OPINION: Better digital measurement

We used to assume that the measurement of digital marketing would be easy - what, with all that data at our disposal. But datum can obfuscate, as much as it can elucidate - especially when it’s being produced at a rate it is humanly impossible to digest, across an ever-proliferating volume of channels.

So, how to measure digital better? Here are a few ideas.

Understand that last click attribution is flawed. This should be obvious but the number of marketers that continue to lean on last click as the way of measuring digital confounds the obvious.

Last click attribution measurement methods will always overstate the ROI of digital, to the detriment of key offline media. There may be a consensus that more marketing budget needs to be spent in digital channels (especially amongst certain types of advertisers – FMCG, anyone?) but relying on the budget allocation implied by last click is an erroneous and round-about way of justifying digital budget.

Marketing mix modelling (MMM) is not perfect (and its imperfections can be saved for another article) but at least it attempts to look at all marketing channels, digital included, as part of a complete set, where the interaction between channels can be measured. If TV advertising is driving a lot of traffic online, and so benefitting digital’s perceived ROI, this will be recognised by MMM, but not by last click.

Don’t let data define marketing’s objectives. Digital is a reckless producer of data but don’t be seduced by its wiles. Humans seem to have some hard-wired belief that a number is much more credible than a notion unsupported by a number.

Digital is replete with numbers, from which a whole slew of metrics can be devised. In our wilful obsession to objectify, however, we frequently let the metrics overwhelm the marketing, and start measuring stuff without due consideration of its relationship to marketing’s objectives. Don’t spend time analysing data and building metrics just because we can. When it comes to digital marketing, ask plainly, How does what I’m measuring link to business goals? What insight reward can I expect for my effort?

Don’t forget the 80:20 rule. Digital marketing measurement is 80 per cent about people and processes and 20 per cent about tools and technology. If you have just sold the crown jewels to pay for a web analytics platform but then failed to support that platform with adequately equipped people and processes, then do your best to get your jewels back.

Remember when we used to watch TV? So often digital’s exponents are siloed from the rest of a business like some rare and exotic bird in the menagerie. This is true for both agencies and clients. Digital is just another channel or touch point, so expose its esoteric delights to the trials, rigour and inspection of anything else marketing, and the business, puts on the table. By opening up the process of digital measurement to the masses, flaws can be identified, more useful metrics determined, and a more digitally-savvy personnel nurtured.

Measure it better. Technology marches forward and occasionally this benefits marketing. It has been reckoned that between 50 per cent and 80 per cent of clicks in China are fraudulent, generated by either students after a quick buck or machines (imagine inventing a click generating machine?).

On the back of this, some nifty work has been done to identify heavy handed clickers, which has resulted in a raft of other applications for digital analysis. Traditionally, a complaint of adservers was their data-dump approach to analytics. These guys in China have turned their technology into a service of active campaign optimisation. Not only do they analyse the deeper data from their superior cookies for you, they also have an account team that manages the performance of your digital activity, weekly. Win-win.

Appreciate that this is an evolution. There is no so-called silver bullet for digital measurement - last click is patently wrong, MMM has its flaws and quality scoring for social is uneven. But don’t let these manifold imperfections dissuade you from that first measurement step, or from considering the higher forms of digital measurement. Considered digital analysis, that is internally robust and consistent, and whose outputs are viewed through a filter of reasonableness, are better than no analysis at all.

Don’t be overwhelmed by digital and its data – its Achilles heal can also be its arrow.