Update 1/22/13: A lot has changed with Facebook and Google in the last couple of years. Please note that much of this post is now out of date.
According to a May 2011 Comscore report, Facebook now accounts for 1 out of every 3 display ads served in the US. This percentage is up from 23 percent in the third quarter of 2010. If you haven’t tried Facebook for patient recruitment, it’s time to start.
Facebook has some very powerful targeting features that can be used for patient recruitment, but for the newcomer these options can be overwhelming. So I’ve put together a 2-part series to help you effectively target your patient recruitment campaign. You can skip to part 2 of the series here:
Facebook Targeting for Patient Recruitment (Part 2)
When launching a campaign on Facebook, your ad clickthrough rate (CTR) is very important. Facebook’s algorithm looks at CTR to determine how many impressions your clinical research ad will receive. If your CTR drops below a certain threshold, Facebook will give your recruitment ad very little (if any) impressions. Of course, you can’t recruit patients if no one is seeing your ad.
Several elements contribute to CTR on Facebook, including ad copy/design and “freshness.” Another important element is targeting, which is the focus of this post. If you are already experienced with Facebook targeting, stay tuned. I’ll discuss more advanced tactics next week in Part 2.
Research the Patient Population
For the purposes of this tutorial, I’ll be advertising for a hypothetical type 2 diabetes study in New Orleans. Even if you already know a lot about a patient population, it helps to look up reference information. Reading up on a condition will often prompt ideas for targeting. I find that a good place to start is Medscape, though you will want to check other resources as well.
The Medscape type 2 diabetes reference notes several risk factors.
- Age greater than 45 years (though, as noted above, type 2 diabetes mellitus is occurring with increasing frequency in young individuals)
- Weight greater than 120% of desirable body weight
- Family history of type 2 diabetes in a first-degree relative (eg, parent or sibling)
- Hispanic, Native American, African American, Asian American, or Pacific Islander descent
- History of previous impaired glucose tolerance (IGT) or impaired fasting glucose (IFG)
- Hypertension (>140/90 mm Hg) or dyslipidemia (high-density lipoprotein [HDL] cholesterol level < 40 mg/dL or triglyceride level >150 mg/dL)
- History of gestational diabetes mellitus or of delivering a baby with a birth weight of >9 lb
- Polycystic ovarian syndrome (which results in insulin resistance)
This information will give us a good foundation for targeting patients who might be a good match for our hypothetical study. Now that we have gathered information about our patient population, we can look at Facebook’s targeting options.
Please note that since this is a hypothetical study, we do not have inclusion/exclusion constraints. In practice, you will want to consider how your inclusion/exclusion effects your target audience, and use the targeting options as needed. If your study has an upper age limit, for example, adjust your targeting appropriately.
At the risk of sounding like a realtor cliché, patient recruitment is all about location. Patients’ proximity to the research site will greatly influence their decision to participate. It is essential that you take advantage of the geotargeting feature available in Facebook ads.
After choosing the appropriate geographic area, check the “include cities within __ miles” box. You will generally select the “10” or “25” miles option. The appropriate choice will depend on several factors, including the geographic region, therapeutic indication, mobility of your patients, etc. In most instances, I would suggest starting your campaign within a 10 mile radius. You can always try expanding your reach later.
Next Facebook provides you with some basic demographic targeting options, including age and sex. Our Medscape article noted that age greater than 45 years is a risk factor for type 2 diabetes, so I’ve set 45 as the minimum age for my target audience.
You don’t need to check the “require exact age match” box unless your inclusion/exclusion criteria prohibits patients outside of a particular age range.
Facebook’s default interest targeting uses broad categories, but I generally don’t recommend you use these. Of course, there are exceptions. For example, broad targeting categories allow you to target parents, which can be of use for pediatric trials.
Click on the “Switch to Precise Interest Targeting” link. This option allows you to select people with very specific interests, as indicated in their profile. Facebook will make recommendations as you begin typing. In this case, I start typing “diabetes” and find that there are a variety of diabetes-related options.
We only want diabetes interests that likely relate to diabetics (or those close to them). For example, one of the choices available is “Diabetes Dude.” This interest is a reference to Wilford Brimley, who appeared in numerous diabetes ads for Liberty Medical. Most fans of “Diabetes Dude” will not be diabetics, so we don’t want to include this interest in our targeting.
With the targeting we currently have, Facebook estimates our ad’s reach at 360 people. Lets say we want to expand our reach. Certainly, the majority of diabetics do not note their condition on their Facebook profile. But they might note other interests that are related to diabetes in some way. We could also add diabetic products to our targeting.
Risk factors can provide additional targeting ideas as well. Since we know that obesity, hypertension, and dyslipidemia are risk factors for diabetes, we could also brainstorm for interests that relate to these conditions. For example, one area to target would be weight loss interests. Be creative! In Part 2, I’ll go over some techniques that are helpful in determining additional interests of your target audience.
Advanced Targeting Options
Facebook’s advance targeting options allow you to target by sexual orientation, relationship status, languages, educational levels, and workplace. Most of these won’t come into play until we discuss more advanced targeting in Part 2.
But for now, make sure that you use the language option to eliminate patients that your site is not prepared to consent. If you don’t have processes in place to consent non-English speakers, then you should only show ads to people that speak English.
Precision & Reach: A Delicate Balance
As mentioned previously, Facebook estimates our reach at 360 people. We’ve targeted our audience quite precisely, which will probably garner us a good CTR. However, this precision also means that our reach is very limited and we are missing a lot of diabetics. To extend our reach, we could try adding additional interests or eliminate our interest targeting all together.
The proper balance for your study will depend on a variety of factors, including your inclusion/exclusion criteria, geographic region, therapeutic indication, demographics of your patient population, and more. For conditions that are strongly correlated to particular demographics, demographic targeting alone may work well. As an example, a site conducting a diabetic trial in New Orleans might have more success with broader targeting than one in Denver. The reason is that disease incidence is high in New Orleans and low in Denver.
Determining the right balance will require planning, experimentation, and most importantly, tracking. Try different targeting options and always track your results. If you don’t have much experience recruiting patients on Facebook, it might be a good idea to start with precise targeting and then experiment with extending your reach.
With a good Facebook targeting strategy, your patient recruitment campaign will have a much higher chance of success. Now that you have the basics down, check out part 2 of this patient recruitment series.
[…] 6, 2011 In part 1 of this patient recruitment series, I went over introductory targeting techniques for research professionals new to Facebook […]