The Beginning of Infinity: David Deutch
The E.T.-detecting cork is a particularly satisfying example – distilling measurements taken at the sharp edge of human ingenuity into a single ‘pop’…
But of course we use proxy measurements all the time – from the volume of applause in an auditorium, to the rumbling of your stomach, to your scores in a test (provided you did well in that test, of course…)
Identifying these kinds of correlations, and chains of connected events, can help us make better, faster, and easier measurements (ice cream sales as a proxy for temperature?) It’s a classic way to work smarter rather than harder.
So let’s see what useful insights we can draw by looking at the hidden chains of events behind some of our Google Ads metrics…
1) Time to conversion as proxy for funnel position
We often categorise segments of our marketing activity by funnel orientation.
This brand+product campaign is oriented towards users with a high intent (as demonstrated by the nature of their search) and we’ll treat its targeting / bids / ROAS expectations accordingly…
But how often do we test those assumptions? And how would we determine at what stage of the funnel a given campaign is actually operating?
One answer is, of course, in the sub heading above.
Time to conversion is a reasonable proxy for funnel-stage, with a shorter time-to-conversion indicating that your ads are targeting users who are close to converting, while a longer time-to-conversion might suggest that your ads are targeting users nearer the top of the funnel, still requiring further touchpoints.
Noting the time to conversion (funnel stage) of any given segment of activity allows you to tailor your activity (ad text / landing page content / targeting adjustments) to these users more effectively, based on some real – albeit indirect – evidence of where those users sit in the buying journey.
2) Bounce rate as a proxy for ‘expectation fulfilment’
Bouncers are users who don’t stick around – so you might reasonably take Bounce Rate as an indicator of landing page quality…
But more specifically, it is a marker of quality in relation to the expectations (or hopes) of the particular set of users landing on that page.
That gives you two lines of enquiry once you’ve identified a high bounce rate: Does the page need changing, or the users? (both are testable in multiple ways).
Any use of Bounce Rate as a success metric comes with the caveat that single-page sessions are not necessarily unsuccessful (but, frankly, they usually are… and GA4 does a better job than UA of addressing these exceptions by cancelling the bounce after ten seconds of dwell time by default).
So Bounce Rate is a useful indication of the ‘right traffic to the right page’ and – as wide and hazy as that target area is, it is nonetheless a crucial one in any marketing endeavour.
Bounce Rate becomes extra valuable in this role when you need to fill the data gap left by a lack of conversions.
As one of the potential inputs for audience creation (and with audiences based on ‘home data’ becoming increasingly important) – Bounce Rate is also a useful variable to play with in narrowing down what you want those audiences to represent.
3) % New Sessions as a proxy for audience growth
What can we tell from the % New Sessions metric as a metric in Analytics?
Context (and caution) are crucial here, since a low % of new users may mean either success in engaging and retaining return visitors, or failure to attract new ones.
So the metric should always be considered in relation to what job you want each campaign to do.
If you want to grow your audience, % new users can give you one very useful indication of how efficient a job your campaigns are doing in fulfilling that aim.
Many of our standard metrics could be described as proxies of a kind, and it’s worth stepping back occasionally to remember the human stories that lie behind the numbers.
After all, an impression is an experience…
A click is a decision…
A conversion is a leap of faith…
And this user-centric perspective isn’t just whimsy. If you want to move the numbers in the right direction, at some point you have to understand what moves people.
But while we are in metrics mode, it’s also worth expanding our analysis palette where we can, by drawing new paths through data points to arrive at new insights.
And hopefully the examples above show that there are new, meaningful connections to draw between our metrics and our objectives, each one building a little more explanatory and predictive power onto our analysis…
Ultimately perhaps even a few more cork-popping moments.
