Now that you know the behaviors that are most important, it’s time to develop behavioral interview questions that will allow you to assess candidates against them.
The most effective interview technique in the world today is to use behavioral interview questions. There are reams of data on the technique, comparing it to others. Until there’s a new way (AI, perhaps), you can’t call yourself a good or professional interviewer unless you’re asking behavioral questions.
The best example of behavioral interviewing today surprises many managers. It’s software developers being asked to write code or debug some code during a technical interview. While most of us don’t think of this as a behavioral interview, it is. It proceeds from the same premise: Past behavior is the best predictor of future behavior. If the primary behavior in the role is writing code, and you can actually assess that behavior, why wouldn’t you?
The same would apply to a role where using spreadsheets was a key factor. There are wide disparities in how people assess their own skills. Some people consider themselves experts in MS Excel, for instance, but can’t create a pivot table. Why not assess their behavior? We have heard surely a hundred anecdotes of managers doing so and finding “experts” who stare at some relatively simple Excel problem and can’t even begin to solve it.
This is not to say that every role lends itself to such a test. It is to make the point that such a test is in fact a form of a behavioral interview. Past behavior/performance is the best predictor of future behavior, and in these cases, objective skills can be behaviorally “tested.”
Of course, it’s not possible* to “test” someone’s ability, for instance, to get a team aligned and fully communicate over a several-month project. So we ask questions about past examples of the behaviors we are looking for rather than directly testing them.
You’ve probably heard or experienced some of the other formats and types of questions that are asked in interviews. Almost all of them are worthless because they are terrible predictors of true positives and true negatives.
But they’re not useful questions unless the role requires such estimation and problem solving on short notice. And while there are such roles (management consulting, for instance), it’s unlikely yours is one of them.
Even the classic question, “Tell me about a weakness/Tell me your greatest weakness,” isn’t all that useful, according to interview results and performance comparisons. While it does get to self-awareness, that’s usually something easily determined by asking a number of behavioral questions. What’s more, high performance comes from leveraging strengths, not from improving someone’s weaknesses. Also, sadly, most interviewees aren’t terribly truthful with their answers.
“Unstructured” interviews have become more popular. They are also equally terrible at predicting performance. This is where the candidate asks all of the questions. They’re certainly easier for the interviewer to prepare for, and maybe even more “fun” for the interviewer. But that’s irrelevant. What matters in interviewing is efficient prediction of future performance.
There’s another “type” of interview that can be called “unstructured.” This is the interview conducted by the unprepared interviewer. This is just an unprofessional way to go about doing the most important strategic job of managers: assessing talent. It’s funny that those who conduct these interviews reject other ways because the other ways don’t have “data” to support them. Maybe funny is too kind a characterization.
If you’re a manager who tends to be analytical and you’re looking for data to support whatever managerial/professional approaches you take: In my more than 30 years of interviewing candidates all over the world, there has never been any system that has any significant predictive effectiveness other than behavioral interviewing.
A good behavioral question has three parts: the helpful lead-in, an open-ended beginning, and the behavior you’re looking for. Put it all together, and it sounds like this:
Each of these questions has three parts. It’s the use of the three-step building process that makes creating a behavioral interview question simple.
We’ve already shown the three parts: the lead, the question, and the behavior. Before we describe how to build the overall question, it helps to explain why the parts are important to the question.
Part 1 is the lead. It’s often skipped, but it’s there for a big reason. The lead tells the candidate exactly what we’re going to expect in the job. We’re telegraphing what we’re looking for. We’re making it easy, but only for those who have what we’re looking for. We can telegraph what we’re expecting a candidate to do in the job because if he hasn’t done it, it will be obvious when we probe any answer that isn’t based on direct and related experience.
Frankly, we want to make it easy for those who have the experiences we want. When decisions are tough, the decision support method ought to “spread the field”: make the worst performer as far from the best as we can get them.
Part 2, the question itself, asks for an expansive answer. Note that in our example question we’re making a request rather than asking a question: “Tell me about a time when . . .” You don’t want to start with a who or a what or a why, generally. Those are certainly good question words for many interview questions, but generally not for behavioral questions. Behavioral interview questions suggest the need for a narrative, a description, a longer answer than most candidates tend to deliver to who/what/why questions.
Part 3, the behavior part of the question, is closely related to the lead. In the lead, we give the candidates an idea of what our job will require them to do. In the behavior part, at the end, we state specifically what we’re looking for from their answers. It’s couched in the form of a situation. What’s interesting about this is that many candidates don’t hear the behavior. They’ve heard the lead, they think they can just talk about a recent experience that matches what we’ve shared, and they either don’t choose well or don’t have what we’re looking for.
But that’s okay, because if they don’t have it, we want to know that, and if they do have it but don’t highlight it, we’ll be probing their answers to see whether it’s there.
There are really two parts to creating the Lead. The first is thinking about the key activities that you expect in this role. The second is actually crafting the wording, which is pretty easy.
For the first part, the activities we expect from the role, use the raw material from your earlier research. Further, here are some questions to ask yourself:
This gives you is a whole lot of “stuff”—they might not exactly be behaviors, but that’s okay. It might be “stuff” like “runs project meetings,” “produces C++ code for our core application,” “inspects bridges and compares them to published standards,” or “creates new product ideas.”
The other way to do this is even easier—ask your directs to answer the questions. You have to change the questions to their point of view, but that’s easy enough. And by the way, give them no more than 48 hours to do it. We’ve found that all the good work happens right in the beginning.
Now for crafting the wording. Simply put the “stuff”—the behaviors, if you’ve been that precise—into a lead-in statement, a stand-alone sentence, with an introductory phrase that approximates, “Periodically we engage in behaviors in this area here.” Some examples:
This is the easiest part of all three. You can choose from a prepared list.
There are more that you could use if you prefer. But don’t spend too much time here: Spend less here and more time on knowing what behaviors matter in the job.
Simply take those behaviors—that “stuff” from your preparatory work—and add it to the question that you started in part two.
Part one and part three are closely related. If you use the exact same words, it might seem funny to you, so change them a bit.