It’s an adage as old as the marketing discipline itself: right message, right person, right time. Easy, right? Well, kind of. As easy as that is to say, creating meaningful buyer personas is often an exercise fraught with areas of complication. One area I often consult around relates to leveraging generalization. A buyer persona that doesn’t generalize to the target audience is not just useless. It is harmful. Let’s talk about why.
The Linda Problem
Apologies to all the Lindas.
Linda is 31 years old, single, outspoken, and very bright. She majored in philosophy. As a student, she was deeply concerned with issues of discrimination and social justice, and also participated in anti-nuclear demonstrations.
Which is more probable?
- Linda is a bank teller.
- Linda is a bank teller and is active in the feminist movement.
If you’re like 80 percent of people, you chose 2. And you chose wrong. If you chose 1, congratulations.
People at all levels of statistical expertise make this error. It is statistically impossible for 2 to be the right answer. Bank tellers who are active in the feminist movement can only be a subset of all bank tellers.
Why People Get It Wrong
This is called the conjunction fallacy. Kahneman and Tversky explain why we do this in terms of representativeness. We can conjure up a very detailed image of number 2. Can’t you almost picture Linda in her beret? This salient imagery makes number 2 seem more likely in our minds. But again, statistically, it’s impossible for 2 to be more probable than 1.
Thinking probabilistically is hard, and for most of human history, we didn’t evolve to think hard. We evolved to think fast. Now that we have the luxury of thinking hard and we have computers to do the heavy lifting, we can avoid committing these errors.
So, as level of detail goes up, generalizability goes down. But, and this is the important bit, perceived generalizability goes up. As we become less accurate, we become more confident.
Applying “The Linda Problem” to Buyer Personas
In creating personas, we create a list of traits per persona. What do they care about? What are their goals? Do they have the means to achieve their goals? How do they measure their progress?
Like most human traits, each persona trait falls on a spectrum and there are typically more people in the middle of each spectrum than on the extreme ends. When we say that Persona X “is under pressure from the board of directors to expand the company’s global footprint”, the actual individuals we talk to are going to vary from a little pressure to a lot of pressure. For some, it may be their core pain point. For some, it is peripheral.
When we pick what the typical person looks like, in order to create a persona, we are narrowing the range of people who can fall into our persona.
Now, what happens when we add another variable?
We have again constricted the range of people who fit our description.
How about two more?
As the level of detail goes up, generalizability goes down.
What is the impact of this?
Overfit messaging to a specific person runs the risk of not resonating with the target audience. We want the message to resonate with the highest number of people while alienating (or boring) the lowest number of people in our target audience. We’ll never get perfect overlap, but we suggest striving for better.
Lack of Situational Awareness in Your Front Line
Second, and more importantly, as we move back toward the concept that people matter in sales and marketing, non-generalizable personas decrease situational awareness in your sales and marketing teams. We have very powerful biases, and personas that augment these biases make our sales and marketing efforts less efficient and effective.
In the 70s, researchers in bell bottoms did an interesting experiment, which uncovered one of the most fundamental errors humans tend to make. If you ask people to watch a quiz show and rate the general knowledge of the contestants compared to the general knowledge of the questioner, people rate the questioner as more knowledgeable. Dubbed the “Alex Trebek effect”, people discount the situation (the questioner literally has the answers on cards in his/her hand) and attribute the person’s demonstrated knowledge to dispositional factors about who the person is. They called it the Fundamental Attribution Error. It is one of the most powerful biases we have. Personas that paint a rich personality of a specific person encourage us to commit this error.
It goes even deeper, to the root of our biology. Our higher-order cognitive capacities are made possible by our huge cortexes, but they are rooted in our lizard brains. This means that abstract concepts are based on perceptual knowledge. Time flows from left to right, I wake “up”, etc. It’s all over the place in the language and how we think. That conceptual knowledge can literally override our ability to perceive the world.
In another interesting phenomenon called “Verbal Overshadowing”, when one describes something conceptually first it actually reduces the ability for people to recognize it perceptually later. For example, if an eyewitness is asked to describe a suspect, they are less likely to be able to pick them out of a line up!
As we abstract, we lose information that lets us treat people as individuals, and this actually limits how we can see/hear/experience people. It decreases our situational awareness to be able to respond appropriately to the signals a person sends.
The lesson? Any abstraction had better damn well be accurate most of the time.
How Do We Fix It?
We don’t just need to reduce harm. We also need to make personas effective. Here are three ways to accomplish both.
Build a Buyer Persona of Differentials
Does this piece of information help you to distinguish Buyer Persona X from Buyer Persona Y? Does this piece of information lead you to treat the personas differently? If yes, keep it! If no, throw it out! Every variable you add creates noise. Be sure that it also sufficiently boosts your signal enough to justify the noise.
This can often require using ranges instead of single values. You’re looking for traits and clusters of traits that create mostly non-overlapping distributions.
Eliminate Pictures of People
If you love introducing bias, simply show your sales and marketing staff a picture of a person. As visual animals, we infer a lot about a person from visual appearance. The amount of scientific literature documenting how pictures bias the assessment of people is extensive. For example, you can give hiring managers the exactly identical resume, but vary the gender, race, or physical attractiveness of the picture, and they come to very different conclusions about the resume. Same for names.
With personas, using a picture of a (fake!) person opens the door for a lot of preconceived notions about the person. As mentioned earlier, this impedes your sales and marketing teams’ ability to hear important cues about who the person actually is.
So, people-based persona pictures add zero value (unless you are hiring exotic dancers or klansmen). Instead, they can cause harm by narrowing how we can think about the person on the other end of the marketing/sales message.
Challenging This Idea
Does that mean we should avoid persona pictures altogether? Two months ago, I said “Yes. Absolutely!”, but then I had a talk with Jamie Kirmess, one of LeadMD’s Principals. It went something like this.
J: “The picture gives people an anchor that sticks in their memory.”
E: “Totally. But, do you want inaccurate information to stick in people’s memories?”
***And then she flipped the script.
J: ”Does it need to be people?”
J: “Does it need to be people, or could we give visual anchors that are actually useful?”
This is why I joined LeadMD.
Instead of using irrelevant (and potentially damaging) pictures of people, why not use something that sums up the core of the persona’s Goals, Priorities, Means, and Metrics? For example, a buyer persona with pain around “trying to handle too many tasks coming in at once” could be represented by an air traffic control radar screen. This way, sales and marketing have a visual glimpse into the persona’s soul without getting distracted by the surrounding meat sack.
Let Percepts Override Concepts
Even with the best personas, real world signals should always override the theoretical persona. If an individual in Buyer Persona X exhibits any of these signals, they tell you how to message that person and whether they have a high propensity to buy. What kind of signals can we look for? These can be intent to buy indicators (e.g. they consume a white paper, etc). They can be an acute event (e.g. they experience a service outage with your competitor). These positive signals are basically the ICP of B2B buyer personas, which should immediately bubble up that individual as a priority engagement target. Other signals indicate key pain points, which allow you to message the person differently in real time.
These signals vary by persona. For each persona, your sales and marketing teams should have a set of signposts to look out for, trigger words, behaviors, and situations that tell them to veer away from the script and toward a new path.
Trust your team members to use your intel to inform their conversations.
Let’s do it. It’s morning again for buyer personas. Instead of personas that sit on the shelf and gather dust, let’s create personas that people actually used to effect positive change in marketing efficiency. Let’s do it!
Meet Eric Smith
Eric Smith is an alumni of the Visual Perception, Neuroscience, and Cognition Lab at Northwestern University. He gets excited about discoveries that lie at the intersection of perception and reality. Professionally, Eric bridged into business as a researcher at NU's Ford Center for Global Citizenship, where he caught the strategy bug. He spent 10 years doing private non-market intelligence and strategy as one of the founding members of Firstsight Group (formerly Diermeier Consulting), where he developed a "data science as intelligence" mindset. Along the way, he has met many interesting people with many interesting problems and has been lucky to be a part of solving them. At LeadMD, he aims to apply a scientific mindset to improving the state of marketing data science and intelligence.