One of the most memorable sessions at the TechHR Singapore 2019 conference was Brian Sommer’s guide to cutting through the hype around HR technology solutions.

There are innumerable market options that leave buyers confused as to what solutions are actually relevant for their businesses. Brian, Founder, TechVentive, explained how to get past the hype and understand the value of HR technology solutions. In this direction, he shared 5 key insights.

Understand what problem you want to solve using HR technology solutions

Brian suggests beginning with separating hype from reality in terms of what HR tech products can truly do, to stay on the correct side of the new technologies coming up.

80% AI for HR is in the recruitment. However, there is a lot of hope and not a whole lot of science in recruiting automation.

To successfully use and get value from HR tech products, organizations first need to understand what problem they are solving. Then they need to upgrade their HR processes to best derive value from the HR Tech products.

If you don’t know what problem you’re solving, you won’t know what to do with the technology.

Separate visual/aesthetic appeal from commercial viability

Many apps might be fabulous to see but are not commercially viable. Be wary of hyperbole. Ensure that the product you invest in solves a real business problem and can feasibly be used on a day-to-day basis.

Understand the Science behind HR technologies

Brian suggests that there are key scientific parameters to consider when evaluating the value proposition offers by tech solutions:

• Co-relation vs Causation
• Opacity vs Transparency
• Trainable vs Black Box (Bad Training)
• Incomplete data sets

Consider as an example an organization decided to offer “financial wellness” courses to help their employees. Their strategy being: Employees are stressed about finances and therefore if we offer financial planning courses, they will be more engaged and therefore more productive. 

"If this-then that” thinking is shaky on causation and focuses on correlation. Just because someone is or isn’t engaged, it doesn’t have an impact on their productivity.

Evaluating how opaque or transparent the logic behind chatbots/machine learning tools is, becomes absolutely critical to whether it will be a viable solution for the organization’s needs.

If the logic is not transparent, the teams will be forever dependent on the vendor to verify and explain the data and to fix any problems thereafter. Consider also the problem of the logic or algorithms being flawed form the perspective of the realities of the organization.

Narrow down relevant products for your business needs & maturity levels

Brian’s advice to clients of HR technology solutions is to challenge everything. He recommends not taking anything at face value until the vendors can prove all the functionality they are promising at the scale required.

Another tip is to talk to the real customers. In most organizations, the CFOs typically sign off on purchase. As a result, they are naturally inclined to give the purchases a good rating rather than critically evaluating the decision. However, the HR teams using the product may not give it a good rating at all. It is important to take into consideration feedback from the actual end user.

On hype without content, Brian offers the following example. A tech vendor pitched as follows, “If you know employees whose blood sugar is low, we can get the employee engagement up.” This kind of hyperbole and poor logic must always be challenged.

1. There is such velocity in the marketplace, chances are that by the time you implement what you bought, there will better apps out there. 
2. Frequent upgrade or replacement cycles will be the norm
3. In an increasingly global workplace, most solutions will need to work harmoniously across different geographies with varying regulatory systems, cultures and philosophies. The product market share and viability will vary; adoption of the technology will vary. For example, one single payroll product may not work globally with perfect precision.
4. The new data protection regulation varies country to country and this will have a huge impact on global applicability of recruitment products.

Understand the unintended consequences of choosing the wrong technology

Perhaps the most important idea to watch out for is that there might be unintended consequences to tech solutions.

In 2014, Amazon developed an Artificial Intelligence recruitment system. The system taught itself that male candidates were preferable. This bias against female applicants crept into the system because the data that was fed to the system was collected over 10 years from primarily make candidates. So the “she” “women” or names of all female education institutions were mentioned, the system downgraded those applications. Of course Amazon shut the AI system down once this flaw was discovered. But it offers a sobering lesson that bias embedded in technology will severely amplify systemic flaws.

Another example for ways a machine learning system can teach itself the wrong logic is the Hawthorne Effect or Gamblers Fallacy, where a lucky streak of the same result coming up, may lead to the machine thinks it's the pattern.

In conclusion, organizations need to critically evaluate their needs before they sign up for a particular HR Tech solution in order to avoid themselves from being swept away by the hype around these solutions and ending up with a technology which is might not be able to solve their exact problem.