How SAP Business Beyond Bias productizes inclusive processes within SuccessFactors – an illustrated review


SAP’s Business Beyond Bias initiative has always been a compelling story. But now comes the hard part: overcoming algorithmic (and human) bias while productizing better processes. Here’s an illustrated tour – and my assessment.

Screen shot, SAP Workforce Analytics

One of SAP’s best stories is its commitment to diversity via its Business Beyond Bias initiative. Powerful videos on diversity set the right tone at SAP, but that wouldn’t hold up in the long haul – not without a deeper effort.

At Sapphire Now 2017, I blogged about my video shoot with Dr. Patti Fletcher, Leadership Futurist with SAP SuccessFactors.

My prior article shows how SAP is tackling the issues, but there’s a big missing piece. Since Sapphire Now, I’ve been tracking how SAP is embedding inclusive processes within SuccessFactors. Tackling “inclusion” in software means confronting algorithmic bias. It means acknowledging that no software is neutral. But it also means helping customers understand their choices. Customers don’t want to be forced down a narrow path of mandatory actions.

In this piece, I’ll delve into several illustrated examples of how SAP is approaching Business Beyond Bias in a productized sense within SAP SuccessFactors. I’ll bring in interview highlights with the Business Beyond Bias team to give context – and I’ll assess what I’ve seen.

SAP shares business case and diversity data

SAP has put out a flurry of content to help customers make sense of the issues. There is an e-book on developing more diverse/inclusive processes with SuccessFactors.The good news? The business case for diverse workforces gets stronger with each data point:

But there is concerning news as well. In a post by Lyndal Hagar and Gabriela (Gabby) Burlacu, SAP shared some numbers from its own Workforce Analytics data. From the blog post:

  • Our data shows that female representation in organizations across the globe has remained steady– and rather unequal.
  • Nearly 75 percent of companies employ a greater number of males than females, with the 10% of companies that are the least gender diverse employing four to five men for every woman.
  • Well over 75 percent of companies in our sample employ more male managers than females, and the 10 percent least diverse companies have six male managers for every female one (a number that is trending upward, not down).

How SAP is productizing Business Beyond Bias

Education and data are helpful, but how is SAP productizing their diversity efforts? Answer: in two ways: by helping companies take better advantage of existing functionality (ebook link), and by embedding specific “Beyond Bias” components throughout the HCM workflow. SAP has identified key areas within HCM that are most impacted by unconscious bias:


For each of these questions, SAP has provided guidance to customers (and web site visitors). The SuccessFactors diversity and inclusion home page is a good starting point, and this PDF brochure provides an overview of each question, and how customers can address each with SuccessFactors.

I am wary of hearing great things about cutting edge products that aren’t yet available. So when it comes to the Business Beyond Bias products line, SAP provided me with the current schedule of four key products, and when they became – or will become – generally available (GA):

  • Mentoring – Went GA in Q4 2016, almost 200 customers running it (see: blog post)
  • Photoless Calibration – Went GA in Q1 2017 (see: blog post)
  • Calibration Alerts  – Beta Q4 2017, planned GA: Q1 2018 (see: blog post)
  • Job Analyzer – Beta Q3 2017 planned GA: Q1 2018 (see: blog post)

Quick look behind Photoless Calibration

Lyndal Hagar, Principal Product Manager – Diversity & Inclusion, SAP SuccessFactors, walked me through the demos. Photoless Calibration, now in GA, is based on a simple but important concept: research has shown that sometimes a photo isn’t necessary to make an HR decision. As Hagar told me:

The classic case is recruiting: Why do you need to see the photo of the candidate?

But photoless calibration has implications beyond hiring. When the Business Beyond Bias team talked with the SuccessFactors Customer Advisory Board, they asked customers about the different decision points SuccessFactors supports. They identified other scenarios for Photoless Calibration, such as when a manager sits down with the HR team to assess performance ratings.  Hagar:

What we’ve done is provided customers with an alternative view. In the calibration, rather than seeing the photo, you see a silhouette.

But with performance management, it’s not just about removing a photo that might incur bias. It’s also about putting information in front of managers pro-actively:

The type of information you would typically glean from a report after the fact is now the kind of information that you would see up front. It’s now provided when you’re sitting there making the decision, thinking about the performance ratings that you’re giving to your team.

Here’s a screen shot example, with performance categories:



SAP is refining as they go. Hagar told me that they are figuring out when it’s useful to include demographic information. So, for example, this screen now includes gender info. That has impact. You can see, for example, that there are more men in the high performance area. Gabriela (Gabby) Burlacu has been heavily involved in human capital management research at SAP (she wrote the e-book I linked to above). As she explained:

The screenshot is actually similar to what happens in the real world, where women get really high performance ratings, but very low potential ratings, so they end up not getting that top box. I think one way you can measure success in diversity efforts is to really take a look at the breakdown, and see if it’s looking more accurate, and/or more looking more gender diverse in each of the boxes.

Quick look behind Mentoring

Mentoring, also in GA, can be used in in a couple different ways. The algorithm matches mentors and mentees. Rather than being assigned a mentor who works in the same office as you, the algorithm matches you based on the skills and competencies you’re looking to grow.  Hagar:

You can configure it to be purposeful in its matching, so if you’re running a women in leadership program, if you want to match women with women, you can use it in that way also.

Here’s a screen shot:


It may not be obvious how this connects to diversity, but here’s my view. For those who are struggling to find community or fit into work social settings, being able to seek out a mentor across the organization could be a beneficial way of accessing know-how (and political pointers) that you can’t get when you are feeling isolated in your immediate surroundings.  I’ve heard plenty of anecdotes lately about the price of cultural isolation at work; I’m sure this type of formal connection program would be useful there.

Quick look behind Calibration Alerts

Calibration Alerts help tie these functions together by pro-actively alerting to situations where companies/managers may need to take action. This functionality stemmed from SAP’s work with customers to identify the issues they’d like to act on in advance, rather than in a static manner via reporting. When it comes to retaining talent, you can’t afford to wait until problems mushroom into a reason for taking a job elsewhere.

Examples of alerts might include: an employee who comes back from a leave of absence, and their manager rates them lower than their previous rating.  Hagar:

The customer can configure what type of alerts they want, so we’ll give them these scenarios about a lack of promotion, or reduction after a leave of absence. They can either run them across the entire population, or they can target them on under-represented groups.

Here’s a screen shot, showing an alert of a “Bob Jackson” who has been rated as a high performer for three years without a promotion:



Quick look behind the Job Analyzer

The “Job Analyzer” may be the most well known of SAP’s new Beyond Bias components (it was featured in several keynotes). The short version: it’s intended to flag up phrases that might be off-putting in job orders, surfacing biases in descriptions we might not be aware of. In this screen shot, you can see the potentially biased words (“superior” and “delicate”) flagged in red on the right hand column:


My take – software isn’t neutral, and the work has just begun

My biggest concern about this post is giving the impression that I believe diversity can be solved via smarter software. I don’t believe that; I know SAP doesn’t believe that either. A mechanistic approach to diversity would surely fail anyone it touches.

But we need to accept that software isn’t neutral – algorithms certainly aren’t. If technology can uncover some of our biases, rather than reinforce them, that’s a worthy effort. There’s no substitute for turning hiring practices upside down. There’s no substitute for a diverse leadership team and board of directors – something SAP itself still needs to progress on.

This isn’t as simple as running a spell or grammar check. If the Calibration Alerts show that a manager isn’t promoting high performers from a certain background, that’s a potent situation. Then there’s the education and cultural commitment to applying these tools. SAP has worked hard on both, but that’s the nature of this one. Work harder until it’s solved. Until then – no excuses.

One thing I respect about SAP’s Business Beyond Bias team is that they don’t sugar coat these obstacles. The team realizes they need to expand many of these tools beyond gendered bias to other forms of inclusion (I should add that Workforce Analytics is another big piece of this puzzle, as customers assess their progress – see featured image above for a screen shot). Fletcher dropped this funny and memorable line:

We’re trying to bring the mountain to the hobbit.

Yep – you can make the journey easier and provide a map, but the mountain looms. And speaking of mountains, the SAP Leonardo team better be working closely with the Business Beyond Bias folks. Otherwise, Leonardo’s sexy tech will be next-gen in name only. The Beyond Bias team says they have good input on the Leonardo side; I’ll ask the Leonardo team about that next time I see them. My next editorial goal here? Profile a Business Beyond Bias customer.

I’ll give Fletcher the last word:

It really is the early days. This is a drum beat that we keep having to keep in time with. There is no end game for this; there are so many different kinds of diversity. We have to figure out how to enable all those key decisions: everything from how to restructure the data that we gather, to who do we hire and promote, and everything in between. I think we’re validating that.

Oh, and if you want to see the rarest thing of all – an exceptional piece of corporate media – look no further than this 2016 video:

End note: special thanks to Geraldine Lim of SAP SuccessFactors communications for her stellar efforts related to the material needed for this piece.

London 2018 UNLEASH banner - diginomica supports

Image credit - Screen shots provided by SAP, including feature image.

Disclosure - SAP paid the bulk of my expenses to attend Sapphire Now and SAP SuccessConnect, where the bulk of my interviews for this story were conducted. SAP is a diginomica premier partner.

    1. Phil Wainewright says:

      Important also to realize that having a diverse workforce is only one part of it – culture must be inclusive too. Therefore the segment ‘How are people managed?’ in Figure 2 is crucial. Otherwise your attempts to recruit, reward and promote in an unbiased manner will be subverted by a failure to include people in day-to-day decision making.

      In that context, I’m impressed by the work that Cloverpop has been doing on measuring the inclusiveness of decision processes in an organization. Because unconscious bias in decision-making, even in organizations that prize their diversity, is a real obstacle to realizing the benefits of a diverse workforce. For more on this, see:

      1. Jon Reed says:

        Phil thanks for the Cloverpop link and reference. We need more vendors to tackle this problem.

        As for your comment: “Therefore the segment ‘How are people managed?’ in Figure 2 is crucial. Otherwise your attempts to recruit, reward and promote in an unbiased manner will be subverted by a failure to include people in day-to-day decision making. ”

        Yep. Thanks for bringing that out. That’s really a core topic we should continue to cover on diginomica from several angles.

        Phil just one add: as I was linking to this piece I previously did, What I learned from taking Facebook’s workplace bias course, I found this line from my final take that I think lines up with what you are saying here:

        “Without incorporating diverse feedback early into design and policy decisions – including the feedback of external constituents – the competitive edge of diversity will be squandered. A more diverse constituency is just a starting point. After that, your culture determines how much of an edge that diversity is.”

        – Jon

    2. greg misiorek says:

      i’m not biased, but…

      this is really long (and very good) but unfortunately behind the paywall and in German: . maybe some of your readers are also subscribers, but what i gathered from it is that finding more about it (bias) is the first best step one can take.

      for the rest, here’s a shorted version and with tongues-in-cheek:

      thx for the nice writeup and the update on SAP’s product pipeline.

      1. Jon Reed says:

        Greg thx for sharing that………

        ” finding more about it (bias) is the first best step one can take.”

        Agreed, though I’d break it out into:

        – doing all we can to uncover and address our own biases
        – understanding and grappling with the reality that our code and our software) includes such biases which must be accounted for
        – hopefully addressing pro-active ways of alerting to potential bias

        (the third one being the focus of this piece)

        No way I can really satisfactorily cover it in one piece but fortunately our team returns to this topic often. But – if you want to get a deeper glimpse into how I think about confronting bias, check this prior piece, What I learned from taking Facebook’s workplace bias course.

        – Jon

    Leave a Reply to Phil Wainewright Cancel reply

    Your email address will not be published. Required fields are marked *