Data Data Everywhere, and Not a Bit of Meaning
Not long after the first commercial sites appeared on the Internet, long before DoubleClick and Mediaplex and 1,000 other 3rd party ad serving solutions came along, advertisers clamoured for a form of measurement that quantified the value of what they were buying.
And web site publishers were all to happy to oblige, armed with log files and reams of data about pageviews, impressions, visitors and clicks…. more data than any offline publisher had ever offered, in more excruciating detail than anyone could have imagined, or wanted.
It all seemed terribly impressive at the time, to know down to the individual viewer how many times a page was seen, or how many times a link was clicked on. And a lot of advertisers felt a sense of comfort with the exactness of the measurement… it all seemed so reliable.
Except that there were always exceptions, and difficult ones to explain in layman’s terms. (This is not to say that advertisers weren’t, or aren’t, smart enough to understand the details… it’s just that, as with most issues relating to technology, they have better things to do than try to tease out the differences between a visit and a visitor, the vagaries of using web bugs vs. javascript to gather usage data, 30 gigabyte logfiles and so on. Marketers never worried too much about cathode ray tubes or circuit boards either, and that didn’t stop anyone from creating effective advertising.)
So it’s useful, I think, in light of the fact that more and more marketers are questioning the utility of the click, to talk about why the click came to be used in the first place. It’s really quite simple:
It was easy to measure.
Since the invention of hyperlinks, the click has been the Internet’s basic unit of interaction. Literally every single piece of measurable data available online today originates from a click. Because it is and always was so essential to the basic functioning of the Internet, keeping a log and then reporting on click activity later evolved quite naturally.
(I’m skipping over an important development here: initially, most banner ads were sold on a CPM basis… this was the model publishers were used to offline, and pageviews seemed equivalent enough to “readers” or “viewers” in print and TV. But this model was dying before it started…. the Internet had no scarcity of bandwidth similar to offline media, and publishers could just create more inventory at will. Impression-based buying has enjoyed a resurgence of late, as more big-brand marketers have shift budgets online… but this is actually impression-based buying of a sort that is fundamentally different than that which first developed, and which was killed off by Google’s popularization of click-based purchasing.)
Now, barely a decade and a half into the commerical web, we have oceans of data derived from that simple action. Clicks-to-conversion, clickthrough rate, and clickpaths. But as the data grows richer, meaning has been harder to extract.
In the end, for most marketers, what matters is what sells. And the link between online exposure and offline activity is hard to measure. Couponing and loyalty programs can help, but they only hit a segment of the audience. Reconciling online ad exposure with shopping behavior can be done - on a sampled basis with tools like Consumer Direct - but typically only with large budgets, and in specific industries.
And if you have enough online and offline data and a decent enough geographic footprint to have measureable differences in your marketing mix by region, you can analyze differences in your spend by channel and see large-scale differences in effectiveness.
But for businesses that transact primarily offline (that is to say, most businesses), a more satisfying (and intelligible) measurement of advertising effectiveness is elusive. The best units of measure are the ones you control yourself - measures of engagement that happen on your own web properties, or ad units that enable and encourage interaction. Use the click as a basic building block of your measurement platform - as a starting point.
2 years ago • Notes