…the “average American” has one breast and one testicle…
– Jon Steel, Truth, Lies and Advertising, (John Wiley & Sons, 1998)
This is arguably among the most important real-world issues out there today (and has been, for some time now). It’s an issue in the offline world, as much as it is in the online world. Except that, online, there are a lot of efforts on to try and get an accurate gauge.
I recently came across readings on the topic, which I finally read this evening, after sufficient procrastination. First, the Forrester release on the topic*. And then, two Coremetrics documents, both by Eric Peterson of Web Analytics Demystified. Peterson’s first line is a great introduction
One of the best-kept secrets in online marketing is that most campaign attribution data is completely wrong and the models used to evaluate campaign performance are wholly inappropriate
John Lovett’s Point of View
The Forrester document discusses various issues, but the bottomline is really their model, which boils down to weighting three points of data, (1) frequency (2) recency (there is a time-period cutoff) and (3) time-on-site, which represents site engagement and interest. The image below, once you parse it, explains it all.
It’s an actionable idea, once you get the measurement in place. However, I’d love to see some testing behind the weightage method. Every minute spent on the website is equivalent to one exposure of the ad. How does one reach this conclusion? I agree that this is extremely hard to measure, but I’m sure there should be/can be research on the weightages.
A strong point that Lovett makes is that while most folks will agree that it’s important, there seems to be a paucity of strong case-studies out there. I also think it’ll be a strong source of competitive advantage, and would sympathise if someone who’s figured out the weightages already isn’t sharing them.
As Eric Peterson quotes Andy Fisher (Razorfish) in his white paper
“Coming up with a good weighting system is hard!”
Eric Peterson & Core Metrics
Eric Peterson wrote a white-paper (and a solution brief) on the topic, which was sponsored by Coremetrics. Half-way through the paper, I lost the thread somewhat, especially since I have never, and don’t, use Coremetrics. The paper starts with an intelligent analysis of the curent situation, which decries the absence of multi-touch attribution. Peterson calls it ‘Inappropriate Attribution’, strongly putting down all commonly used attribution systems. He then suggests the 3-touch view (something that Coremetrics does), which is where I started to lose the plot a bit. Perhaps when I get to fiddle around with Coremetrics tool, it might get clearer.
Peterson goes on to classifying each campaign as an acquisition, persuasion or conversion campaign, based on the the Appropriate Attribution Ratio (AAR). In other words, the performance of the campaign would define the kind of campaign that it is. This, I have a disconnect with. I advice clients to think about campaign objectives when planning them, not when reviewing the results. Because, campaign objectives would define the campaign copy, ad placement, and overall campaign design. Of course, this system can be used to see if the campaign met the desired objective or not. I’d also like to nitpick about the term ‘acquisition’. Peterson uses it to refer to acquisition of a lead, while a good part of the industry might refer to it as a sale. It’s a nitpick, but take note, or you might be confused when reading the paper.
Just before I was to post this, I figured out that this is part of Coremetrics’ overall masterplan to track individual users on the net. It has sparked off a debate between individual tracking and aggregate tracking. A subsequent post will parse that.
In all, an important topic, and it’s likely that there will be more posts on this as I explore.
*Note: I got the Forrester document from here, where you can download it if you offer pimp yourself as a datapoint for their database.
The New York Times predicts that data is the future of work.
I prefer this to Hal Varian’s previous statement about statisticians being the hot profession. Simply because statistics is only one of the many skills an Analytics profession has.
To quote the NYT,
As suggested by Daniel Pink’s assertions on the rise of a right-brained working elite, the ability to extract stories from a world of increasing and abundant data will be increasingly critical to many industries
It’s analytics that the future belongs to, not just statistics. That’s because statistics is the science of techniques, while analytics is a way of thinking. It’s larger.
Confession: I once applied for the role of a Quantitative Marketing Manager at Google. I got no response. After making enquiries, I got the impression that they wanted someone with a ‘statistics PhD or masters degree’. While I understand their insistence on a degree, they must understand that a practitioner who’s answered business questions from clients has a far better chance of making a serious business impact. You don’t always need a complex technical model. Yes statisticans are more technically skilled, but you don’t always need one. From further informal enquiries I know that the job never got filled for an entire year. Rant over.
I’d like to repeat that I’m not knocking statistics, I’m just saying that this is bigger. Business thinkers, database folks, visualizers and artists, all will play an important part in parsing all this data that does and will abound.
I got around to visiting Web Wednesday in Singapore, by the river, a couple of days ago (staying out late on a school night!)
It was a fun evening – the talk was short and simple, and not immensely educational or controversial, but the mingling was fun – I met all sorts of interesting folks.
I look forward to the next one too. Sign up here if you’re interested.
There’s a lot going on in the world of Mobile Analytics, and I wish I could post a longer, more detailed, point-of-view on it.
But since I’m busy, here’s a set of links to kick start your reading.
The Truth about Mobile Analytics – white paper and presentation.
I got an invite from Percentmobile, which I’m trying out with one of my pro-bono clients.
From the Omniture Blog, here’s something on how to measure mobile user action actions in Site Catalyst.
Here’s a quick look at two of the tools to check websites for analytics code implementation.
The Web Analytics Solution Provider was built by Immeria and is probably the best tool out there for auditing a website’s analytics code implementation. It provides a whole bunch of solutions, and I’m told that the purchased version is quite powerful – I presently use the Firefox plugin, and it’s most satisfactory. The only drawback is if you’re a non-Firefox user.
It’s a useful free tool to scan an entire website for tracking code implementation – which makes it a very useful starting point for all kinds of exercises. I’ve commisioned a sample scan for one of the non-profit websites which I help pro-bono, and expect a report soon. I’ll post again if there’s anything awry with the report.
A very simple plugin that I use on my Firefox which just simply tells you what cookies are being used – it’s not exactly a web analytics tool, but is a useful tool to keep an eye on how websites are tracking you – something one needs to continously do, as a professional.