Australian Popularity contest – May 2007

We had a lot of positive feedback the last time we showed the performance of Australian presences in SL, so we thought it time to revisit. One thing that hasn’t changed since last time is the fact that the two major presences are the BigPond and ABC islands.

bigpondpopularitymay07.jpg

abcpopularitymay07.jpg

Some other Australian presences and their respective traffic:

1. Australia Sim -2776
2. Terra Incognita (Uni of Southern Qld) – 937
3. Esperance Island (AFTRS) – 244
4. RMIT Ormond Island (Public) – 298

Some of the traffic figures continue to place Australian locations well and truly in the big leagues. That said, the traffic statistics themselves are at best indicative, change every day and are also being removed in future updates. Hopefully the replacement functionality will provide an even clearer picture.

(Methodology: one-time sample (Saturday 26th May at 1.15pm AEST) utilising SL search functionality)

Comments

  1. Thanks for the detailed comments Gary – the single-sample methodology is certainly a flawed one and a rolling sample is a much better way.

    The point of the article was more general in saying that the Australian presences are still performing well. I think any claim for one ‘main meeting / social spot’ is premature and even may never be realised, particularly as more presences are established. That said – what do others think? Do you tend to think of one area in SL as your main social spot? And is it an Australian or international one?

  2. Thanks again Gary – point taken on the use of ‘a’ – that said I’m still very interested in what the wider Australian SL community’s perception of their favourite social spots are. Anecdotally, the campfire area outside the Billabong Bar and the ABC Sandbox are two of the main ones but interested where else groups gather.

    On traffic stats: my understanding is that LL intend on providing an alternative to the current traffic stat option. One of my main gripes with the current system is that it advantages the ‘bigger’ presences’. A company that can afford to buy four islands and combine them is going to get a huge amount more traffic (assuming they create something people want to come to) than someone who owns a mainland sim. It’s not unusual to see a hundred people on an island suite whereas any mainland sim will fall over around 50 people most times. Traffic stats are affected by this and hopefully any replacement traffic metrics take this into account.

  3. Ok, I may be wrong here – but doesn’t the ability to get larger numbers of people on one sim improve if there are nearby sims linked to share prim count etc?

    That was my point – happy to be corrected of course πŸ˜‰

  4. Gary – thanks for the explanation – I’d mistakenly applied some mainland sim stuff (linking parcels) to islands. And no-one is happier than me that both major aussie presences are doing well. I’m still looking forward to an improved traffic metric though as the current one is too indistinct.

    Tateru – you raise one of the more interesting issues about all of this – the methodology. Any methodology is open to criticism and I’ve always been up front that mine is a snapshot only. I’d be fascinated to see more discussion on the issue as I know SL Insider has an excellent regular feature on traffic stats as well.

    Again – a very worthwhile discussion to have all round.

  5. I believe there’s a better traffic system in regard to the way the numbers are calculated. I don’t believe the algorithm currently used is the best way and obviously neither do LL. Taking my own modest presence in SL – on search today it has a traffic of 520, yesterday it was 7. The big jump is due to half a dozen new avatars attending an education event on my parcel. I’ve had a couple of events that have sent the traffic to over 2000. Is that an accurate reflection? Not really – it’s actually a well used resource for new aussies, but not to that level. Hence you’ve not seen a series of articles claiming SLOz HQ is as popular as some locations on ABC and BigPond πŸ˜‰

    I agree totally that whatever is in place should be as publicly viewable as they currently are.

  6. Yep, I actually understand the concepts of mean, median and mode…

    My points are:

    1. That the algorithm as it currently stands places too much emphasis on some factors (percentage time versus total time in-world) and not enough on others (idle time, activity-specific metrics). Whatever LL replace it with may not be better, but I remain hopeful.

    2. That the current traffic stats are transparent, but really only to experienced SL users. A new business or NGO can easily be snowed by traffic figures. In an ideal world the figures would be as finessed as those available on the web – and it probably will end up that way.

  7. Thanks for the detailed comments Gary – the single-sample methodology is certainly a flawed one and a rolling sample is a much better way.

    The point of the article was more general in saying that the Australian presences are still performing well. I think any claim for one ‘main meeting / social spot’ is premature and even may never be realised, particularly as more presences are established. That said – what do others think? Do you tend to think of one area in SL as your main social spot? And is it an Australian or international one?

  8. Thanks again Gary – point taken on the use of ‘a’ – that said I’m still very interested in what the wider Australian SL community’s perception of their favourite social spots are. Anecdotally, the campfire area outside the Billabong Bar and the ABC Sandbox are two of the main ones but interested where else groups gather.

    On traffic stats: my understanding is that LL intend on providing an alternative to the current traffic stat option. One of my main gripes with the current system is that it advantages the ‘bigger’ presences’. A company that can afford to buy four islands and combine them is going to get a huge amount more traffic (assuming they create something people want to come to) than someone who owns a mainland sim. It’s not unusual to see a hundred people on an island suite whereas any mainland sim will fall over around 50 people most times. Traffic stats are affected by this and hopefully any replacement traffic metrics take this into account.

  9. Thanks Lowell. A much larger range of brand figures which show comparatively how Australian brands are doing is published every monday at http://www.theprojectfactory.com

    Rather than say this is at best indicative (which you keep saying) perhaps the readers would want to know why it is the ONLY way to compare like-for-like stickiness to sims – as it goes beyond just ‘visits’. To help people I list below how the figures are worked out. Also this changes daily so you do need to consider mapping a rolling average of these figures. Finally if LL remove these figures then we could end up with the equivalent of a dot com bust – at the moment being able to see these figures openly is a bit like being able to see ‘web traffic’ stats in the early days of the internet vs companies over inflating their figures. I for one think we need to put pressure on LL to NOT remove these figures.

    Also you need to add the figures up for the Australian sims – ABC TV Australian dont own the Virgin nightclub obviously! I just did a check on the figures as they have updated this morning and here are the latest (you can check inworld)…Notice that the small 16th of a sim bar area of the Pond has reached record levels, showing how the Oz community now see that as a main meeting/social spot.

    BigPond – TOTAL 18004
    Billabong bar 8014, Pondex 2731, The Pond 2594, Uluru 996, Harbour Bridge 814, Pyramid Spa 744, Ponderosa 648, Club Illusion 526, The Dome 454, Pondex 273, Opera House 210

    ABC – TOTAL 11376
    Sandbox tower 8687, Broadcast Tower 1172, Triple J 593, Eco House 346, Amphitheatre 310, Media Pod 181, Dreamtime Cove 87

    Reporting dwell statistics in Second Life as a measure of engagement Traffic is a number allocated to each parcel which is based on the amount of residents who visited, and the time spent on that parcel out of their total time in world that day.

    Every user gets a set number of traffic points to give out during the 24 hours between midnight and midnight. The user’s points are then evenly divided between those parcels. So, if I was online for 1 hour and spent 20 minutes on resident A’s parcel and 40 minutes on resident B’s parcel, resident A would get 33% of my points and resident B would get 66%. Alternately, if I only spent 5 minutes online and spent all of it on resident A’s land, they would receive all of my points.

    Gary

  10. Ok, I may be wrong here – but doesn’t the ability to get larger numbers of people on one sim improve if there are nearby sims linked to share prim count etc?

    That was my point – happy to be corrected of course πŸ˜‰

  11. My original line – “a main meeting/social spot.” Notice the word ‘a’ – not saying one, which you turned it into?

    Sure there may be more, but at the moment that single ‘spot’ is busier than ALL the other branded areas in SL (expect pontiac and ABC). Obviously it will not compare strongly to various and many clubs (where there is endless rights infringing music). Not sure why the word premature comes into this – the’ Bong’ area as it if affectionately known has been between 3000-8000 for the past 3 months!

    We should be celebrating this space which actually does not rely on any major Aussie ‘icons’ to promote it as I designed it as my original creation based loosely on a few look and feel pics I took when extensively travelling in RL in WA and QL in 2003.

    BTW (rather than posting to another post) the quality of the new Australia sim is very poor, the builder has seeked no permissions (using pics of uluru and aboriginal art) and actually cheapens the Australia presence globally – see this image for instance, Uluru is turned into a lump of I don’t know what! http://www.flickr.com/photos/garyhayes/515323325/
    Also across the flyover road that cross uluru (!) there is a shopping mall full of ads (which looks nothing like the supposed Darling Harbour complex!). Now that really is being disrespectful.

    Best Gary

  12. Your assumptions are again out.

    ABC is one sim and is number two brand using these figures. The Pond figures, two thirds of the traffic is from one sim The Pond – the data above shows that 12000 out of 18000 is from that one sim.

    Best Gary

  13. Thanks Lowell. A much larger range of brand figures which show comparatively how Australian brands are doing is published every monday at http://www.theprojectfactory.com

    Rather than say this is at best indicative (which you keep saying) perhaps the readers would want to know why it is the ONLY way to compare like-for-like stickiness to sims – as it goes beyond just ‘visits’. To help people I list below how the figures are worked out. Also this changes daily so you do need to consider mapping a rolling average of these figures. Finally if LL remove these figures then we could end up with the equivalent of a dot com bust – at the moment being able to see these figures openly is a bit like being able to see ‘web traffic’ stats in the early days of the internet vs companies over inflating their figures. I for one think we need to put pressure on LL to NOT remove these figures.

    Also you need to add the figures up for the Australian sims – ABC TV Australian dont own the Virgin nightclub obviously! I just did a check on the figures as they have updated this morning and here are the latest (you can check inworld)…Notice that the small 16th of a sim bar area of the Pond has reached record levels, showing how the Oz community now see that as a main meeting/social spot.

    BigPond – TOTAL 18004
    Billabong bar 8014, Pondex 2731, The Pond 2594, Uluru 996, Harbour Bridge 814, Pyramid Spa 744, Ponderosa 648, Club Illusion 526, The Dome 454, Pondex 273, Opera House 210

    ABC – TOTAL 11376
    Sandbox tower 8687, Broadcast Tower 1172, Triple J 593, Eco House 346, Amphitheatre 310, Media Pod 181, Dreamtime Cove 87

    Reporting dwell statistics in Second Life as a measure of engagement Traffic is a number allocated to each parcel which is based on the amount of residents who visited, and the time spent on that parcel out of their total time in world that day.

    Every user gets a set number of traffic points to give out during the 24 hours between midnight and midnight. The user’s points are then evenly divided between those parcels. So, if I was online for 1 hour and spent 20 minutes on resident A’s parcel and 40 minutes on resident B’s parcel, resident A would get 33% of my points and resident B would get 66%. Alternately, if I only spent 5 minutes online and spent all of it on resident A’s land, they would receive all of my points.

    Gary

  14. So I will correct you…prim count and agent (avatar number) is fixed to each sim (a quarter of a processor).

    The limits are – 15000 prims and you can control the number of agents on a sim (depending how lagged you want it to be). I know some owners who have set their agent limit to 120 and got that many on. People talk about oh you can only have 30 avatars in one sim (true about a year ago) now you can set the upper limit. 60-70 is actually more comfortable on the series 5 processors, and that is getting better. It really comes down to how many attachments and scripts each avie carries. For example you could get 100-130 noob avies (jeans/tshirt) on a sim without major performance hit.

    Best Gary

  15. I’d like to see the project factory’s actual methodology before necessarily trusting figures that implicitly self-promote.

    Is it published anywhere?

  16. Sims can’t share prims or other resources.

  17. My original line – “a main meeting/social spot.” Notice the word ‘a’ – not saying one, which you turned it into?

    Sure there may be more, but at the moment that single ‘spot’ is busier than ALL the other branded areas in SL (expect pontiac and ABC). Obviously it will not compare strongly to various and many clubs (where there is endless rights infringing music). Not sure why the word premature comes into this – the’ Bong’ area as it if affectionately known has been between 3000-8000 for the past 3 months!

    We should be celebrating this space which actually does not rely on any major Aussie ‘icons’ to promote it as I designed it as my original creation based loosely on a few look and feel pics I took when extensively travelling in RL in WA and QL in 2003.

    BTW (rather than posting to another post) the quality of the new Australia sim is very poor, the builder has seeked no permissions (using pics of uluru and aboriginal art) and actually cheapens the Australia presence globally – see this image for instance, Uluru is turned into a lump of I don’t know what! http://www.flickr.com/photos/garyhayes/515323325/
    Also across the flyover road that cross uluru (!) there is a shopping mall full of ads (which looks nothing like the supposed Darling Harbour complex!). Now that really is being disrespectful.

    Best Gary

  18. Gary – thanks for the explanation – I’d mistakenly applied some mainland sim stuff (linking parcels) to islands. And no-one is happier than me that both major aussie presences are doing well. I’m still looking forward to an improved traffic metric though as the current one is too indistinct.

    Tateru – you raise one of the more interesting issues about all of this – the methodology. Any methodology is open to criticism and I’ve always been up front that mine is a snapshot only. I’d be fascinated to see more discussion on the issue as I know SL Insider has an excellent regular feature on traffic stats as well.

    Again – a very worthwhile discussion to have all round.

  19. Your assumptions are again out.

    ABC is one sim and is number two brand using these figures. The Pond figures, two thirds of the traffic is from one sim The Pond – the data above shows that 12000 out of 18000 is from that one sim.

    Best Gary

  20. So I will correct you…prim count and agent (avatar number) is fixed to each sim (a quarter of a processor).

    The limits are – 15000 prims and you can control the number of agents on a sim (depending how lagged you want it to be). I know some owners who have set their agent limit to 120 and got that many on. People talk about oh you can only have 30 avatars in one sim (true about a year ago) now you can set the upper limit. 60-70 is actually more comfortable on the series 5 processors, and that is getting better. It really comes down to how many attachments and scripts each avie carries. For example you could get 100-130 noob avies (jeans/tshirt) on a sim without major performance hit.

    Best Gary

  21. I’d like to see the project factory’s actual methodology before necessarily trusting figures that implicitly self-promote.

    Is it published anywhere?

  22. Sims can’t share prims or other resources.

  23. I believe there’s a better traffic system in regard to the way the numbers are calculated. I don’t believe the algorithm currently used is the best way and obviously neither do LL. Taking my own modest presence in SL – on search today it has a traffic of 520, yesterday it was 7. The big jump is due to half a dozen new avatars attending an education event on my parcel. I’ve had a couple of events that have sent the traffic to over 2000. Is that an accurate reflection? Not really – it’s actually a well used resource for new aussies, but not to that level. Hence you’ve not seen a series of articles claiming SLOz HQ is as popular as some locations on ABC and BigPond πŸ˜‰

    I agree totally that whatever is in place should be as publicly viewable as they currently are.

  24. Yep, I actually understand the concepts of mean, median and mode…

    My points are:

    1. That the algorithm as it currently stands places too much emphasis on some factors (percentage time versus total time in-world) and not enough on others (idle time, activity-specific metrics). Whatever LL replace it with may not be better, but I remain hopeful.

    2. That the current traffic stats are transparent, but really only to experienced SL users. A new business or NGO can easily be snowed by traffic figures. In an ideal world the figures would be as finessed as those available on the web – and it probably will end up that way.

  25. Tateru – very simple.

    Step 1 – Go inworld.
    Step 2 – Type SEARCH/PLACES. Put in some of the big brand names like, Microsoft, IBM, L Word, Pontiac, BigPond, ABC etc:
    Step 3 – Look at the numbers. Add them up to get a total for the brand.

    Clear for you? In otherwords these figures are published all the time for all to see. This is why I keep saying if they are removed we will have no open figures to compare and don’t understand why Lowell thinks there is a better traffic/dwell system.

    Gary

  26. You can also get the figures using the backend LibSL to automate the process (which we do too to cross check) – then apply simple rolling average equation to smooth them over the week.

    But I would suggest if you are concerned about self-promotion just go and look yourself. The figures are obvious and visible to all.

    Gary

  27. It would be if it was averaged over a longer period cause here your just talking about spikes. Your rolling average is therefore probably around 300 or so…Is this math 101? Take a series of readings over 14 or 21 days then do an average!

  28. Tateru – very simple.

    Step 1 – Go inworld.
    Step 2 – Type SEARCH/PLACES. Put in some of the big brand names like, Microsoft, IBM, L Word, Pontiac, BigPond, ABC etc:
    Step 3 – Look at the numbers. Add them up to get a total for the brand.

    Clear for you? In otherwords these figures are published all the time for all to see. This is why I keep saying if they are removed we will have no open figures to compare and don’t understand why Lowell thinks there is a better traffic/dwell system.

    Gary

  29. You can also get the figures using the backend LibSL to automate the process (which we do too to cross check) – then apply simple rolling average equation to smooth them over the week.

    But I would suggest if you are concerned about self-promotion just go and look yourself. The figures are obvious and visible to all.

    Gary

  30. It would be if it was averaged over a longer period cause here your just talking about spikes. Your rolling average is therefore probably around 300 or so…Is this math 101? Take a series of readings over 14 or 21 days then do an average!

  31. (note to self. Do not use the less-than symbol in a comment)

    I had this one guy who used to visit my store parcel. He’d turn up regularly when the neighbours two parcels north were having avatar-sex, and stand on my roof, snooping.

    He was worth 200 points of traffic for his 20 minute stay. Conversely, I’m worth less than 1 point of traffic virtually everywhere I go. I’m either not there for long enough, I cross a property line, or it’s such a small percentage of my total time, that it just doesn’t factor in at all.

    Oh, and traffic isn’t being removed from SL – it might not even be being removed from the search system. It’s just not going to be used as a *ranking* determinant for search results.

    As for the methodology, most of the venues we’re measuring the majority of their land is unlisted parcels, with traffic not available by search. That means a heavily trafficked site with a listed parcel may appear to be busier/more popular than an even more heavily trafficked site where much of the build is on unlisted parcels.

    Already tried the search-and-add methodology, and it gave us a highly skewed (and easily gameable) result.

  32. (note to self. Do not use the less-than symbol in a comment)

    I had this one guy who used to visit my store parcel. He’d turn up regularly when the neighbours two parcels north were having avatar-sex, and stand on my roof, snooping.

    He was worth 200 points of traffic for his 20 minute stay. Conversely, I’m worth less than 1 point of traffic virtually everywhere I go. I’m either not there for long enough, I cross a property line, or it’s such a small percentage of my total time, that it just doesn’t factor in at all.

    Oh, and traffic isn’t being removed from SL – it might not even be being removed from the search system. It’s just not going to be used as a *ranking* determinant for search results.

    As for the methodology, most of the venues we’re measuring the majority of their land is unlisted parcels, with traffic not available by search. That means a heavily trafficked site with a listed parcel may appear to be busier/more popular than an even more heavily trafficked site where much of the build is on unlisted parcels.

    Already tried the search-and-add methodology, and it gave us a highly skewed (and easily gameable) result.

  33. Hi Tateru,

    Yes I actually travel to parcels too that are unlisted and add them to measurements as I have noticed quite a few ‘brands’ now removing themselves from search because they obviously dont want the world to see how ‘low’ there actual passing/dwell traffic is.

    I disagree with your comments as both you and Lowell seem to concentrate on spikes vs long term dwell usefulness (now that is a skewed perspective too). Like talking about an election and saying well my three friends voted for candidate X who didnt win there it is rigged?!. We have been taking readings over the past few months and the ‘order’ or popularity is pretty consistent. Here is the latest order based on all the parcels for each of these brands. I have removed 2 brands who refuse to have their parcels listed in search or even inform us of their full range of parcels. This strikes me as ‘cover-up’ poor performance (given that both are big builds – designed to keep avies on their land). So pulling out one or two examples, like your 20 min one stop person, or a group of friends staying most of one evening are just more examples of spikes.

    To me this is about comparison over longer periods of time. I will say again, comparison over long periods NOT one off, anecdotal events. Until there is an alternative, fair, agnostic system (vs companies and sims just telling us how popular they are) then this remains. Gary

    Positions based on dwell for we 28 may:
    (range from 19 000 at top to 48 at bottom.)

    BigPond
    Pontiac
    ABC TV Australia
    NetG Training
    Nissan
    Mercedes-Benz
    IBM
    Dell
    Mazda
    American Apparel
    Showtime (L Word)
    Microsoft
    MTVN
    Toyota
    Reuters
    AOL
    Sears
    Samsung
    Sony Ericsson
    Adidas Reebok
    Starwood Hotels
    Sony BMG
    STA Travel
    BMW
    Circuit City
    ABN AMRO
    Cisco
    PA Consulting Group
    Intel
    Sun Microsystems
    Sundance Channel
    Sprint
    The Infinite Mind
    Text 100
    Wired Magazine
    CNET

  34. Hi Tateru,

    Yes I actually travel to parcels too that are unlisted and add them to measurements as I have noticed quite a few ‘brands’ now removing themselves from search because they obviously dont want the world to see how ‘low’ there actual passing/dwell traffic is.

    I disagree with your comments as both you and Lowell seem to concentrate on spikes vs long term dwell usefulness (now that is a skewed perspective too). Like talking about an election and saying well my three friends voted for candidate X who didnt win there it is rigged?!. We have been taking readings over the past few months and the ‘order’ or popularity is pretty consistent. Here is the latest order based on all the parcels for each of these brands. I have removed 2 brands who refuse to have their parcels listed in search or even inform us of their full range of parcels. This strikes me as ‘cover-up’ poor performance (given that both are big builds – designed to keep avies on their land). So pulling out one or two examples, like your 20 min one stop person, or a group of friends staying most of one evening are just more examples of spikes.

    To me this is about comparison over longer periods of time. I will say again, comparison over long periods NOT one off, anecdotal events. Until there is an alternative, fair, agnostic system (vs companies and sims just telling us how popular they are) then this remains. Gary

    Positions based on dwell for we 28 may:
    (range from 19 000 at top to 48 at bottom.)

    BigPond
    Pontiac
    ABC TV Australia
    NetG Training
    Nissan
    Mercedes-Benz
    IBM
    Dell
    Mazda
    American Apparel
    Showtime (L Word)
    Microsoft
    MTVN
    Toyota
    Reuters
    AOL
    Sears
    Samsung
    Sony Ericsson
    Adidas Reebok
    Starwood Hotels
    Sony BMG
    STA Travel
    BMW
    Circuit City
    ABN AMRO
    Cisco
    PA Consulting Group
    Intel
    Sun Microsystems
    Sundance Channel
    Sprint
    The Infinite Mind
    Text 100
    Wired Magazine
    CNET

Your comments

Previous Posts