This blog post is based on the tenth episode of the data-driven recruiting podcast hosted by CodeSignal co-founders Sophia Baik and Tigran Sloyan. You can find and listen to this specific episode here or check out the video version embedded below. When identifying talent data sources, it’s essential to focus on accessibility and accuracy. Just because a candidate may have one source readily available, doesn’t mean it’s highly accurate. And those that are accurate may not be widely used. Let’s jump into the top five talent data sources and see how they rank. Coursework and GPA These two data points are most relevant if you’re hiring current students or recent graduates. Because almost anyone who attended a school or took online courses will have them, they rank as a medium on the availability scale. When it comes to accuracy, these measurements can be misleading. At more demanding schools, a C grade may indicate you studied and worked hard. Alternatively, some schools don’t offer challenging coursework, and all A’s could indicate an average student. This metric could be more accurate if you’re familiar with the school itself and its level of difficulty. On the overall accuracy scale, it receives a medium. References and Referrals Referrals aren’t often readily available because you can’t ask for them upfront, unless you know the candidate fairly well or happen to know one of their colleagues. This becomes much less likely with high volume recruiting. Even if you did ask and receive referrals or references, there’s a big potential for bias. Having someone gush about a candidate is valuable, but only if they worked closely with the individual and can offer concrete evidence of their skillset. In those cases, these tools can provide a lot of value. For references and referrals, availability is low, but accuracy is high. Competitions and Online Challenges Certain roles and environments, like a data scientist or university setting, are more likely to have exposure in these areas. While succeeding in a competition or online challenge is difficult to fake, very few people actually do them. They’re very hard, and even an experienced individual may not place in the top results. Plus, they’re often very public, and many don’t feel comfortable putting their name out there. This means the availability of data is limited. That leads to a low availability score but high for accuracy. Resumes We can probably all answer the availability question when it comes to resumes – they are everywhere! Every applicant will likely produce a resume, but don’t assume that means it is accurate. Resumes, by their nature, allow the individual to fill in the details. A rich pedigree may help, but even that isn’t accurate because attendance at a top tier school can’t validate a direct skill. Resumes have a very high availability but low accuracy. LinkedIn Believe it or not, LinkedIn may rank higher than resumes for availability. This is because it’s not a part of the application process. Anyone can search and find your profile. And although LinkedIn encourages the use of endorsements, which you won’t find on a resume, they should not be readily trusted. You should expect to see less exaggeration on LinkedIn than a resume because it’s public and can more readily be validated. This leads to slightly higher accuracy and availability. So, what about specialized sources for software developers? Stack Overflow is a question and answer website and a common hotspot for software developers. Once online, however, most people are just consuming, not engaging. To have higher accuracy, and even availability, there has to be interaction. An updated profile and active participant in the platform will potentially have posted endorsements and various metrics by which to validate a skill. In those cases, it’s a much better signal than a resume when available. GitHub Inc. is a collaborative code community. The accuracy can be questionable because codes can be copied, and individuals typically only upload their best projects for display. And it’s not very high on availability. Again, just like Stack Overflow, an active user will offer more data than a disengaged one. One data point that offers the highest level of accuracy is a direct measurement of skill. Unfortunately, it’s not yet readily available, and just like references, you have to request it. With CodeSignal, we’re changing that and envision a scenario where candidates come through the door with certifications. Although we’re still in the early stages of this industry shift, we’re excited to see an increase in the availability of this highly accurate tool. When you’re thinking about talent data sources, be sure to remember and assess the availability and accuracy of any given tool. Although it’s great having something that is readily available, you’ll be compromising the value if it’s not accurate too.