State of the web: Australia’s online traffic

What about technology websites? 🙂

Would be interested to see some stats around user engagement — for example number of comments.

You didn’t mention gumtree.com.au the market leader in Classifieds in Australia< Bit of an oversight

I think the CPM up top is off. Mainly because of the $1.9m revenue ~50% is search (ie is derived from clicks not impressions). If you take that out and re calc the average CPM again back of envelope sits at more like >$5, this feels intuitively more correct.

Interesting analysis overall though. Thanks for it.

Would love to see the page impression figures on the non audited sites if they turned off auto refresh … Odd to see austereo doing 3x the imps of myspace with 1/3 the users.

Love the analysis but think itd be more robust using netview and not MI as Netview is harder to game than MI.

I know this isn’t the core of the article, but in talking about an ‘average CPM’, this is only useful if you’re trying to reach an average user with an average message. Once you start to care about targeting through context, and attention/engagement through quality content, then naturally the CPM will rise.

It’s like expecting an average CPC across every search term on Adwords. It’s irrelevant and distracting.

good stuff, i’d like to see this every month.

would it be worth using daily average UBs as the topline figure? ABCe in the uk just started using this rather than monthly as it helps get over probs of cookie deletion = more accurate. just a thought

Ebay?

be good to see sports clubs, sports codes, sports news, games and how they stack up against other online properties?

Thanks. Would be made incredably helpful with an insight into site usage and engagement, in addition to advertisers. Who is advertising most? Who’s being origonal?

Tim,
Interesting review – my topline feedback – avge monthly UBs are a waste of time as there is so much duplication and frankly little credibility in, as you say, a country of only 22m with 17m active online. If you are going to use MI at all, use average daily UBs as you did later on.

Personally, and this is not yet IAB policy (!) I think the sooner we kill off UBs as an audience metric the better, so would have to agree with Ben S to use Netview for now, even though we know many smaller sites aren’t yet picked up.

With the advent of hybrid, hopefully sometime in 2010, we will all have access to even greater accuracy and again people based not browser based metrics.

Game attempt to gauge avge revenue and cpm per page but too may variables here – # of ads per page, pages not carrying ads, differing cpms per ad on the same page, video, sponsorships, house, direct response, bonus – the list goes on.

Hi Tim,

Fabulous series, will look forward to your regular analysis. Personally, I struggle to find relevant statistics and analysis on Australian usage / traffic re: social media, plenty of global analysis out there, however locally the information is few and far between. If anyone know of Aussie Social Media stats I am missing out on please let me know.

Cheers, Lauren

Good overview Tim.
General note: this is Australian traffic only and based on sites that are being measured with Nielsen tags. If there are sites missing (like gumtree, eBay), they choose not to be measured this way.

@jonathan, official industry rankings are already based on Average Daily UBs. Since this article uses monthly UBs instead, think of this as unoffical custom rankings.

The Campaign Brief figures to date should be higher:
* Campaign Brief – Daily UBs 787, similar to BRW
* Campaign Brief WA – Daily UBs 227, similar to AFR Smart Investor

Keep an eye out for more detailed ‘Targeted Business’ and ‘Information Technology’ Category Reports to be released by the ABA shortly.

I think you’ve got your revenue / impressions ratio around the wrong way. 4.9b impressions and $156m revenue is 30 impressions per $1, not $30 per impression.

Which makes an awful lot more sense. A few cents per impression makes much more sense than tens of dollars.

thanks for this Tim. What would be interesting to see if there has been any decline in traffic since this time last year, as time spent in social media, particularly Facebook, increases.

I would also be happy to see this report monthly

I would suggest Nine News heads the list of News sites due to the amount of Hotmail users being redirected to ninemsn after logging out of Hotmail. The amount of UB’s vs PI’s shows it isn’t very sticky site, especially when it has half a million more UB’s but half the amount of page impressions.

Tim – Average daily UBs are a better measure of audience than monthly UBs, particularly for regularly read sites such as news, sport and finance.

Dumb question, but what percentage of the overall 88-million+ UBs is rolling in from overseas? At both Fairfax and News, a link from Drudge or Instapundit means an immediate surge, so are these stats included?

Another issue: When Rupert’spayywall goes up, does he put traffic magnets like Andrew Bolt (3 million a month) behind it? If he does, how long does Bolt stay on the reservation?

And a final point. If you want to see how mass delusion has become the mother’s milk of News Ltd.’s executive cadre, read Mark Day in today’s Oz. For the second week in a row he is preaching iPad salvation — a case he is mounting without actually understanding diddly about the Web, surfing habits or, indeed, the iPhone (see his aside about Skype users).

Efforts to algorithmically deal with unique browser count inflation have so far seemed to consider the problem as intractable. It only takes a relatively small proportion of the audience to be using security programs that regularly clear their cookies to continue to inflate these numbers. So expect UB numbers well into the 100 million in 12 months time.

It will continue and has been good for the industry overall for many years but means those who need to know useful numbers have to use whatever metrics works for their particular needs.

Thanks for clarifying that reference Tim. In fact, Habbo is in great shape both in terms of user numbers and revenue. It’s highly profitable, and its audited figures by Nielson are looking very healthy here and globally. Example ….

Habbo Australia has almost 4.7m registered characters and has shown consistent growth since its Australian launch in 2004. Habbo Hotel globally, remains the world’s largest world for teenagers and has 162 million registered characters – up significantly over a year ago and continues to climb. Pretty impressive given globally Habbo is ten years old!

Already in April, the average unique browsers are 26.1% higher than March. In March there was a 6.7% growth in unique browsers in comparison to February.

In regards to time on site, Hitwise figures show that Australian teens spend nearly twice as much on Habbo to any other social networking website. In March Australian teens spent an average of 47:05 minutes per user session in comparison to Facebook with 26:33 minutes per user session, MySpace with 20:55 minutes per session, Bebo with 20:43 minutes per session and You Tube with 18.49 minutes per user session.

Jeff Brookes

No mention of of porn.

What porn is there in Australia anyway? Most people browsing or genuinely interested in porn are going to visit an overseas site with fewer censorship concerns.

CarsGuide- UBs: 35,262 PIs: 15,203,966 – 431 pages per person?

(Mumbrella adds: Please see the comment stream below: that number was a daily rather than monthly UB figure)

@Jim It’s one of those awful websites that reload every time you select something from the drop-down box. You will need to do at least 5 (maybe more) reloads to perform a single search.

Great post. I haven’t seen this kind of analysis anywhere else (excluding internal reports) so I’m keen to see this series continue.

I think it would be useful to see who is advertising – who are the big spenders and are these advertisers using other media – tv, print, radio etc.

Or the war.

I’d love to see more of this overview. If it was possible, how about including traffic to Australian facebook ‘business’ pages? Not fans or members, but UB’s. Guess that would be hard to get, but interesting to see if fb offers a serious ‘alternative’ traffic route.

Thanks for all your hard work!

Any rankings on Networks?

I think Take40, Nova and InTheMix are all audited sites too.

Does http://www.abc.net.au/news really not rate in the top-20 news sites?

“Hotfrog is not a particularly useful directory (hence having an average site visit duration of just 1m,40s), yet still tops the list.”

I’m not a fan boy of Hotfrog, but a low average site visit duration is not necessarily reflective of a lack of usefulness. In fact, the opposite may be true.

Like a search engine, Hotfrog’s goal is to deliver results. Someone searching for a business is not going to stay long if they find the business they are looking for. The same goes for Yellow, Truelocal and others in this space. A low number here could be the desired result.

Disclaimer: My company consults in this space but does not work for Hotfrog.

Good article and great points raised through the comments – I’d also like to see this monthly.

I agree Tim re Bank West and their advertising revenue potential – there are a lot of non-traditional media sites out there who could consider developing advertising as an incremental revenue stream. Also, could Nielson’s Market Intelligence Lifestyle vertical do with some further segmentation?

As a few have mentioned, and it should be noted, these figures and commentary cover sites choosing (read: paying) to be audited by either Nielsen’s Market Intelligence product or the ABA. There are many sites who may not make the top rankings in some categories but are certainly relevant to advertising buyers – however the cost to audit it prohibitive.

And a question for Paul Fisher – any update on the hybrid?

Cheers – Denise

Tim – Hi,

Thanks for the report.
The Carsguide figures you have quoted are incorrect, and this has lead to some discussion (@Jim and @amused).
The correct UB figure is 762,314, not 35,262. @Jim, this equates to approx 20 pages per visit.
Please feel free to visit http://www.carsguide.com.au – you will note certain assertions about the site refresh etc are incorrect.

Howard
(Carsguide Marketing)

Interesting to see this covers Aussie eyeballs only and there’s discussion over how to find valid sources and construct useful reporting models.
There’s probably also a valid debate about whether foreign eyeballs have a value too, given that the ‘world’ can generally access the same sites, may travel to Australia, or in some way be influenced by advertising and/or content of various Australian sites.
The general argument that an Australian advertiser should only be interested in local eyeballs as they’re paying the bill may not always be the best approach – horses for courses.
We’re playing around with a ‘borderless’ approach to Australian, NZ and Pacific radio brand websites to see how stations perform or rank when taken outside their licence areas or licence type and compared on a global basis.
You can read the initial results at http://www.radioheritage.net ‘Borderless Radio’ and if anyone has better source suggestions than Alexa [!!!!] that a non-profit organization like ours can freely access for this project, please yell out.
From the initial Australian report we compiled, advertisers are probably missing viable markets through sponsorship of community and narrowcast stations!
There are some fantastic niche stations and programs out there.
Actually though, it’s scary that so many radio brands across the Pacific have such dysfunctional or non-existent websites in 2010.

I think Fish said it all very well. Monthly UBs have become probably the most meaningless metric going around town.

When I first started getting agitated over monthly UBs they were around 18m – way too high when the population was 20.5m but internet penetration was only around 75%. That was around 2.5 to 3 years ago. It hit around 35m about the start of last year. It’s now 86m!

Put simply this means that on average, people clear cookies weekly. The average hides the problem though – some people never clear their cookies, while others do so more frequently. The impact of this was that Average DAILY UBs are starting to move from a reliable metric (we used to say maybe 1% inflation) to a now suspect metric. My guess/estimate/gut feeling is that daily UBs are now probably over-stated by the 5%-10% range. With the rate of escalation of cookie deletion as evidenced by the growth in the monthly UBs it won’t be long before the daily UB count is unreliable.

I agree with Ben – we need to focus on people as all other media do – UA and not UB. Roll on the hybrid!

@Prat is right, the sites from mcm (take40), DMG (Nova) and Sound Alliance (inthemix) are audited, but mcm and DMG only have the green ABA audit tick from April onwards. The Allure Media sites (kotaku, gizmodo etc) and Gizmag are also new audited sites.

For the benefit of unaudited pubishers, I hope to share more about the common measurement pitfalls occurring, because we’ve been finding heavy overcounting and UNDERcounting across the board over the last 6 mths. You’d be surprised how differently some publishers have been counting things like rich media, photo galleries, slideshows, eMags, ajax pages etc. This can greatly impact their market metrics.

Publishers should ensure their measurement techniques aren’t lagging behind current ABA rules and their competitor’s practises.

Does anyone now exacly how the IAB determines their estimate on annual spend? We’re not one of the major publishers but ours is’nt in there from what I know.

@Jeremy.

I believe that the major publishers, aggregators etc “pool” their data through PWC – though I believe that Google aren’t included in that pool. This pool’s share is then estimated and cross-checked with other data sources (such as revenue estimates), and then the pool is grossed-up to provide the market estimate.

I could be completely wrong on this one, so people closer to the process may have greater clarity. Given that no-one can be forced to contribute their revenue data this appears to be a quite robust approach.

Hi Tim, no sport? This is the time of year that afl.com.au and nrl.com start to produce traffic numbers in the millions and overtake many traditional News sites for audience.
David

The people who’ve brought up the fact that off-site traffic and engagement (Facebook fan pages, YouTube videos, iPhone apps etc) aren’t measured are spot on…and I don’t envy the people who have to figure out a viable way of measuring it!

Hi Tim

APN Regional News Network missed out on inclusion in the News & Weather category. Hopefully this will be rectified with Nielsen next month, but the network has seen significant increases in both UB’s and PI’s in March 2010.

UB’s – 912,594 up 125% yoy, March 2010 vs March 09
PI’s – 7,930,992 up 202% yoy, March 2010 vs March 09

This growth is the result of the importance of local relevant news to people in regional Queensland and northern New South Wales.

Page impressions are so much easier to lie about than UBs although they are pretty easy to fudge too

If only Nielsen had some competition – their methodology is flawed

AutoRefresh users and we all know who they are (Austereo) are so obvious but they are allowed to continue to do this although it is completely misleading to the advertiser – this practice destroys CPM for everyone else

There is another problem with page impressions – the deeper a person gets into a linear sequence site the less chance they have to click – so this methodology is also flawed – deeper inventory is also worth less – eg. search result pages

@Real Yield I just wanted to clarify that (potentially) part of the overall problem is that we (i.e. Nielsen) actually measure page counts with and without autorefreshes so you can actually choose whichever you prefer (and no doubt many agencies who use our products do exactly that).

Nielsen’s methodology for the Market Intelligence product (page tagging based) measures pages SERVED which is neither a correct or incorrect way of measuring website performance but is certainly a figure that many within the industry want us to measure since it corresponds to the inventory sold in a typical campaign. It is a limitation of this type of methodology that without publisher cooperation we cannot easily distinguish between an auto refreshed page and a user refreshed page.

Nielsen’s methodology for the NetView product (panel based) measures pages REQUESTED by a human user in the active window (i.e. minimised and non active windows won’t count) which again is neither correct or incorrect but is also a figure that many in the industry want to know because it corresponds to the amount of inventory that is requested (and likely viewed) by human eyeballs. The limitation of this methodology is the volatility of measurement when using a sample to measure pages viewed.

The Business category above is a bit too broad, unfairly mixing general biz news, personal wealth, online banking etc with B2B sites. So to assist in benchmarking B2B sites, here is a custom report drilling down to Targeted Business websites. http://bit.ly/9Hwzwy

So if you have a website trying to reach industry-specific professionals, a daily reach of between 500 – 2000 UBs is strong, depending on how niche the industry is (after all there is a limit to how many gas pipeline workers you can reach online per day).

To save Tim from self-promotion, Mumbrella is #3 (as opposed to #34 above) and has very high loyalty (do we all need an intervention?). CBS Interactive’s BNET leads the pack in terms of daily reach.

Let’s get a sport category in there!

And BTW – daily UBs surely favour sites that have daily visitation, that is, news sites. I wonder if the IAB’s position on the best measure of UBs has anything to do with the fact their most influential stakeholders are the owners of news sites and have the most to gain from this…?

I think the most interesting thing from all of the above is how many people disagree with the figures and have different views on what the data means!!

Online is undoubtedly a great medium but using it to advertise a product is as crazy as paying to place a new brand of yoghurt on the set of neighbours and expect the stuff will fly off the shelves!

Melissa, I would have thought that a daily measure for a daily-used medium would be the logical way to go. Heaven help us if TV starts talking about monthly data – the numbers will be HUGE!

As a matter of fact US data is now showing that the #1 usage is now Facebook having just pipped Google which is back into #2 – neither of which are IAB (Australia) members, so little to gain there!

where is the sport category?
very un-Australian!!!

When it comes to Categories I guess you could go to thousands and then all sites would rank somewhere. Lies, Damn lies and Statistics!! Gotta love them.

I want to see a measurement that shows interaction with advertising content. If you like a category that measures effectiveness. It would be great if i knew that 1 in a thousand visitors would end up on my site and purchase something as an average. Surely it is possible to measure this and much more valuable than UB’s and PI’s. Which still sound like something out of a bad John Wayne Movie.

For those of you who don’t know Movie star, died before you were born, played cowboys and his real name was Marion!

LMAO Ben. I could have got them to a billion MUCH quicker. Simply add up the seconds and not the programme episode averages. Let’s see … around 48 minutes an episode is 2,880 seconds, with an average of around 1.1m per episode last year that gets us around 3 billion “people-seconds” per episode! Yes, lies, damned lies and statistics as Mark Twain famously said.

Mark, while I whole-heartedly agree with your sentiment around measurement of ad content, we need to look at who should pay for such measurement. The publishers do not own the ad-content – they own what surrounds it. Should they be beholden to low metrics because of poor creative? Isn’t this the responsibility of the client/agency?

I remember a famous story from the US in the early 90s when one of the agencies bailed up one of the TV broadcasters and complained that the ad minutes in the programme (I think it was Hill Street Blues) were lower than ‘what we paid for’. The broadcaster quickly pointed out that they had spent several million dollars producing that programme and attracted milions upon millions of viewers – and in 30 seconds their lousy ad lost them 10%+ of their hard won audience. The suggestion of adding a surcharge was even floated – good creative or else!

Hi Tim

Whilst Alexx does clarify this with the comments section.

Would really apprecaite you updating this post with DMG site figures being flagged as audited…

Thanks
Mo

Shopping sites would be a great inclusion!

Is anybody from Mumbrella going to respond to the query at 1pm this afternoon re the ABC’s web site traffic?

Grono thanks for the reasoned response. But if you can record a unique browser, surely it’s not that hard to set up a spider tracking where that browser goes. And subsequently if that browser then moves through to a purchase point or even an involved reaction. To an advertiser that would be so much more valuable.

@John Bennetts – look above in comments, they aren’t included in Nielsen’s data.

Agreed Mark – given that the period you track is short (one day if not one session), otherwise cookie deletion etc. will inflate your audience but destroy your unique tracking metric.

It sounds like you are focussing more on “website analytics” – with the emphasis on the word “site”. There are many packages out there that track usage on a site (or group of sites), and get closer to the “point of purchase” and how that user navigated to the “ah-ha, I’m gonna buy (or whatever)” point. These tools do analyses that web analytics don’t do. The objective of “web analytics” is to quantify (at a higher level) the market or the audience (not the browser). Web analytics tend to be more of a strategic rather than an implementation tool. That is, there is a place for Nielsen’s hybrid to quantify the market, and a place for webiste analytics tools to track usage and navigation on and between sites at a micro level.

Having said that, I still have some issue with attribution to “the last click”. In many (not by all means not all) cases this is erroneous – and tends to favour search over brand advertising. The more importance we point on measurement closer to the bottom of the funnel, I fear that we will divert funds away from creating consumer demand to help re-fill the funnel!

Great article. Any chance of covering retail? Monthly?

Just wanted to jump here on the Mark/Grono thread.

Mark – unfortunately as many advertisers have very different campaign objectives it is very hard to provide meaningful aggregated performance data. Yes, if I am an online retailer click data or online sales conversions are possibly the important metric (and I agree with JG here on the major problem of last click attribution) but many other campaigns want to drive a different kind of response (people buying a product in a supermarket, voting a particular way or using less electricity).

As an industry we definitely want to be able to identify the audience of a website and move towards the audience of a campaign but there are other factors in a campaign’s performance – creative quality being a major one.

I love that so many people are this passionate about online data. It would be good if you could pose some of the same data demands on offline media!

And as a number of people have pointed out a hybrid system bringing together the best of panel and tagging methodology will go a long way to help address many of the issues mentioned above.

@Mark and @John Grono

The only way to track people’s behaviour across multiple un-related sites, i.e. from News.com.au to SMH.com.au, is through panel based tracking. E.g. Nielsen’s Netview data.

However this wouldn’t track actions such as purchases effectively for a couple of reasons
– some of this will be done in secure environments (HTTPS) which Nielsen doesn’t track
– when you’re looking at purchasing actions too few people on the panel will be doing these to allow an accurate projection of the behaviour of the population.

Hi Jon. Good observations – panels are (at least at this point in time) the best way to provide some handle on the totality of online activity. I hesitate to use the terms “track” or “measure” – we’re producing most likely estimates.

However, there appears to be an underlying assumption that purchases are conducted online. What about someone who sees an online banner and purchases offline – clothing would be a good example. Also, what about people who see offline ads and go online to purchase.

My personal belief is that you can not design a panel either large enough or representative enough to “track” all the marketing and advertising stimuli that lead to purchase (either online or offline). I believe that the most promising method is multivariate non-linear microeconometric analysis. We have to remember what Einstein said “not everything that can be measured should be, and not everything that should be measured can be”. This is where I see the role of sales modelling flourishing.

However, for such models to provide realistic and robust outcomes we need robust metrics for each of the major media. The hybrid solution proposed by Nielsen et. al. would go a long way to getting robust online data into such models. Further, I believe that any medium that can not provide timely, granular accurate data to support such modelling in the years to come will suffer as clients increasingly demand deeper guidance on where to spend their next marketing dollar.

Tim, I’ve not had the opportunity to read through all the comments so this may have been brought up already but were there any categories relating to Beauty, etc?

@BenS – Can you shed light on why a site has to pay Nielsen to have their site stats included in MI? I mean they charge agencies to access the product, which must be a decent revenue stream. Surely they could just offer a discount on the product for those that provide the data?

It just seems a flawed system to me. They will never get 100% of the market, and it only caters for the big players in the industry. For example some music or fashion blogs may get a decent amount of UB’s, UB frequency and ASD. Sure they’re not full of ads (that will come) but these kind of stats would be very useful to PR’s for example.

I mean they’re just aggregating data into some tables, and they must know that not for profit or even government funded sites (ABC, etc) are never in a position to pay them. But they’re all an important part of the online eco system.

It’s an area ripe for competition, if only someone had a means of detecting and measuring all AU traffic without charging people to provide the information!

Now that would be a more useful investment by Chairman Conroy Tze Tung than a filtering system!

@Gai

“I love that so many people are this passionate about online data. It would be good if you could pose some of the same data demands on offline media!”

Here here! Very good point.

Tell me how online data is any more a valid currency than people meters? I think that the bus shelter monitoring service can claim a degree of greater clarity than the online data right now.

Hey Dex. Let me see. The argument goes something like this.

The publisher owns the site. The publisher then sells the advertising space to generate revenue and make a profit. But you think that the media agency (the buyer) should pay the costs to process the tags to make sure that the publisher can be trusted and is not pumping up the numbers? “Trust me, it’s from the server log” just doesn’t cut it.

In every audience measurement system I have seen anywhere in the world, the publisher pays the lion’s share – because they stand to make the most financial gain. Media agencies typically spend 8%-12% of their revenue on audience and advertising research. Is your organisation stumping up that sort of amount?

To say that Nielsen “are just aggregating data into tables” (and by implication that the data should be dirt-cheap or free) holds as much credence as saying the publisher is just aggregating some HTML and Flash into a webpage (how hard can that be!) therefore the ad-space should be free. By the way, I’m sure that the ABC pays their fair share regarding all audience measurement!

I’d be interested if someone from Nielsen could jump in with the number of tags that they process each month just here in Australia to add some perspective – Stu, Matt ?

@ The Honorable Mr Grono – Each month sees Nielsen process something like 5-7 billion pages / tags from Australian websites.

Bloody hell Stu – we don’t want that moniker catching on! (No chance I expect).

Thanks for the tag count. Some really rough back-of-an-envelope scoping numbers – even if you could collect and process a million tags for just one dollar we’re still talking a processing fee of $70k p.a. Now I suspect that I am out by at least a factor of ten (i.e. $10 per million tags sounds more reasonable) so I reckon $700k p.a. in processing alone.

I just don’t get why there is this mass expectation in the online world that everything should be free (and I hate paying bills as much as the next person).

As always, the data provides a useful indicator and its use as a trending measure rather than an absolute measure is what is most useful.

Digital has made its own cross to bear, if the USP of digital was its measurability then it is expected to live up to that. I personally feel that even with all the grey areas, the reasons for weighting some spends towards digital is quite clear.

I would love to see the whole “engagement” metric thrown away, just as hits were a decade ago.

Melissa – “…I wonder if the IAB’s position on the best measure of UBs has anything to do with the fact their most influential stakeholders are the owners of news sites and have the most to gain from this…?

No it doesn’t! The IAB’s position on the best measure of UBs has everything to do with what the buyers are telling us they want and need to grow online advertising expenditure. Simply put, that is accurate, credible, meaningful online audience measurement data in a user-friendly UI that they can more quickly, easily, simply plan and buy online media campaigns with. Ideally, they then want to be able stack up an online media schedule with other media, especially TV.

As John Grono – the MFA’s rep on the IAB Measurement Council – points out, many of our members, in fact the majority don’t have news sites but do attract substantial daily audiences. (Correction here JG – FB and Google ARE BOTH IAB Australia members!)

Denise Shrivell- hybrid has no specific delivery timeline but progress is good both by Nielsen and Comscore on their resepctive hybrid OAM products. We are also working on a standard measurement of online video, both content and advertising.

I think the interest in Tim’s story – kudos Tim for once again generating so much debate – shows that the measurement issue is still the biggest issue facing this industry, globally.

My view is we are about 12 months away,maybe less, from a much improved online audience methodology – hybrid, standardised metrics, people based not server based, an accepted online video measurement system, a better UI for media planners and buyers, reach & frequency planning tools, and the start of an online audience measurement data set that can be integrated with other media measurement eg TV and mobile.

As the old adage goes, it won’t happen overnight, but it will happen.

Apologies for the brain-fade Fish. I actually meant on the Measurement Council of the IAB.

I have to echo what Fish has said abouy hybrid – it won’t happen overnight, but it will happen. No-one should forget that the hybrid nut has not been cracked anywhere in the world yet, and that Australia is definitely at the forefront and leading the charge.

Dex in #81: If I was Nielsen, I’d be stoked with Tim publishing MI figures because it further validates MI in the industry, which is a huge cash cow for them. I paid for it one year myself on a high-traffic site, the costs were astronomical compared to the benefit I got (so yes, I have a bone to pick, not denying that). If Tim was publishing Netview figures surely Nielsen would send him a C&D because they don’t make nearly as much money out of that.

Google could decide tomorrow to publish a similar list from Analytics and knock Nielsen out of the water. I wish they would.

Paul, can you please enlighten me as to how Google would track people online who don’t use Google? Audience measurement systems need to cover all sources of traffic. Are you aware that there is a whole world out there online beyond Google? I sure hope you aren’t working in analysis of the online space!

Everyone uses Google, John. How does Nielsen track people online who go to sites that don’t pay Nielsen, John? No one is perfect, but Google has greater reach than most.

@Paul it’s not really the same thing. The sites that I most regularly visit are bookmarked. I don’t go to google for those sites. Therefore google analytics would miss all my traffic.

Further, once I’m at the site google has no knowledge of how long I am there, or which pages I visit.

Google analytics is great for analysing activity on google, but not on specific websites. It’s a different area.

amused: You do not understand Google Analytics.

Paul, I can see why GA seems like an attractive solution due to its ubiquity, but it’s a non-starter as an industry solution because Google is a publisher and it’s a conflict of interest for a single publisher to be operating the web ratings system (because the ratings directly influences where the dollars go).

Maybe they could release a cheap and nasty tool like alexa that would be used by fringe operators, but it couldn’t be the currency.

Very amusing Paul. Or were you serious when you said “everyone uses Google”?

Let’s see, Google has around 13m UAs out of 17.5m UAs – that’s only 4.5m active users a month who don’t use Google. That’s a pretty big hole in “everyone”. It also disguises the fact that not all of the online usage of active Google users is done via Google -and I’d wager that it is only a small proportion overall.

Nielsen’s approach is to sample around 6,500 people and track EVERYTHING they do. The software tracks every click – even when they aren’t online – and tracks every application (e.g. Word). It also knows which applications have the focus and are being actively used. This sort of “people profiling” provides insight into what is actually happening on the machine once a file is served. There is no “flitering” done for client vs. non-client. Of course non-clients don’t get a free ride so those sites don’t get broken out.

Of course no sample is perfect and for sites with small traffic (and little funds to pay for such research) there is insufficient active sample for them to be reported – but in my experience those sites are not bought on the basis of volume but on strategic fit irrespective of audience size.

And … ‘amused’ may or may not understand Google Analytics, but they sure seem to understand online and online audiences.

Don’t google run a spider across all web sites? The spider travels perpetually across the one billion sites and their average of 56 pages each scanning for change and recording interactions. Thats how it stays fresh and keeps us dictators up to date!
Regards
R. Mugobwe

RM – yes Google does run spiders across the web. It is searching for content – web pages – and not for usage. Put another way, that crawl is about as useful as an ashtray on a motorbike.

Of course, Google et. al. spiders, crawlers, bots etc end up contributing around 20% of all pages on the web – I sure hope they are being excluded from traffic counts (I know MI and NetView do).

@John Grono: Out of interest John do you have an agenda against online?

I agree that Google is not the be all and end all, however Google Analytics is extremely accurate when we want data regarding searches. 80% approx of Aussies use Google to search, so Google would be a very good tool to monitor where people are going via searches. Agreed Google cannot tell us which sites we frequent via our bookmarks, unless of course we have the Google toolbar installed, in which case is certainly could do (like Alexa…)

Not all, but many people ‘Google’ the site they wish to go to. Check out Google analytics and see the websites that are searched for. Check out their site traffic on a shoddy independent, like Alexa – it is pretty relative.

Big picture: we need an ABC online auditing tool. Either way the analytical and web monitoring tools already in place for online blow Radio, TV and print out of the water(.)

@Googler Perhaps we could get 1000 people nationwide to have boxes next to their computers, Iphones, homescreens etc and then press buttons when they are on certain websites – that’ll get more accurate results…

Can you look at sport in isolation as well please, Tim?

for everyone requesting tim to provide data i have a suggestion

invest in a neilsen subscription like most of us do … these numbers shouldn’t be free.

Hi Googler – most definitely not, as those who know me will attest to.

My goal is basically independent, robust audience data for the online industry – this is what clients demand. It is also exactly the same goal as for television, radio, magazines, newspapers, cinema, out-of-home etc. Each of these media is working to improve their current status quo and deliverable, which is a really healthy sign for audience measurement.

I agree that with respect to monitoring individual transactions, the monitoring and analytics packages for online are mind-boggling and streets ahead of all other media. Set-top-boxes in television offer promise – but also pose their own set of issues that need to be resolved.

The problem I have is that online does not have robust measurement of its audience as a quantum. We can tell within the micro-second what individuals (well, browsers – we don’t know a lot about the individual) are doing – where they came from, where they go to, how long they stayed etc. However, as a totality we do a very poor job of measuring OUR MARKET. We do a great job of measuring OUR SITE.

One only needs to look at the 80+m UBs (and growing every month) in MI to realise that the UB metric is sadly broken – a metric which seemed to be the defacto currency a year or two ago. The NetView panel data makes much more sense as a quantum (not as a diagnostic), has improved in the past 12 months, and still needs various improvements to move with the market.

A media planner needs robust audience data in order to create their communications plan. It’s very hard to confidently recommend a medium (let alone a site) to a client that lays claim to monthly UBs (let’s call it reach) of 80+m in a population of 22.5m. This is a fundamental issue that desperately needs solving for the good of the medium! Clients see 80+m and get rightly confused.

Just as I do not think that server-side based metrics can measure the audience for the medium (I stress the medium – not a site), I also don’t think that a panel of n=6,500 people can measure either the medium or the site to the level that we need – some sites will be ommitted as too small, and some will have data that bounces around way too much. But, the fact is that cookie-based measurement for an audience is now so over-inflated that the under-reporting of panels is preferable because it errs on the side of caution – which clients seem to prefer.

So, how do we get out of this conundrum? I believe the solution is a hybrid of these two methodolgies. This is something I have had numerous discussions with both Nielsen Online (Stu Pike “gets it”) and ComScore (Josh Chasin in the US is an acolyte for ‘hybrid’).

Hybrid takes the benefitsof server-based measurement – traffic – along with the benefits of panel-based measurement – people’s behaviour – and fuses them together. At its simplest you “clean-up” the quantum from the server-traffic (remove the spiders, crawlers, international traffic etc), and then apply the ‘people-based’ learnings from the panel (cookie deletion rates, mutliple computer usage – same person different IP, multiple person usage – same computer multiple people etc).

By merging these data sources you can ‘adjust’ the traffic to reflect the people quantum. After all, clients want to buy people and not traffic. You end up with people-based data that is correlated to and reflects the server traffic (which is a panel bugbear for sites).

As you rightly point out we also need an audit service to make sure that everyone plays by the same rules (for example, move your tag up the page and watch your PIs increase, auto-refresh rates, auto-play, click attribution etc). The IAB is working with the ABA and the MFA to draft these rules in order that sites can “pass audit” or at least report PIs to reflect these different scenarios.

Of course NONE of this will replace the website analytical tools which abound in the market. Google Analytics et. al. will remain in the marketplace for many, many years to come doing what they do best – analysing how people utilise and navigate sites.

Cheers.

Re comment 103 “invest in a neilsen subscription like most of us do … these numbers shouldn’t be free.”

Perhaps not, but there’s a big discussion to be had about Nielsen’s monopoly position and how much they charge agencies for subscriptions.

Agreed “Anonymous”. (By the way Tim, in my browser every comment appears as #1 – I am using IE 8.0.6001.18904).

There is an even a bigger discussion to be had as to whether there should be an open market – that is, anyone and everyone can quote whatever numbers they want using any software and algorithms – or whether we decide on an agreed “currency”.

Agreed currencies have the benefit that every publisher and every buyer are trading under the same conditions. They are generally reached by a consensus of seller, buyer and client, and “annointed” for a period of some years (to provide trading stability and lower cost over the longer term). They are also open to continuous review and inspection by representatives on a tripartite review committee – the IAB’s Measurement Council.

Currencies can have the drawback of “locking in” to a system that in an extremely dynamic market can quickly became yesterday’s thing and be surpassed by newer and better technologies. Currencies can also over the long term become pricey – I can never recall a currency going down in price despite new entrants to the market, which is why the term of the agreement is crucial so that the market offering and market prices can be regularly reviewed.

I also note that in Australia there are tripartite media currencies for: television (metro, regional and subscription); newspapers and magazines (for circulation, with readership being done by Morgan – the de-facto currency – but subject to a RFT from parts of the industry); Out-of-home; radio; and cinema. None of these media seem to have been harmed by having a single currency agreed upon and reviewed by seller, buyer and advertiser.

@ Anonymous

How many agencies actually pay Nielsen more than our products are worth to them?

I think the answer surely is zero (at least for all our repeat clients). If our products aren’t helping an agency do their planning better (and hence be more efficient and make more money) then we have no value and eventually no place in the industry. The reality appears to be that our products are of value to agencies and publishers just like they are in other markets around the world. A general rule of thumb in media seems to be that 1% of industry revenues are spent on measurement (split amongst all participants) and the Australian online industry is no different. That leaves 99 cents in every dollar spent on advertising for agencies and publishers to carve their own profit from. Sounds reasonable to me but I will acknowledge I’m in an obviously biased position.

I also think there are many fine companies out there that would argue against Nielsen being thought of as a monopoly but of course I’m not about to give any of them a free plug.

Mr Grono I agree with most of your comments regarding online measurement in the thread above.

I think the main issue with a hybrid method is deciding whether you calibrate data to the site centric measurement results or to the panel based data. I accept that the panel based measure can help you clean up many of the problems confronting site centric measurement. Most researchers are uncomfortable with the fact that site centric is a measure of browser activity not an individual (person). I accept that you can model this data to take into account its failings to create a surrogate respondent. The crucial issue for me is the number of things in site centric data that need to be modelled or calibrated for and the fact that they are ever changing (new browser technology or security technology on the browser side and deviousness of page set up and site design to rort the measurement on the publisher side).

Online gives the impression of being the most measurable media, in reality it is proving to be a very difficult audience measurement nut to crack. Don’t get me wrong there are numbers coming out of everyone’s ears, but few make sense and many contradict each other.

As you say uniformity of procedures for measurement are crucial. The science (it does exist sometimes) should require you to demonstrate an ability to replicate results. Unfortunately one site or even page with two different companies tags will often (usually) produce different results. Two different panels in the same market measuing the same sites will do the same, in many cases the results are alarmingly different and equally worryingly showing no consistent relativity between sites measured.

We are at the end of the day in the estimates business. I am less confident with online audience measurement estimates than any other media.

‘Researcher’, I am in total agreement with every issue you have raised in your erudite post. I suspect that you should be working on this project with us as your input would be invaluable.

My gut feel is that it will end up being site-centric down in the same way that the MOVE data has been generated – though that was simpler as the objective was “average daily contacts” as an OTS to be converted to an LTS. The difference between MOVE and what we are facing here is that the ‘known quantum’ with MOVE was verified and transparent – generally official government statistics. As you explain in your second paragraph there are many ways that the site-centric data can be ‘ever-changing’. I see the ABA as having a role here checking/enforcing compliance to agreed rules once they are established.

Researcher, should you want to ‘chew the fat’ about these issues further, please do not hesitate to contact me. My contact details are available from the MFA website http://www.medifederation.org.au … if you don’t already have them.

I would love to John, however my company would take a dim view of my activity on these forums. My comments and views don’t always match theirs.

I may however take you up on the offer in the future if circumstances change. I don’t work in the region any more but am still actively interested in the market and obviously still working in the audience measurement business.

Obviously this data is interesting to many, well done

please add health, healthcare, health products & services to the mix, maybe as subsection of Lifestyle, cheers

Number 1!
Number 1!
Number 1!

Great article.

you guys so need some visualisation sotware! agree need health also government

Online shopping sites would be great to be included..where is ebay, amazon or dealsdirect

I’d love to where this claimed traffic for Hotfrog is coming from.

A couple of years ago I would have said “fair enough” but for example where the Reed properties including Hotfrog would do very well for generic searches like:’home cleaning sydney this is no longer the case.

Google has penalised these general directories and they do not seem to get in the 10 ten generic search results very often and as we know what else matters?

Hello,

I would like to know where I can get a list of Online Media reps in Australia and the publishers or networks they work for ? Also if anyone could tell me which organization sells the inventory for ninemsn.com.au that would be great.

Thank you kindly.
Ruchika