" world of analytics would certainly help- the ability to track individual customers across their whole purchase journey, online and offline.

over 5 years ago

Avatar-blank-50x50

John Gallen

Hi there,
Good point but not particularly new.

I myself don't refer to 'unique visitors' anymore since I read that IAB was to set the standard term as 'unique browsers'. That said, I rarely hear too many others use the term 'UB', as yet. Bad habits are the hardest to break.

Do you have any ideas on the implications of this for 'total visits' and 'avg visits per user'?

On a broader note, its funny that the medium most touted as being so accountable and measurable is so incredibly flawed on these metrics.

Nice post, makes me feel good that I've been steering my clients in the right direction, thanks.

John

over 5 years ago

Avatar-blank-50x50

Berit Block

Hi Matthew, very interesting post and good to see this discussed and explained! We've also recently published a blog post discussing one of the most acute pain points our industry has experienced over the years – the debate over which “unique visitor” count is right.

You can read the full post here:
http://blog.comscore.com/2011/03/digital_analytix_iab_abce.html

over 5 years ago

Andrew Lloyd Gordon

Andrew Lloyd Gordon, Digital Marketing Expert, Speaker and Trainer at New Terrain Limited

Hi Matthew

I'd agree with others here that this information isn't new but it's still useful to be reminded of it.

Of course, whilst it's old news, it's surprising how few people understand that web data can sometimes mislead.

If we move to the use of 'Unique Browsers', do you think this would encourage organisations to build client login areas, paywalls etc in order to gain a clearer idea of how many actual visitors they have?

And do you think that the only web metric we can really trust is that of 'Conversions' or the 'Outcomes' our websites generate e.g. sales, downloads, customer enquiries etc?

over 5 years ago

Toby Kesterton

Toby Kesterton, Head of Digital at Lab Lateral

Good article - Perhaps the coming EU cookie directive will sharpen minds about cookie practices & reliance.

over 5 years ago

Avatar-blank-50x50

Matthew Tod

Thank you all for your comments, and while it is not new to you digerati, how many times do you read about those fictional 'Unique Users'? Everywhere! In fact I think we should start a campaign to haul across the coals every pundit who uses the term 'Unique User' in any article or presentation!

In terms of what we can trust the answer is events, and little else. Any tracking that relies on cookies is, to my mind, totally flawed. Sorry but stringing a sequence of events together using a flawed cookie model is simply no good enough anymore.

Bring on the EU I say, ban a useless tool! OK, a bit radical there but talk about shutting the door once the horse has fallen over and died!

As pointed out, if you want to track people you need to get a login, and the question then becomes how will you achieve that? Force it Facebook style? Make your site so useful people really do login - Amazon style? Or bribe them with points on a clubcard? It is sure to be an interesting area over the next few years!

over 5 years ago

Avatar-blank-50x50

kimatgiffgaff

Great article thanks. We have been using Google Adplanner Unique (cookies) metrics as a way to benchmark against over sites. It may be that even if it's overestimated, it's overestimated to everybody, so at a like for like it's still a fair comparison. Would you agree on that?

I always wondered how Google makes that estimation on real users visiting the site. It would be useful if they integrated that metric with Google Analytics.

over 5 years ago

Avatar-blank-50x50

kimatgiffgaff

I forgot to add, I assume this also has a massive impact on Google Analytics metrics. Unique visits are actually x3.5 times less that what they seem, and what about New visitors? As they are cookie based could we assume the same deviation?

over 5 years ago

Avatar-blank-50x50

Peter O'Neill

Hi Matthew - great to see you are continuing this debate. There have been various comments here from people agreeing that unique visitors is an outdated metric and agreeing it not accurate. But it is the implications of visitors being inflated that is more important, not the actual number being wrong.

Strategies are being decided based on intelligence using data such as the reported Visits to Conversion, the proportion of Return Visitors or using a Visitor based attribution model. Due to the issues you raised, these strategies are based on bad intelligence. The actions being taken are not optimal for that business.

Unfortunately, there doesn't appear to be an easy solution. More people are demanding visitor based reporting with various technologies promising exactly that. I think we need to be more creative and develop methodologies based on valuing website behaviour so can evaluate performance at visit level.

My thoughts on visitor counts being inflated can be found at http://www.l3analytics.com/2010/03/10/what-if-visitor-counts-are-inflated/.

over 5 years ago

Avatar-blank-50x50

Matthew Tod

@Kim In general errors can be ignored if they are constant, so comparisons are ok. BUT if the error is not evenly distributed then comparison does not work. So for example a site which attracts an older demographic, who typically use one computer and don't wipe cookies, will not look the same as one that has Econsultancy readers who are digitally savvy.

New visitor numbers - well what really mean is New Cookies placed, and this has limited relationship to the metric we really desire which is "new people who have not visited our site before". If you look at New Visitor conversion rates they are always flattered by research activity that has taken place on other computers, so it is not really that useful.

@Pete - I agree completely! Any strategy that is based upon Unique Users derived from an analytics tool is flawed. The only way to know the real number of people is to use a panel based survey from companies like Comscore... who now own Nedstat which might lead to something interesting!

over 5 years ago

Ben Sidebottom

Ben Sidebottom, Media Systems Manager at Essence

But the cookie issue is not something new to people, surely? This is a well known problem and is kind of accepted. Unfortunately there are little ways around this and with the new eprivacy directive, this is going to get even more inaccurate!

over 5 years ago

Avatar-blank-50x50

Alex Yefetov

Not to mention that some if not most internet providers use dynamic IP i.e. new IP for each online session...

over 5 years ago

Avatar-blank-50x50

Andrew Hood

Hi Matthew

Good point well made on poor decisions being made based on misinterpretation of the unique visitor metric.

I've always liked Eric Peterson's pyramid of metrics, and I think it's still highly relevant today (despite it being more than 5 years old and looking like the Illuminati gone wrong!):

http://web-analytic.blogspot.com/2005/08/pyramid-model-of-web-analytics-data_11.html

Personally I still think UVs and cookies are important, and we should focus on getting our heads around them rather than throwing them out. I've seen bad decisions made on visits, page impresions, conversion rates, BARB ratings, opinion polls, P&L, balance sheet... generally based on a misinterpretation of what a number means rather than the accuracy.

The Google Ad Planner stats are great; equally they "improved" their (still undocumented) user counting methodology this month and the ratio changed dramatically:

http://doubleclickadvertisers.blogspot.com/search/label/DoubleClick%20Ad%20Planner

A brief sample seems to indicate that the Google view is now substantially less cookies per device than it was in February (the "improved" view). Or everyone has suddenly ditched their laptops and iPhones.

Relatively speaking it's still a fair benchmark of market share. Absolutely speaking the seismic plates of the Internet have apparently shifted according to their stats. That in itself perhaps tells an important story about how these metrics should be used.

How many (or few) participants in the Comscore panel are being aggregated up to the 153m estimate? I dread to think. At least with cookies the limitations are transparent and vaguely predictable.

Back on topic: there are two separate but equally important points in your post.

1. The Unique Visitor metric has its limitations as a benchmark of market share (of online audience). But assuming the majority of online audience is anonymous, it's going to be a challenge to resolve that, and we need to contrast with other far less representative routinely used stats (e.g. BARB sampling <0.1% of audience to draw conclusions on TV ratings).

(Incidentally, I'm not sure rebadging Unique Visitors as Unique Browsers does anything but confuse matters more).

2. Cookies are pretty poor at keeping state for customer interactions over the long term, and we need to have stronger identifiers.

Matthew - your point on incentivisation here is bang on: long term, why do you consent to be tracked if there's nothing in it for you? This is also where all the privay issues kick in (another topic for another time!)

Unfortunately I suspect the EU are confusing the two things above and as a result nobody will win...

PS: Anyone from Google here: please can you burn the GA dashboard report where you say "1000 Absolute Unique Visitors" and then on the same page "1000 PEOPLE visited your site". You know it's not true. We know it's not true. It's not helping people get their heads around what unique visitors actually means.

over 5 years ago

Avatar-blank-50x50

Frank Doscher

Thank you for an interesting article.

I am working on an e-business plan; as you well know, accurate projection of unique visitors plays a critical role at accurate estimation of revenue and other factors.

I would greatly appreciate it that you could assist me with finding material exploring the conventional wisdom on this topic and any other related issue.

about 5 years ago

Comment
No-profile-pic
Save or Cancel
Daily_pulse_signup_wide

Enjoying this article?

Get more just like this, delivered to your inbox.

Keep up to date with the latest analysis, inspiration and learning from the Econsultancy blog with our free Daily Pulse newsletter. Each weekday, you ll receive a hand-picked digest of the latest and greatest articles, as well as snippets of new market data, best practice guides and trends research.