Digital advertisements appear to be more effective than their actual performance due to the notion that ads on the internet boost click-throughs. People who see ads will purchase the product even when they don’t see an advertisement. It’s possible to cut your advertising budget while earning more.
Digital advertising is a wildly underrated. A comprehensive analysis of advertisements on eBay discovered that brand search advertising effectiveness is underestimated as high as 4,100 percent. A similar analysis of Facebook ads showed a figure of 4,090 percent. It is apparent that companies do not know how to answer the questions asked by John Wanamaker , a 19th-century retailer: What percentage of the advertising budget of my company was spent?
This question needs to be addressed. It’s not a lack of data that caused the issue Wanamaker ran into, but the fundamental misinterpretation of causality and correlation.
The Conversion Fallacy
Marketing reps frequently sell ad space to customers and say that advertisements may induce or trigger behavior changes. This is often called lift. It is believed that the conversion rates are the standard method used to support their claims.
I ask my students to picture me in front of the door and handing out flyers advertising the course the day they start. Then, I question them “What is the conversion rate of my advertisements?” Since 100% of the people who saw my advertisements “bought” the course or enrolled in it. I always correct them. I I ask: “How much did those advertisements affect your behaviour?” They all respond, “Not at all,” as they were already enrolled in the course prior to the time they saw the ad. Thus, while the conversion rate is 100 percent, the amount of lift it generates and the amount of change in behavior that it triggers is completely negligible.
The example I have chosen, though it’s not complex, shows the way that confusion about the effects of lift and conversion could cause issues when evaluating the ROI of marketing. The biggest companies pay a lot of money to consultants to direct their advertisements to those who are most likely to purchase their goods. The conversion process from click-to-cash isn’t profitable if it isn’t targeted at people who don’t have the requirements to purchase the items. Advertising is all about getting customers to buy your product (or give money to a cause or get a vaccination) which they wouldn’t otherwise be able to.
Measuring lift
Let’s suppose that we wish to determine whether (A) an individual’s life earnings are lower due to the fact that it was when they joined military (B). It’s not possible to compare salaries of people who are in the military against those who don’t. There are many variables (C) which could result in variations that we cannot observe in the raw figures.
Jobs that pay better such as those with higher salaries are less likely to be drafted into the military. This is known as B-caused A. The people with higher education and abilities are less likely to be a part of the army. (C is the cause of both A and). What may appear to have a causal link between lower wages and military service could be an indirect consequence of other influences. It is essential to consider other variables and still discover the most appropriate connection to investigate.
It is possible to do this by forming a team called a control. Randomly assigning people to military can result in the group that is treated (or those in the control group) sharing the same educational background and skills and also their gender, age and temperament. If there are enough records to be able to assess the distributions of all visible and unobservable traits among those who were placed in the control and treatment groups. This will help us understand why results differed among the groups. We are able to say with certainty this is because their experience in military is what could influence their earnings.
This is where the issue lies. Scientists would have a tough to justify a research which randomly enlisted individuals in the military. This is known as “natural experiments” and are sources of random variation that replicates an experiment that is random.
Josh Angrist used a good natural experiment to determine the impact of military service on pay. It is the draw lottery which was forced on U.S residents in the Vietnam War. Every male citizen was issued an draft lottery number. The numbers were randomly selected to determine who was selected for the draft. The draft lottery was a natural experiment that created a random variations in the probability of being drafted into the military. The variation was utilized by Angrist to estimate the effect of causality on the salaries of military service.
Like my experiments with my weather experiment, Christos Nicholaides and me made it an ideal study to determine the impact of social media’s messages on exercising behavior. People who run more frequently have more people who run with them. But, the variation in weather has helped us understand the extent to which the social messages from friends led us to exercise more.
It is easy to see that ads online are more effective that you expected. Yahoo! For instance, Yahoo! however, the majority of the sales were made by repeat customers who clicked the advertisement. 93% of sales were made in brick and mortar stores , not on the internet. The conventional theory of causality in online ads that states that viewing causes a clicks and ultimately to purchase, doesn’t reveal how ads affect the actions of consumers.
Causal Marketing: The Benefits
These results could provide a reason for why Procter & Gamble (the granddaddies in marketing for brands) could improve their the performance of their digital marketing despite having their digital marketing budgets reduced. Marc Pritchard (P&G’s Chief Marketing Officer) reduced the company’s spending on digital ads by around $200 million which is 6%, in 2017. Unilever reduced its digital advertising budget by nearly 30 percent in 2018, which is more than in the year 2017. What would the outcome appear like? The outcome? The result? 7.5 percent increase in P&G’s organic sales growth for 2019 and an 3.8 percent increase for Unilever.
The two companies also altered their media budgets to a more focused focus on frequency, as measured by views and clicks instead, they focus on reach, or the amount of customers they interact with. Their prior data had shown that certain customers were repeatedly bombarded with ads on social media from them between ten and twenty times every month. The bombardment was causing a decrease in returns , and could even upset faithful customers. They reduced the number of ads by 10%, and transferred the advertising money to new customers who were not seeing advertisements.
To better understand the motives that drive purchase, they looked at first-time buyers in detail. This enabled them to identify potential customers. In their earnings call for the fourth quarter in 2019, the company revealed that they’d moved away from “generic demographics” like “women aged 18 to 49” to “smart market” including first-time mothers as well as washing machine owners.
John Wanamaker’s query could be answered with the flood of personal, detailed information that is generated online. These data can be utilized by marketers to determine the most effective messages and which don’t. Make sure you differentiate between causation and correlation like P&G, Unilever, and don’t target those who have already been loyal.