IAB ‘retires’ unique browsers measure as reliable metric

Online advertising industry body The Interactive Advertising Bureau (IAB) Limited has announced it no longer feels unique browsers is a reliable metric for online traffic.

The announcement:

IAB Australia has issued a Notice of Intent to the industry advising it is retiring the Monthly Unique Browser (UB) metric in Australia as of 1st February 2013. Following that date, IAB has advised the Monthly UB metric should not be used as an audience proxy and in line with this move Nielsen will stop reporting this metric within its Market Intelligence product. Instead, the use of a people-based metric Unique Audience (UA) should be used.

The Monthly UB metric is being retired as it is no longer in line with the active online population or even total population. Monthly UBs now total over 90 million in Australia – outnumbering the actual human online population by a factor of more than five. The decision to retire the metric was made by the IAB Measurement Council with the MFA and has the support of AANA.

Gai Le Roy, IAB Australia’s Director of Research, commented: “The Monthly UB metric has been used in the Australian market for more than 15 years but now that it no longer equals a person, it has outlived its usefulness as a proxy for monthly visitation to website figures and should not be used.

“As an industry we are looking for more accurate and meaningful data and it’s clear that the Monthly UB metric is become increasingly inaccurate, less useful and even misleading. Indeed with the proliferation of devices and browsers in use by consumers, the disparity between browsers and actual people will only accelerate,” said Le Roy.

While the Monthly UB metric is being retired, the Average Daily UB metric may still be used as the IAB Measurement Council has determined that the active online population in Australia is aligned with the Average Daily UB numbers.

In conducting its review, the Measurement Council found that while the Monthly Unique Browsers inflation is most noticeable at a Market Aggregate level, the top five Publishers in Nielsen’s Market Intelligence (MI) often have a greater number of Monthly Unique Browsers than total Australian Active Online universe of 16.1 million people – an impossible and implausible amount.

Source: IAB press release

Comments


  1. willemrt
    3 Dec 12
    12:01 pm

  2. Wait.. monthly UB’s.. unreliable.. Daily = OK?

    That is fairly laughable, but then again it is the IAB. Did someone say banner ads are awesome?

  3. Kento
    3 Dec 12
    12:57 pm

  4. The problem with this is that Nielsen’s UA measurement is even worse. I’ll take five times out with UB over what they purport to track with UA…

  5. Youneak Yousa
    3 Dec 12
    1:12 pm

  6. UB’s have always been a bit of a ‘bullsh1t’ measurement, especially when the usual suspect publishers auto-refresh – totally flawed for their sites.

    Engagement is what it is all about. I want clicks that convert for my clients.

    Will the IAB be putting out an updated white paper / one pager on the latest measurement methods, why / how these were decided on etc?

    @willemrt Why are you ridiculing the IAB? Is there something I don’t know?

  7. Anonymous
    3 Dec 12
    1:38 pm

  8. I think I am with @willemrt – monthly figure unreliable but daily ok?

    What if, as I do, a user accesses the same site from a laptop, tablet and mobile each day?

    Cue angry comment from Paul @ IAB …

  9. Ben S
    3 Dec 12
    2:10 pm

  10. Bold. This wouldn’t have been an easy one to push through the IAB board. The industry has enjoyed the big UB figures for years, even when there’s 5 browsers for every 1 human here in AU.

    IAB should be applauded for moving this forward.

  11. John Grono
    3 Dec 12
    3:31 pm

  12. I think those that can’t see why an ‘average daily UB’ is VASTLY better than a ‘monthly UB’ – to put it quite simply – should not be engaged in the buying and selling of digital media.

    If you stop and think for a minute or two the device duplication and cookie duplication HAS to be many magnitudes higher when measured over a month than a single day. I would estimate that Average Daily UB still overstates the actual number of daily users, but the percentage would be single digits unlike the triple digits with Monthly UBs. However, Average Daily UB is still a reliable measure of traffic to a website, and a usable cohort for the total daily audience to a website.

    Kento raises the issue of the Hybrid UA data. To say it is ‘worse’ than Monthly UBs is frankly laughable. Yes there are issues with it for the long tail, and yes we’re working on alternatives. But there is NO way that the variance on UA exceeds that of Monthly UB.

    In fact Kento, it could be that for ‘small sites’ it may be that the ‘noise’ in the UA data is simply too large and we may have to revert to trading on Average Daily UBs for ‘small sites’ purely for stability reasons and forego demographic reporting. If I can give you an example. One publication I looked at had less than 30 of the 8,000+ sample browse that site in a month for months on end. Even if the sample was out by 100% – we’d still be looking at 60 out of 8,000+. In anyone’s language that is a small percentage of a large sample indicating a small audience. Meanwhile the daily traffic data was all over the place but was ‘gospel’ even if it made no sense.

    The problem is that the panel can’t access all users – e.g. government, financial, defence etc – and is subject to the usual sample variability. The problem with server side data is that it has no knowledge of ‘the people’ being served those pages and what they are doing (if indeed they actually do see the page – I currently have 20+ tabs open all contributing to ‘time spent’ metrics while only this tab is active). Clearly we need to merge the two sources together to apply empirical user beahviours to robust server traffic. Hybrid is a long way down that route but it still has its problems – none the least being the reporting of small sites.

    Overall the problem is that the market (buyer and seller) has become addicted to Monthly UBs just like an addict becomes addicted to heroin, and we’re witnessing the withdrawal symptoms.

  13. Anonymous
    3 Dec 12
    4:48 pm

  14. @John Grono

    “If you stop and think for a minute or two the device duplication and cookie duplication HAS to be many magnitudes higher when measured over a month than a single day. ”

    On the contrary, if I access a given site on 4 devices in one day, rather than a month, depending on the method of calculating the monthly uniques, then the quadruple counting of my devices causes a large variance as a proportion of the whole figure than it would in a month, making the daily figure more likely to be prone to error than the monthly.

    Would love to hear from the ABA on how they calculate the Monthly Uniques but I hope they haven’t been counting my 4 devices every day but just the once? Assuming not, the shorter period does not cure the issue.

  15. John Grono
    3 Dec 12
    5:46 pm

  16. @Anonymous.

    So given your example of 4 devices in a day (uncommon but not impossible, but will become increasingly more common) yes there is a quadrupling in the count.

    But if you measure this exact same setup over a month and rely on cookies as well (which are commonly and regularly deleted) then you not only get device duplication but cookie duplication. As an example, let’s say that every day you access a site on your (i) work PC, (ii) home PC (iii) smartphone and (iv) tablet – there is your quadrupling for any given day. But let’s also say that you run security software once a week each week on both the PCs and it clears cookies You would have 4 x home PC, 4 x work PC, 1 x tablet and 1 x smartphone in this scenario – a 10x inflation in ‘uniques’. Of course not everyone uses four devices every day of every month, so yes that 4-device day is relatively larger, but temporally across a month the over-statement is larger on the associated ‘uniques’ figure, and continues to inflate the longer the reporting period.

    The key thing is that, ANY server-side metric can NOT count device duplication. That is why we use a panel to at least estimate and take these factors into account. So basically tag-based and cookie-based metrics will always inflate uniques. ABA use a tag-based system so their data will be inflated.

  17. Anonymous
    4 Dec 12
    8:05 am

  18. Thanks John,

    Nothings perfect, as you say.

    How will sites spin this to advertisers?

    Yahoo’s Spreets, for example, will have to admit that its UA is only 22,000 users rather than 390,000.

    And Eatability, just bought by Optus for $5m, is actually talking to 36,000 users, not 964,000

    We’re just going to quote our Google Analytics figures!

  19. John Grono
    4 Dec 12
    9:13 am

  20. @Anonymous. The MFA and the advertisers have been pushing for the removal of this clearly wrong metric of Monthly UBs for quite some time, so agencies will be (in general) accepting. Mind you agencies are guilty of being addicts that couldn’t get off the junk being peddled, so there will be a cold turkey period for all.

    Also, don’t forget that media agencies work across all media, not solely on digital. When doing annual media plans, it has been clear for years that Monthly UBs were inflated so were in the main ignored at that stage of the process.

  21. Me
    4 Dec 12
    10:56 am

  22. Meh, there’s an entire underground industry devoted to inflating every known online measurement, UBs, traffic, likes, followers, friends…

    We all know everyone’s cheating – it’s like professional cycling, may the best doper win.

  23. Shamma
    4 Dec 12
    10:59 am

  24. Daily UB is a good indication of consumer loyalty to a site. The reality is, most sites do about 4-8% of their monthly uniques each day. Some do less than that and their monthly UB is spiked by one or two stories that pull in a huge amount of their ‘users’.

    PLUS it brings online in line with TV which reports daily ratings. No more, we have 950,000 monthly users (who use the site on average 1.1x a month due to search, hence we really have about 30k daily users)

    Let’s get these video sites and video networks doing the same. The ‘users’ tag sites throw around (ie ‘we make a difference in 9m Australian’s lives a month’ is completely redundant’). John and the IAB measurement committee have done the best thing here.

  25. Offal Spokesperson
    4 Dec 12
    2:00 pm

  26. Until they start reporting on the number of people browsing that have “Adblockers” active…then all the figures are redundant.

  27. Me
    4 Dec 12
    2:52 pm

  28. @Offal Spokesperson

    Conservative figures I’ve seen place use of adblock plus at about five per-cent.

    Increasingly government departments, universities and some big businesses are rolling it out to improve privacy, reduce traffic on their networks and lower the risk of exposure to malicious sites and software. IT areas love it for these reasons.

    I just love it because it blocks annoying ads.

  29. Sceptic
    4 Dec 12
    3:19 pm

  30. Trust the ad industry / media to fudge the stats, even when they can be clearly reported.

    Auto refresh – publishers

    B*llshit metrics – agencies

    This is the new world – why don’t we report the facts and sell around the true value instead of guess work, like traditional media? Tsk, tsk.

  31. Sam G
    4 Dec 12
    6:09 pm

  32. Too soon, sorry.

    Yes, monthly UBs are inflated. And yes, they should be phased out.

    But “small publishers” should be disappointed a solution wasn’t put in place for monthly measurement before bluntly discrediting the one metric they have to measure their audience.

    I’d be interested in the ABA’s point of view on this.

    Why not trialing an NZ-like monthly UA averaged over 3-6 months to increase the sample sizes, or supplying tagged state-based UA data for sites with <300,000 UAs?

    Effective Measure have been gifted a golden egg.

  33. Buyer
    4 Dec 12
    7:09 pm

  34. Really Sam. So far the ‘phase out’ period is best measured in years and it’s too soon? And it is not about “the one metric they have to measure their audience” as they have Average Daily UBs which is a much more realistic and reliable measure.

    It seems that certain parties are quite content to delay and delay so as to stick with a flawed metric as it best suits their business model. I’ve also heard on the grapevine that rolling averages are being looked at – and yes it would be nice if we had them now – but that is hardly a reason to keep Monthly UBs around any longer.

  35. Shamma
    4 Dec 12
    7:25 pm

  36. Sam G – the small publishers aren’t really a concern for the IAB, one could argue that is by design.

  37. Good Moron
    5 Dec 12
    3:14 pm

  38. I’ve been thinking about this whilst having a look at some daily UB figures on Nielsen and comparing them with the claims made by publishers.

    The monthly UB figures are nonsense in most cases.