The State of Interview Scheduling Technology

Interview scheduling is a micro-niche activity that penetrates our attention because we all must do it and it can be a surprisingly headachy time waster. You would think that scheduling interviews would be automated by now. 

And sure enough, we do have tools; indeed, most ATSs have some sort of scheduler. The problem is: Most tools are not very good. Still, though, there are effective solutions being developed. 

The Basic

A basic level of automating interviews can be found in free/inexpensive tools like Calendly and Vocus. These tools allow candidates to pick time directly off the interviewer’s calendar. They will automatically block the chosen times in the candidate’s and interviewer’s calendars, send reminders, and load the relevant information into the relevant systems (e.g., a candidate relationship management database). 

Every small organization should be using a tool in this category. If you are doing one-on-one interviewing, then these are excellent tools and provide a huge gain in efficiency.

The Sophisticated

If you have multiple interviewers, then the level of complexity is such that basic tools like Calendly do not do the trick. What you need is a tool that understands things like “Either Ali, Frances, or Fernando should do the first interview for any administrative jobs,” or “For the technical interview for software roles, Rosanna should do the interview, but if she is unavailable, it’s OK to get Regina to do it,” or “John and Mary must be at the final interview for engineering jobs.”.

There are sophisticated tools, such as, that can handle this complexity. In Goodtime’s case, one of the elements that allows flexible and sophisticated scheduling is the use of tags. For example, you can tag all those individuals who can do first interviews for a certain job category. With this information, the scheduler will know whose calendars to look at for a given candidate and find an acceptable time. 

One clever use of tags is to tag any female engineering candidates and then have the system make sure that the interviews are set up such that you never end up in a situation where all the interviewers are men — and vice versa. 

The scheduler is also set up so that it knows your interviewing process. For example, you may have a process where a screening interview is followed by a technical interview for certain job categories. If the scheduling software knows this, then it can set up the whole series of interviews that a candidate needs.

Because these sophisticated systems need to know a fair bit about your interviewers, interviewing process, and job categories, they can take several weeks to set up. However, once set up, the whole process of scheduling interviews is largely automated.

The Artificially Intelligent

It’s now possible to deploy AI to do interview scheduling with tools like These tools can handle a recruiter saying things like, “I need to schedule these 10 interviews, and put as many as you can on next Monday and Tuesday.” This is possible due to a natural language processing engine that can make sense of what was said, as well as a symbolic process engine that is smart enough to figure out how to deliver a good solution to a somewhat ambiguous request.

Article Continues Below

Fans of AI will recognize that the symbolic engine Evie is using is what is called GOAFI (Good Old-Fashioned AI), rather than machine learning. It’s the right engine for this particular task.

Jin Hian Lee, Evie’s founder, says that both the natural language processing and the symbolic engine can work extremely well because scheduling interviews is such a narrow domain. The AI doesn’t need to have a wide range of knowledge, just those few things that are involved in scheduling interviews.

Lee points out that while the domain is narrow, it is very complex. He describes it as a “mountain in a molehill” — it looks simple, but as anyone who has scheduled interviews in a large organization knows, it can get very messy. 

You might like to think of AI systems working in a narrow domain as crossing the boundary from “providing automation” to “being autonomous.” An automated system does a set of tasks in a predefined way. An autonomous system is smart enough to navigate its way to an answer even if the goal is a bit ambiguous or conditions change along the way.

Ultimately, TA pros need to be increasingly familiar with the complex world of automation so that they can decide how to deploy any number of solutions from the above three categories of interview-scheduling technology.

Granted, most recruiting pros will never need to program in Python or build graphs databases or develop adversarial attacks to test the robustness of machine learning. However, they should show an interest in these things since technology mediates everything in HR and TA. At the very least every TA department needs a fair number of people who do show an interest in tech. This is the future.

David Creelman, CEO of Creelman Research, is a globally recognized thinker on people analytics and talent management. Some of his more interesting projects included:

  • Conducted workshops around the world on the practical aspects of people analytics
  • Took business leaders from Japan’s Recruit Co. on a tour of US tech companies (Recruit eventually bought for $1 billion)
  • Studied the relationship between Boards and HR (won Walker Award)
  • Spoke at the World Bank in Paris on HR reporting
  • Co-authored Lead the Work: Navigating a world beyond employment with John Boudreau and Ravin Jesuthasan. The book was endorsed by the CHROs of IBM, LinkedIn and Starbucks.
  • Worked with Dr. Wanda Wallace on “Leading when you are not the expert” which topped the “Most Popular List” on the Harvard Business Review’s blog.
  • Worked with Dr. Henry Mintzberg on peer coaching, David’s learning modules are among the most popular topics.

Currently David is helping organizations to get on-track with people analytics.

This work led to him being made a Fellow for the Centre of Evidence-based Management (Netherlands) for his contributions to the field.