Empowered with this kind of approach, brands can see results such as an increase in performance, decrease in cost, and even a reduction in CO2 emissions.
Statistical Process Control (or SPC) is a phrase you may have never heard before; we’re almost certain you’ve never heard it come up in the context of digital marketing.
Over recent weeks and months, the team here at boa has started adopting the approach when discussing our business challenges – whether those of our clients or those of our agency.
We believe that SPC has a lot to offer to digital marketers, and the technique applies directly to what we do here at boa every day: running, reporting on and improving the performance of (primarily) online advertising campaigns.
In this blog post, we’re pulling back the curtain on boa’s novel approach to campaign monitoring and execution and how we leverage a statistical approach to help clients see an increase in ROI, reduction in costs and decrease in campaign carbon emissions.
Defined broadly, Statistical Process Control is the use of statistical techniques to control a process or production method. These techniques were largely developed and applied over the past century in the context of industrial manufacturing to solve what one of its pioneering thinkers described as “the problems of economic production.”
In short: SPC exists to improve efficiencies, reduce waste and increase profits.
The aim is to move towards prevention-based quality control instead of detection-based quality control – e.g., for manufacturers, it’s the ability to predict the behavior of a production process.
When you put it like that, maybe SPC and paid advertising have more in common than you think. It’s just not something that a lot of agencies (except yours truly) leverage.
PPC (short for “pay-per-click”) refers to online marketing; it’s a big part of what we do here at boa. While most forms of digital advertising don’t operate on a pay-per-click model, to many people PPC and online ads are synonymous.
Advertisers use PPC to identify and purchase ad space on the most effective and relevant channels, at the optimal time, for the least amount of money. PPC advertising looks different from platform to platform, but in general (if the campaigns are working as they should) the return on ad spend should be high.
In other words, PPC campaign management is a classic example of a “challenge of economic production.” So, while they may seem unconnected at first, the goals of PPC and SPC are very similar.
When you’re making decisions with data, it’s incredibly important to understand what your data is telling you, rather than what you think it’s telling you.
A shoe manufacturer doesn’t want to produce a flawed shoe and have to guess where in the production line it all went wrong. They need to know how, when, and where along with the ability to go in and fix their product. That way, they can meet consumer quality standards.
The same can be said of paid campaigns. When we know exactly what our data is telling us, we have the complete story and can make smart decisions as the campaign is unfolding.
At boa, we’ve begun to incorporate SPC methodologies to help us continuously monitor and optimize our campaigns. We strongly believe in the potential for these methods to achieve:
What makes SPC methods so powerful in manufacturing is the continuous collection of product and process data, rendered into intelligible feedback from that process, so operations managers can quickly identify, fix and/or prevent defects before production is adversely affected.
But in any production process, there is variation that is both expected and acceptable.
One of the essential tools that SPC methods commonly rely upon is a control chart. This tool surfaces key performance details about a process in order to enable both a meaningful understanding of the performance of a process over time and its continuous improvement.
Control charts include an acceptable range of variation of the data – formally, the upper and lower control limits – within which the process can be expected to perform, provided that no changes are made. Any data points falling outside of this range should be investigated, but are not in and of themselves a reason to change the process.
In other words, because some variation is expected and because this degree of variation can be known, managers are less likely to overreact to any one signal or data point. Over time, the result is a more consistent, efficient process.
Digital advertising campaign management involves large amounts of data, collected from various sources and by varying methods. That means it isn’t always obvious what the most impactful choices are, which makes it difficult to know what the most impactful action would be – or indeed, if action should be taken at all.
In these instances, PPC ad campaigns can often devolve into a lot of trial and error, with tangible costs along the way. Without the right expertise – and an appreciation of variation – campaign managers can over-optimize their ad campaigns, making constant edits to account for what they perceive as unusual changes in performance. This often leads to the adverse effect of pushing their campaigns further off-course.
By applying SPC methods, we gain greater confidence in our understanding of campaign performance, and we can anticipate the degree to which the numbers can be expected to “wiggle.” This in turn enables us to better predict future outcomes, by identifying meaningful trends and patterns in campaign results over time.
Empowered with this kind of approach, brands can see results such as a decrease in cost, increase in ROI, more qualified leads and even a reduction in CO2 emissions.
From a pure implementation perspective, running a PPC campaign has perhaps never been easier than it is today. The digital advertising ecosystem has become increasingly standardized, and the biggest ad-buying platforms have worked to become more accessible for advertisers of all sizes.
But as easy as it can feel, PPC campaign execution is still a complex technical process with a lot going on behind the scenes. And the issue with any complex process is that users can face an overwhelming amount of data and many possible optimizations to consider – even when sometimes, the best course of action isn’t doing anything at all.
They say that paid advertising is both art and science. We think it’s actually math, and more specifically, statistics. Or at least, our approach to it is.