Personalizing Messaging Services to Support Smoking Cessation

read time - icon

0 min read

Every day, 100 Canadians die of smoking-related illnesses. The old-fashioned habit is one of the deadliest in the country, accounting for more deaths than obesity or high blood pressure. In fact, up to half of all smokers will die of smoking-related illnesses — and on average, they’ll die 10 years earlier than non-smokers. 

But smokers are never a lost cause. Over time, smoking cessation can reverse the harmful effects of cigarettes. Just one year after quitting cigarettes, studies find that people are already at a 50% lower risk of having a heart attack. 

The problem? Getting smokers to quit. Smoking cessation is a notorious behavioral problem, with researchers estimating that it might take the average smoker more than 30 attempts to quit before they actually succeed. The stickiness of this habit makes it an ideal target for evidence-based behavior change interventions. 

The Canadian Cancer Society (CCS) is the largest national cancer charity and the biggest cancer research funder in Canada. They approached The Decision Lab for help designing an SMS-based service to support Canadians as they try to quit smoking. We channeled the latest research on health behavior change into a program that empowers smokers to take back control of their own health habits.

Applying a behavioral lens

TDL started by uncovering the key barriers and drivers of efficient, effective digital smoking cessation programs. Our research team analyzed the structure and impact of 15 existing smoking cessation services, conducted a specialized literature review into established and emerging strategies, and surveyed 122 smokers and former smokers about their needs and preferences around digital smoking cessation aids.

We leveraged the COM-B framework of behavior change as a lens to guide our research. COM-B is a widely used framework to help behavioral practitioners design effective interventions, by first characterizing behaviors in terms of Capability, Opportunity, and Motivation and then identifying factors that may present roadblocks in these areas. 

Breaking things down with machine learning

Our external user research informed our team on overall attitudes towards smoking cessation in Québéc, the target of CCS’s program (and our own home province!). Our structured interviews were able to bring to light the invisible barriers and considerations of our target user base.

But smokers aren’t a monolith. It goes without saying that there’s a lot more to people who smoke than their cigarette habits: they have unique personalities, live in different circumstances, have different social networks, and so on. 

All of these things impact an individual’s relationship to smoking, and how their quitting attempts will play out. That being the case, any effective smoking cessation program has to incorporate an element of personalization: understanding what barriers are holding a particular individual back, and delivering targeted, timely interventions that provide the specific supports they need.

We used machine learning algorithms to facilitate this personalization. While it might not be possible to account for every single thing that makes an individual unique, what we can do is identify broad patterns across the entire population of smokers, and then develop personalized interventions based on their distinct characteristics. 

In the end, we found 4 distinct personas, each with different barriers and drivers:

High Smoker Identification: Those who see smoking as a key aspect of their identity. For this group, the biggest barrier to quitting is the fear of losing the company of other smokers.

Low Risk Perception: Those who believe smoking isn’t as risky as other risky activities. Although many in this group lack the social support to quit, the majority would attempt quitting if a close friend did the same. 

Low Risk Tolerance/High Social Support: Those who understand the risks of smoking and have cessation support from their loved ones. This group usually has high self-efficacy (i.e. they are confident in their ability to quit if they wanted to). They’re enthused about cessation if connected to others who are attempting to quit. Their biggest barrier is low motivation, due to their low perception. 

Low Risk Tolerance/Low Social Support: Those who understand the risks of smoking and don’t have cessation support from their loved ones. This group struggles the most with motivation. They don’t believe cessation services are accessible and don’t trust their providers. These users are best supported by external advertising, in order to build the trust necessary to join a cessation program. 

Conversation trees and behavioral design

With our 4 key personas identified, we set about building conversation trees that would make up the content of the SMS program. 

As with any behavioral health intervention, there’s no-one-size-fits-all approach. Every user will have their own goals and requirements to succeed. To reflect the diverse needs of Canadians trying to quit cigarettes, each communication flow included a timing indicator, based on the preference of the user, to target key behavioral change periods. The content also varied, based on the categories of barriers and drivers, including peer nudging, motivation, self-affirmation, distraction, risk awareness, and rewards. 

We developed an SMS conversational flow based on our identified population clusters, and found the most impactful interventions for each persona based on their identified characteristics and needs. For example, those with Low Risk Tolerance and High Social Support are in need of higher internal motivation. This population responds most positively to tying their cessation plan to personal goals, like being healthier or saving money. Users can receive regular updates on how much money they’ve saved by not purchasing cigarettes, or how much their risk of heart attack has lowered. On the other hand, users who lack social support can participate in leaderboards, including the ability to send and receive kudos from other users. 

We also made sure to stay up to date with users as they progressed along their journey. Frequent reminders may be helpful at the beginning of smoking cessation, but can distract users from their success later on. The frequency of messaging in our design is amplified in the beginning to make the program more salient, but gradually decreases content delivery as the user approaches the final phase, when frequent reminders about smoking become more harm than help. 

One day at a time

Smoking cessation is often framed purely in terms of willpower — either you have the presence of mind to white-knuckle your way through quitting, or you don’t. 

Not only does this approach obscure the many psychosocial factors that influence smokers’ attempts to quit, it invites shame into the equation: shame as a strategy to pressure our loved ones to lose their smoking habit, and shame as emotional response among smokers who have tried and failed to quit. Research shows that all of that is counterproductive.

In our work with the CCS, we built a program that’s actually attuned to people’s individual needs. Smoking cessation is still hard — we can’t change that. But we can meet people where they’re at, and empower them to understand the real drivers of the behaviors they’re seeking to change. For the millions of smokers in Canada who are trying to quit, our work will be a valuable source of support.

Read Next

father and daughter eating french fries together
Case Study

Burgers With Balance

Promoting childhood nutrition in partnership with one of America’s largest restaurant chains

a boy writing on a notebook
Case Study

By the Books

Closing gaps in student achievement through evidence-based curriculum selection

Notes illustration

Eager to learn about how behavioral science can help your organization?