Posts by Jamie Anstice
-
Hard News: Wikileaks: The Cable Guys, in reply to Simon Grigg,
And then it all takes another twist:
Although we should probably note that's an opinion piece from his Australian lawyer, who may not be an unbiased source.
-
@Dave:
Ctrl+alt+delete
the ultimate contract negotiation technique.I'm sorry, Dave. I'm afraid I can't do that.
-
Rich, Metservice claims fine, strong norwest (and kids don't actually notice the weather unless it's actually sleeting).
-
So does anyone know the best way to measure online traffic?
Let's face it, it's very important to have that information if you are a marketer, its basically critical and online needs to start presenting clean and readable data so it's easy to sell to ad buyers.
Marketing & publishing have different needs - marketers want to know how traffic arrived at the site, what happens on the site, and how often (and why) the users return. Publishers (and advertisers) need to know the demographics of the visitor base, and the number of unique eyeballs for ads (which is not to say that demographics isn't of interest to marketers, but it's not as vital as it is with advertisers).
On the publisher side, Nielson and Comscore are the big players, with support from Quantcast, Compete, and Google's Ad Planner. They get their data from a mix of (hopefully anonymised) ISP data, toolbars & panel members, and some of the service supplements this with on-page tracking codes. These services all measure in different ways and have subtly different ideas as what constitutes a unique user, and hence give different traffic figures (it's connected with extrapolating up from a small panel to the whole (usually US) web population, and also attempting to discount traffic by people accessing a site through multiple computers/browsers/deleting cookies/etc.
There was a storm in a blogcup a few months ago when comScore was charging multiple thousands to install & verify their on-page tracking code (which they used to triangulate with their audience panel data) - a lot of people thought it sounded like protection money, but from experience I know how hard it is to accurately instrument a site of even moderate complexity.
For those who aren't wanting to spend 10K per year, Quantcast has a similar service for free (but doesn't have the same weight with ad agencies), and Google's Ad Planner pulls in data from Google Analytics to supplement their other data sources if it's available.
None of the tools compares exactly with each other, and they tend to look very different to the figures from something like Google Analytics - oddly enough this topic turned up on Reddit in the last week (where there was a lot of noise about how the services are all rubbish, and one interesting comment from an actual ad buyer about what they look at when they use these tools: http://www.reddit.com/r/blog/comments/cq1lf/why_do_experts_misunderestimate_the_size_of_reddit/c0udcli ).
These work best in the US, where the majority of the panel data comes from, but Nielson and (I think) Hitwise both have NZ-specific data available, but be prepared to put your hands in your pockets.
-
This is exactly right and it's a total pain in the arse. When you're supposed to report on 'time on site' as a measure, and you know that it's essentially meaningless ...
Despite all the high-precision numbers, web analytics is more like meteorology than accounting (or possibly numerology) - but without a century of theoretical underpinning. Analytics isn't about graphs and tables of numbers, it's about telling stories (grounded in some sampled version of the truth), and then working from those stories to evaluate where you've come from, and work out directions to go next.
Analytics lives in a gray area between IT & marketing - the stories you can (easily) tell aren't the ones that the IT people need to hear, and they often discount web analytics because of this, which is unfortunate as there's a often a reasonably high level of integration required to plug in an analytics tool (or two or three) to collect enough data to tell good stories.
Measures like time on site are pretty poor proxies for measuring site success (are longer visits good because the user is engaged, or bad because content is hard to find and understand?) - better is to work out key actions for your site, then (as close as possible) directly measure these actions, and leave aside all the rest for slicing up your key action measures to tell good stories.
-
[T]he Times insists average duration of visits has stayed relatively consistent
Wearing my web analytics hat, most web analytics systems only track a user once per page (as close to the start of the page load as possible), and time-on-page is the difference between the load time of one page, and the load time of the next. The last page of the visit doesn't have a next page, so there's no time recorded for that page, and visit duration is really time-on-site-except-for-the-last-page. Single page visits tend not to contribute to the average-time-on-site at all (just like your batting average in cricket doesn't get divided by your not-outs).
-
is it wrong to think the robot would be better than the cricket commentators we have have the moment on the telly?
I'd have the robot batting at number 3.
-
Avatar is apparently playing in 3D in Hokitika... (and Nelson and Tauranga, Mt Maungauni, and Te Puke). It's not just the big smoke (and Hamilton),
-
I've been thinking a lot about this business scene recently, as I've got a new venture about to taxi down the runway, and while there are a lot of issues around market size & isolation & access to capital & all those usual things, I think that the Bach-Boat-Beemer checkout point is at a natural inflection in an entrepreneurial business, where the owner who is talented at making boats or software or cheese or whatever reaches a point where to get bigger needs less technical expertise and more managerial skills. There's got to be a conscious decision to get managerial - it's an easy non-decision to keep doing things the way they've always been done.
I think that NZ accepts entrepreneurs (especially those who make things rather than move other peoples' money around), but distrusts professional managers as at best a bunch of flash harrys who come along after all the hard work has been done & take all the credit & make all the money, and at worst hatchet men for the already ultra-rich.
I wonder if this is tied into the access-to-capital story - in a lot of places angel & VC money comes from entrepreneurs who have cashed out and then invest in the next round - if NZ business owners putter along quietly and don't have big exits then the capital gets stuck and doesn't roll forwards.
-
Apropos of nothing, I suspect that Peter Dunn has just been served a lesson in getting what he asks for.