I recently asked an acquaintance of mine who works in marketing for a major franchise if he could supply me with the latest Neilson Netratings Report for Real Estate. Last year I received a few reports and whilst looking through them I came across some information that confused me. So I checked again and now I a even more confused.
Neilson Netratings rank sites who input their website code. It is pretty much only for big business as charges are pretty steep, but basically it works by monitoring visitors behaviour on websites.
Most of the major portals and franchises use it to gain advertising and a little chest thumping.
I am not disputing who is the most popular real estate site in Australia (for those of you that don’t know it is realestate.com.au by a long streak) but one thing that does not add up is the amount of page views the average user looks at when visiting a site.
Here are realestate.com.au’s February 2006 impressive results:
Amount of Visitors = 2,681,788 (yes that is 2.7 million)
Page Impressions = 91.5779 (average page views per visitor)
Average Stay Duration on Site = 10.56 Minutes (per visitor)
(Update) It is not Average Stay Duration – it means Average Session Duration.Thank you for helping me with the missing informaton. Please read Simon Baker (CEO Realestate.com.au) comments below.
Now unless I am a dummy, (and I may well be) this information tells me that for every visitor who comes to this website they view on average just over 91 pages ……………… in 656 seconds!
That is 7.16 seconds per page view! This is not enough time to load a single page on an average 56k modem.
On one of my portals the average time a person remains on the site is a about the same but page views are around 20 per user, giving them enough time to actually read a page and view some photos.
I may be reading the document wrong, but I have checked with a few people and if anyone can clarify this then please let me know.
20 Comments
George Watts
I agree, it doesn’t add up, I think these figures have to be inacurate in the extreme and I also think they can be manipulated despite all the hype to the contrary.
A con job in my opinion but I’m an advertiser so what else is there to go by ?
peter
George, I do not think it is a con job, but I just do not think it is very accurate, there is just too many questions is poses, and I think that nobody ever asks these questions, newspapers and the media just report it. I don’t for one second think it would dramatically change anything as I think realestate.com.au are so far ahead anyway, but I just think it is too in accurate to be used for anything but marketing purposes.
Simon Baker - CEO realestate.com.au
Peter
The average page duration on http://www.realestate.com.au is around 20 seconds not 7 seconds as reported above.
The reason for this is below …
ASD does NOT mean Average Stay Duration it means Average SESSION Duration.
Therefore the missing piece of information is the number of sessions per visitor per month. For http://www.realestate.com.au that is 2.92 for February.
Therefore, the avarege visitor viewed 91.6 pages over the course of 2.92 sessions each of which is 10.56 minutes in length for a total visitation time of 30.8 minutes per month.
Therefore, following your logic above, that is an average of around 20 seconds per page not 7 seconds as reported above.
The point is, Nielsen//NetRatings is the most accurate measurement mechanism we have and extremely more accurate than traditional newspaper reporting measures that are certainly significantly less accurate and much less frequent.
12 publishers, including some agents, sign up and use this mechanism and therefore all sites are reported on an equal measurement basis. While some may not believe the numbers, they cant dispute the relative size of the sites due to the same reporting mechanism.
peter
Thanks Simon, actually it still seems puzzling to me, 90 something page views per visitor is an astronomical figure even over 2 sessions. I just dont understand how someone can look at so many pages so quickly (as an average), even with broadband + firefox. The figures just seem to be way to high.
I understand that the software can only be accurate to an extent and if there were errors they would be consistent across all sites, so I am not disputing the size of the gap between each of the 12 sites. It is just the pages viewed.
I will test some pages tomorrow and see how long it takes me to look at a list of pages, I also understand that many users would look at photos and leave the property very quickly if it does not suit and go onto another until they find one they like.
Maybe everyone is using Firefox and getting through the sites listings alot quicker!
Have a great weekend and don’t buy too many more portals, or I will have nothing left to write!
Paul R
I’m an agent and I dont believe it either.
What about all the real estate sites that have their properties on their sites own web linked from portals like realestate.com.au ?
If realestate.com has 3000 agents with this set up realestate.com.au could be getting traffic from 3000 web sites where the visitors have never heard of or even been to realestate.com.au
That makes any comparision between the portals totally inaccurate and dont try to tell me they filter those things out because I don’t believe that either.
It is the only thing we have to go on but less and less people like me take any notice of it and as far as prices go I find the entry level fee is useless and you have to go above that which is obviously the plan and you have to spend far more then you anticipated to keep up.
Thats life I guess, but these figures aren’t fooling anyone anymore and I hope there is better measurement tool out there before long but I doubt it.
Simon Baker - CEO realestate.com.au
Peter
I have some further information for you that backs up the page time.
Firstly, according to Keynote, the system that measure page down load times, the average download time per page on broadband is ~9 seconds while the average download time on dial up is ~49 seconds.
Secondly, what we also know is that 63% of users are broadband, 15% are dial up and 22% unknown.
Therefore, assuming that that unknown split 50:50 – we can therefore see that 74% are broadband @ 8 seconds and 26% are dial up at 49 seconds.
Based on these weightings, the average time is 19 seconds per page while aligns very closely with the Nielsen//NetRatings times.
The bottom line is that Nielsen//NetRatings appears to be generating quite accurate results that align closely with other monitoring and measuring systems.
peter
Paul
Geez get on your high horse!
Look I have previously asked realestate.com.au before about this and I have checked out a few sites, there does not seem to be any code in these sites, so I do believe them.
This is not what my article is about, my article is all about accuracy, it is not having a dig at realestate.com.au. This site is clearly the leader, my article is about in accuracies in data, if it is inaccurate in one it is inaccurate to the same degree in all so I doubt it would have any effect on who is most popular and by how far.
Some of the information just does not add up as far as I a concerned.
You see to be well informed about this technology, however I doubt many other people understand it,
All I want to see is a statstical data system that we can all participate in small, medium and large businesses.
At the moment there is only 12 real estate websites Australia wide that have this code in them – this is not a decent snapshot of the real estate industry as I know that even my site in Tasmania would be ranked in the top ten judging by these statistics.
I doubt any other website/portal in Australia would get close to the top 5 but who knows, with only a fraction of rea estate websites using this data it is impossible to tell.
Maybe I might develop a tool we can all participate in one that can filter out certain aspects of sites.
So Paul, take it easy and please feel free to respond!
peter
Thanks again Siimon, but does this mean that it takes on avegare 19 seconds to load a page?
Paul R
Peter, the code is not in the agents web sites it’s in the listings the visitors to those sites click on – so when a visitor goes to agent a’s site and looks at their listings the traffic is actually attributed to the portal , realestate.com or domain.com EVEN THOUGH THAT VISITOR HAS NOT BEEN TO REALESTATE.COM (OR DOMAIN OR WHOEVER IT IS)
UNDERSTAND ? these portals have their figures substantially boosted by traffic (visitors) that may not have even heard of them.
The more sites that use this system the more traffic goes to the portal, as realestate.com has 6000 agents ? perhaps half of them use the system thats a big distortion in traffic numbers and it is not the same for other sites who dont have as many agents linking through.
In short portal A might get more actual traffic than portal B but portal B has larger numbers via this linking from people who have never even been to the portal so in that respect the accuracy of the figures in very much an issue and may not be an accurate guage of how many people actually visit a portal.
peter
Paul R
I have checked out a few sites with your comment and I think you are wrong, every link that begins with agents.realestate.com.au does not from researching 10 sites have the code !– COPYRIGHT 2003 NetRatings Inc. installed.
Of course any site that is http://www.realestate.com.au/agentsname will include the code but this is fair enough
Paul R
Im not an expert in these things but probably because it links back to the portal it activates the code there, same thing really.
Scott Hampson - Web Specialist - realestate.com.au
Paul, you are partially correct in your assumptions.
Some website portals out there do include traffic from data embedded into agent websites.
This will serve to increase the percieved traffic figures.
Realestate.com.au’s system is set-up to know whether or not a listing is accessed from the main site, or from a subsidiary product, such as listings embedded into an agent’s website.
The traffic from agent sites is not tracked as part of the Realestate.com.au traffic stats.
It is only tracked to provided the agent with a full picture of where users access their listings.
peter
Yes, I can see Scott that you are correct with embedded code. I have yet to see an agents site with this included. I still maintain however that it is nearly impossible for there to be around 90 page views per visitor.
NSW Agent
Interesting, I believe the ACCC will be taking an active interest in the way web sites calculate their figures and sell themselves to the public. Not before time.
Elizabeth
I agree with NSW Agent!
It will be interesting to see how the ACCC handles the larger portals and franchise groups who appear to be doubling up on visitors to their sites.
I first noticed this when I clicked on an image and saw it was coming from a different site.
The franchise groups get traffic by embedding images taken from their own sites and displayed on homehound
eg. This L.J.Hooker listing on Homehound (or any LJH listing on homehound with a hooker add on it!)
http://www.homehound.com/consumer/propertyDetail.do?
propertyId=872556&searchId=6191601&sortOptionId=6
The test-drive-a-career advertisement is taken directly from their site.
http://www.ljhooker.com/hhbanner/homeloan1.gif
so here both homehound and ljhooker are getting a visitor, but the customer only visits one site.
And Century21 has a Homehound logo at the bottom of its pages (not on the home page though, seems you have to search for a property) that is taken directly from Homehound.
eg. http://www.homehound.com/consumer/images/proplogo.gif
Hookers and Raine&Horne also have this. Ray white appear to have the image taken from their own site, so it looks as if they are not engaging in this activity.
All this seems to prove is that they are out to just make themselves bigger and bigger!
I may have this wrong – Peter can you confirm the above?
peter
Elizabeth, I am impressed! What you have said actually does make alot of sense and it is something that I have never thought to look into.
I will send this through to homehound for them to confirm (or deny) but it really does seem to be there for all to see.
Good get Elizabeth…..it will depend on how they treat this in their statistics, but as we all know – most companies just like touting big numbers, I however look at enquiry rates, as this is the only true gauge!
Regards Peter
Elizabeth
Hi Peter,
So did Homehound ever get back to you? Or are they still trying to get an excuse organised through their legal team.
Maybe they should also run past their legal team the contradictions they are telling the market!
The embedded link they have on the franchise sites says that there are ‘200, 000 + properties for sale, to rent or lease’. See http://www.rh.com.au/
But their latest newsletter that I got via email says that there are only 198, 426.
B2C investigations are what the ACCC really go for – I think that they have a lot of explaining to do.
Either that, or they just simply cannot count.
Dodgy!
Eddie Thomas - GM homehound.com
Dear Peter and Elizabeth,
Your pursuit of accurate and truthful figures is commendable and I assure you that homehound.com not only agrees but is a driving force behind the push for accurate reporting of online statistics in the real estate business. This is precisely why we participate in the Nielsen//NetRatings Site Census and Market Intelligence programs. Homehound.com does not engage in any activity or attempt to artificially inflate our traffic figures.
In response to the specific points you have raised about homehound.com:
1. The L.J. Hooker image (advert) displayed on L.J. Hooker listings on homehound.com.
L.J. Hooker is an active supporter of Homehound.com and their advertising messages on their listings are indeed served from the L.J. Hooker web servers. This allows L.J. Hooker to change the content of their advertising whenever they please.
The important thing to note here is that this is an image that does not contain any Nielsen code (or any other measurement code). Without Nielsen code, no user session or page view is recorded by Nielsen when this image is served. The only thing that happens in this instance is that our page makes a request to the L.J. Hooker server to deliver the image. No user session or page view is recorded. It is simply a task that the web servers do in the background with absolutely no effect on homehound.com traffic figures reported by Nielsen.
2. The homehound.com image displayed on franchise groups’ websites:
This is the same scenario as the above, yet reversed. So again, no user session or page view is recorded for homehound.com. This is because our server simply delivers an image (without embedded code) to the franchise group website when requested, and no record of this request is recorded by Nielsen.
3. 200,000 listings.
Our listing numbers obviously vary from day to day – in fact we have essentially doubled our listings volume in less than 12 months. For the most part, we have well over 200,000 active listings on any given day. However, on the particular day we sent out the newsletter with the numbers you quote we had slightly less than 200,000 – approximately 0.7% less. As I said, homehound.com is a driving force behind the push for accurate statistical reporting, so on the day we sent the email to the industry we stated the exact figure at that point in time
Nathan
Hi all,
We operate in a different classifieds vertical, and we use Netratings to measure the traffic of a number of our sites.
Our Netratings statistics DO vary from our other measurement software, such as Google Analytics and our internal web server reporting, but usually not by too much ( /- 10%).
More so in the past than recently one of the key metrics of a web site has been page views as these determine the amount of ad inventory available (with the increase of video, ajax etc the time spent on a site is steadily becoming more relevant though).
All sites want to report traffic as high as they possibly, and legitimately, can so they count as many page views as possible. For classified advertising sites this means displaying photos of the item for sale (car, boat, house etc) in a html page where the NetRatings code can be used and allowing the user to browse every photo (resulting in a recorded page view for each photo). If you have a look at the HTML source code on a photo page for realestate.com.au you will find the NetRatings code.
So regardless of whether you consider viewing a photograph as being a valid page view or not it is technically a page view (and within the rules of NetRatings terms and conditions). Consequently it is very easy to view 90 or more pages in a month while you are looking for a property (in fact if you are actively in the marketplace you are going to view many, many more than this).
Nathan Huppatz
Web Publications Pty Ltd
Paul D
A property is a property is a property. It doesn’t matter whether there are 2 photos or 25 photos, it should count as 1 separate viewing for one separate person. That is the start and finish of it. You can’t measure things differently. You can put whatever spin on it you like, but one person looking at one property should count as one property view. Newspapers love to say their readership is about 3 times the circulation numbers, due to the fact that people live in households where there are about 3 people. Never mind the fact that mum is too busy to read the newspaper and wraps the garbage up in it, and the four year old cuts it up with the scissors and makes papier mache out of it.