Business decision-makers often face questions of causality: How much will sales increase if I increase my advertising budget? The answer matters when companies need to redirect budgets to keep the business afloat.
Recent research by AC Nielsen shows that advertising spending is down by 20% and it is likely that it will continue to fall. As the lockdown kept people in their homes the out-of-home category of advertising became redundant. Yet, screen time has increased, and even though people are spending less, they are still shopping and your brand should not self-isolate.
The Financial Mail reports on the findings of an Opinium study that customers want:
- simple business updates as regulations keep evolving,
- advice on dealing with the crisis, and products and services that may be useful,
- stories of innovation and inspiring ideas, and
- how companies are making a difference to the lives of the broader community, including their employees,
People want connection, and branding is about connecting with people and building trust.
But knowing that Facebook has the most users, Instagram has been growing, or that TikTok is a growing social media platform in South Africa, does not solve the causality problem. You might end up paying a brand ambassador, but never knowing if he or she is making the difference. Of course, there are some cool software platforms that help you track and manage the work of “influencers”, but there are real benefits to randomised control trial (RCT) experiments.
Economists have been using RCTs for a while in micro-development policy: do giving laptops to kids, improve learning; should aid agencies give mosquito nets away, or sell them at a very low price; should we give food parcels or cash? The key is to figure out what the baseline, or “control”, is and to then carefully add a “treatment” and measure the effect.
For a business that wants to try out a new advertising campaign or a promotion, the key is to design the rollout in a way that you can learn something. For example, advertise or change prices in randomly selected areas and not in others, and try to measure the difference.
Setting up a good experiment is hard, but can be very profitable.
Read more about it here: