What John Snow can teach us about advertising

10/02 By Katia Eliseeva

Key learnings summary:

  • Relying on correlations can be dangerous.
  • Understanding the causal link between advertising and conversions is essential for running an effective advertising campaign.

Advertising and medicine may seem like completely unrelated disciplines, but below the surface, they share important similarities. Today’s brand marketer’s serve consumers ads because they believe that advertising will directly or indirectly lead to more sales, either through immediate purchases or longer-term brand awareness. Similarly, doctors give patients drugs because they believe that the drugs will alleviate or cure their diseases. For a doctor, the effectiveness of a particular drug is based on the series of clinical trials that it must pass in order to be approved for its intended use. The purpose of these trials is to establish a causal link between taking the drug and the improvement of symptoms. It is not enough to say that people who take a certain drug also tend to not have a disease because it is possible that some other factor is influencing both the drug taking and the disease.

Shouldn’t our belief that advertising works be based on similarly robust evidence? How does anyone know which campaigns are influencing sales?

Though the idea that diseases could be cured by treating the sick with medicine came from ancient Greece, randomized control trials and proofs of causality are much more recent phenomena. For example, take the outbreak of cholera in London in 1854, due to a contaminated water supply near Broad Street. To combat outbreaks of diseases such as cholera, it was common practice for residents of large cities like London to dig cesspits under their houses in order to reduce odors. Although this made everyone feel safer, it didn’t stop the spread of cholera, which is actually a water-borne disease. One such cesspit near the Broad Street pump began to leak and contaminated the water supply, leading to an outbreak of cholera in 1854.

John Snow (not the King in the North), was a British anesthesiologist who lived in London at the time. He proved – by mapping victims’ water supplies – that people who lived in the same neighborhood and obtained water from the Broad Street pump had a significantly higher incidence of cholera than those who did not. He used this data to convince local authorities to remove the pump’s handle. As a result, cholera cases plummeted and the epidemic was stymied. By studying the location of water supplies, John Snow had stumbled upon a “natural experiment”, which is one way to establish causality.

In advertising, to ensure that we don’t make mistakes that come from relying on associations, we need to understand the causal relationships between ads and sales. Serving an ad to someone, and a product then being purchased, does not necessarily mean that the ad caused the conversion. Perhaps the purchase would have still occurred even if the ad had not been served. Moreover, it is possible that the consumer had shown interest in the product earlier, visited the website, and was then re-targeted with the ad expressly because of this interest. In this scenario, the purchase actually caused the ad, not vice versa. If we knew which consumers would buy our products without targeted advertising, then we would only advertise to the consumers to whom it would make a difference. Similarly, if we knew that one type of ad would be more effective in convincing a consumer to convert, then we would always serve that ad. We need to have knowledge about these causal links in order to know the best way to advertise.

In general, knowledge about what actually causes diseases and how to identify those causes has led to people living longer, healthier lives. In advertising, we too can benefit from these methods. By running experiments and conducting careful observational studies we may learn that certain kinds of ads and exposure frequencies are much more effective at driving sales. On the flip side, we may also learn that advertising to some consumers doesn’t cause them to make purchases. We can then use this knowledge to design better, more effective advertising campaigns.