BetterUp: Finding the best coach for you at the right moment

BetterUp’s algorithm can predict your needs, match a coach, and measure effectiveness of your development. Sounds terrific, but should we rely so much on algorithm?

Imagine that you graduated from HBS and started your new job. You are working hard every day to make a difference in the world. As you take on more stretch opportunities, you encounter various challenges – managing time, striking the right work-life balance, or maybe even finding purpose at your work. But you no longer have your section mates or your LPA friends that you can ask for advice nearby. So you attend a learning event provided by your firm, but it is not tailored for your specific needs…

Meet BetterUp. BetterUp provides a personalized learning and 1:1 coaching opportunities for professionals, empowered by their AI and analytics. For instance, they can identify the needs of organizations and individuals based on the demographic data and behavioral patterns. They will then match you with a perfect coach based on 150 unique variables with 97% accuracy. (i.e., % of employees satisfied with their original coach) In addition, their analytics will aggregate the data from those coaching sessions to identify company’s stated culture and behavioral gaps. Their customers include Airbnb, Linkedin, and Lyft, and they have increased 45% in employee productivity, a 63% in retention, a 35% in organizational agility, and a 29% in employee engagement. [1][2]

While I was intrigued by their comprehensive services as well as the integration of  algorithm and behavioral science, two questions came to my mind.

  1. Coaching: how do they ensure the quality and how does their matching algorithm actually work?

I was curious how they ensure the quality of coaches given that they have 1600+ coaches in their directory, and how they are actually matching those coaches with clients. Since they don’t disclose much information publicly, I tried to find some research papers from their scientific advisory board and data science team, in vain. Therefore, as a last resort, I applied to become BetterUp’s coach to find out. (I happened to take 100+ hour coaching training last year and I am now in certification program, so I qualified for application…)

Quality of coaches is ensured by its rigorous selection process (8% acceptance rate) as well as continuous learning opportunities. Coaches are selected based on a) credentials (e.g., ICF certified = minimum of 60 hours of accredited training and 100 hours of coaching) and b) familiarity with BetterUp’s behavioral science approach to coaching. In terms of b), once a coach passes the initial resume review and written interviews, they will receive 5-6 hours of BetterUp’s training, and then conduct a demo session. After they join, they have access to various resources in BetterUp Coach Community.

In terms of matching, based on the questionnaire they asked, I would assume that they will be using demographic data (e.g., years of experiences, industry, areas for expertise, name of training institute to identify the types of coaching) as well as behavioral data from written and demo session.  In addition, I think they might be using data from their Coach Community and conduct analyses (similar to to identify coaches with similar approach to improve their matching in the future.

  1. Overuse of algorithm? 

Their comprehensive use of algorithm is impressive, but I think it could be overused. For instance, if you let the algorithm decide your development areas, would that hinder the opportunity for you to ask feedback from your team members directly? If the employees know that the company is collecting data from your coaching sessions, would they feel less inclined to use the service? As in any cases we’ve learned, I believe it is crucial for the companies to be thoughtful about how to integrate these tools, instead of using the algorithm to all possible cases.





AI in Predicting Candidate Potential


Locked in by Algorithms?

Student comments on BetterUp: Finding the best coach for you at the right moment

  1. Great post! A couple of thoughts:

    BetterUp’s claim to 97% match accuracy/satisfaction may not be all that meaningful. To really understand what it means, we need to be able to compare it to a baseline of how satisfied the employees would be without the algorithm – e.g. if we let them manually pick their own coach, or if coaches were manually assigned. Given how stringent BetterUp’s hiring process for coaches seems to be, I suspect the baseline satisfaction rate would be quite high to begin with, so 97% isn’t that impressive.

    Completely agree that companies should be thoughtful about algorithm use. Employees need to be treated as partners in the use of these algorithms, and that means respecting their privacy in coaching sessions. To fully develop, employees need an environment of psychological safety to discuss and learn from mistakes, and a company collecting data on what’s said in the sessions would undermine that.

  2. This was a great read, Miho!! I also agree with K’s points on understanding baseline satisfaction to see the improvement that BetterUp is providing.

    I’m really curious about the focus on aligning algorithm selection towards the culture of the organization over time. I definitely see this as a helpful accelerator to the tool’s effectiveness; however, I wonder if this functionality can scale appropriately? As firms grow larger and larger sub-cultures (that aren’t necessarily bad) are bound to emerge. For example, at a large tech company with over 100k FTEs, the culture differences between senior sales executives and engineering executives will definitely be different. I wonder if BetterUp can (and/or should) control for these cultural differences within an organization? How do they work with their clients to make these decisions?

  3. Hi Miho,

    Interesting thoughts! Satisfaction in coach-coachee pairs is a really understudied topic – we’re not really sure what makes for a coach coach-coachee pair, so like you, I’m curious about how they’re matching people. While I think matching on demographics is an ok start, I think there might be more room for improvement when thinking about relational preferences (i.e., “the soft stuff”). Further, I wonder if left to decide for ourselves, we’d even make a good decision. We might think that we’d want a coach that’s similar to us, but in reality, coaches that are similar to us aren’t good at developing us. That’s where I think the massive amount of data that BetterUp has could be really helpful because we’re able to see how certain variables in a coach-coachee pair do in terms of satisfaction and development/ performance.

    I know BetterUp well (partnered with them on some research) and while I agree that there’s potential for an overuse of an algorithm, they do a fantastic job of letting coachees (1) understand the ways their data is private and protected and (2) see and use their data to their advantage. This is the real benefit of BetterUp – they provide executive level coaching for people at all levels of an org,

    Thanks for the post!


  4. Hi Miho,

    Thank you for this interesting post! I also wonder how much value their personalization and data is adding or if the improvements measured are just a function of the fact that employees are engaging with a mentor (any mentor). In other words, does the AI matching really help improve the outcome?

    I also think BetterUp’s human + algorithm coaching approach is better than Quantified Communications’s fully automated approach. With QC, i thought the analytics was interesting but hard to internalize. I think it would have been helpful if had a human coach to help me interpret the data and elaborate on ways i could improve.

    Thanks again!

Leave a comment