Liam Brennan (@LCBrennan), Global Director of Innovation, MediaCom is one of our industry’s preeminent thinkers in innovation and NDA’s monthly columnist.
The average New Zealander owns approximately six sheep. But you’d be incredibly lucky to find a sheep, let alone six of them, if you wandered into a typical suburban house in Auckland.
Although New Zealand has one of the highest numbers of sheep per capita (recently knocked off the top spot by Mongolia), Kiwi sheep are mostly owned by a small number of shepherds, and as a result, the most common number of sheep owned by a New Zealand household is zero.
This scenario (and logic flow) seems fairly obvious, but it illustrates an important point about reading deeper into data, and never taking a simple aggregate number at face value.
“Being able to understand the mean (six), median (zero) and the mode (also zero, but the typical Kiwi shepherd owns around seven thousand sheep, so an extreme outlier) helps tell a much clearer story of what’s going on i.e. New Zealand has a quite a lot of sheep for a small country, but sheep distribution is concentrated around a small number of shepherds and not evenly across the 4m population.”
As an industry we have an odd habit of heroing new, and often more complex, ways of working, particularly when it comes to ‘left brain’ activities (e.g. coding, data science etc.). And whilst many of these fields can and do add value, they shouldn’t be a substitute for simple entry level analyses like the example above.
With the increase in data availability and the subsequent time required for cleaning, manipulation, showcase and application, we’re often likely to leave it to the experts to do this analysis, view through a templated dashboard, and not invest time in asking simple questions of the insight, or indeed the parameters of how that data was collected in the first place.
This is particularly pertinent as we (hopefully) soon venture out of full COVID restrictions and start to make future decisions based on 2020 data norms that are far from normal.
For example, early in the lockdown, as chart was widely shared chart on social media showing an almost ‘hockey stick’ like growth of ecommerce as a proportion of overall US commerce activity, gaining ’10 years growth in under three months’. But what exactly was driving that growth?
More detailed US Census Retail sales data was released last month (August 2020), and as expected, it indeed did show that ecommerce’s overall share of non-grocery retail grew from 17 to 22% from Q1 to Q2 2020, an incredible 32% proportionate growth, but against a market that had contracted 4%.
Ecommerce grew due to a shift in activity from closed stores to online channels (the nominator), but also as a proportion of commerce due to the decline of overall market (the denominator). When recalculating that growth number, it’s still an impressive 28% quarterly ‘jump’, but also shows a roughly $30bn revenue loss where ecommerce didn’t fit consumer needs, or simply wasn’t available.
Observing aggregated or data points without context, such as the scenarios above, can make for great PR releases, fodder for retweets or can hide bad news in a shareholder report, but they’re problematic when trying to steer a business along a direction of travel.
If one was to look at that ‘32% growth’ number in isolation, your outcome would likely be an urgent need to pivot to ecommerce. But the wider story tells one of substitution rather than gain, particularly when looking at markets who went through the pandemic earlier like the UK.
Recent ONS data shows that that US number will likely shrink in Q3. In the UK, non-grocery ecommerce doubled from 30% to almost 60% from February to April, partly due to overall contraction of the market by a whopping 35% (again, the denominator). In July, that ecommerce number sits at a much more ‘normal’ 42% (and falling) in an overall market much closer to pre-COVID levels.
A laissez faire attitude to data analysis can be dangerous, missing vital ‘clues’ to what is actually going on, drawing incorrect conclusions, and as a result, making flawed recommendations with real business impact.
No one should be expected to have a deep understanding of the New Zealand sheep industry, but we should always know to apply high school level statistics and analysis in our everyday roles, interrogating what the numbers mean and putting them in wider context.
For if we don’t, we’re giving someone an opportunity to pull the wool over our eyes.
