OBDynamics FAQ's
-
How does OBDynamics work?
As a machine-learning program, our platform can be taught to make any number of inferential assumptions about candidates. For instance, we can identify high performers from a wide pool of people who have similar qualifications and backgrounds by comparing things like time to promotion across wide cohorts of people. If the average time to promotion from associate to vice president at Company A is 3.2 years but your candidate did it in 2.7, this is a possible indicator of high performance. Add to that the fact that your candidate graduated from college after seven semesters rather than eight, and you’ve got another indicator. OBDynamics can make these kinds of inferences across thousands of candidates simultaneously, helping us to sort talent cohorts by performance.
OBDynamics’ AI inferences can be individual or population-spanning in nature. On the individual side, it can do things like identify people from diverse backgrounds even if their resume and bios don’t explicitly note diversity, or identify people who are likely to have high-level U.S. government security clearance even if they do not actively publicize it.
On the population side, we can combine data about geographical variations in compensation, taxes, school systems, population gains, and demographic changes to infer things like the desirability of a specific location for a specific type of employee. -
How did we get 1.1 billion profiles?
The information comes from a variety of publically available sources, including social media accounts, corporate bios, and resume databases—information that individuals freely put online. Our value add is that OBDynamics’ AI platform does not merely scrape all this information into a pile; instead it combines and compares data points from these disparate sources, allowing us to build more comprehensive profiles for each person or company.
All of this data is publicly available and ethically resourced. It is GDP and CCPA compliant—meaning that it adheres to the high standards of privacy required by European and Californian privacy laws.
-
What are the OBDynamic Products?
OBDynamics is essentially three different products, which can be packaged together or sold separately.
1. Talent Intelligence. This customizable analytical tool is designed to provide our clients with a detailed map of a specific talent market. The map is highly adjustable, allowing clients to see how the geographical distribution of a specific type of talent changes according to things like historic job titles, skills, diversity characteristic, and years of experience. These results can be further juxtaposed against other forms of macro data like compensation and cost of living.
On its own, this product is not designed to provide name generation for a specific candidate search. Instead, it is a zoomed-out overview of a talent pool, meant to give our clients the information about the talent market that they need to move forward with expansions, relocations, or the name-generating phase of a search.
2. Talent Streams. This is a largely middle-market recruiting product in which we identify and deliver a tailored shortlist of ready-to-contact candidates to the client. It’s best used in instances where the client needs to fill a hard-to-find position that does not warrant full executive search—or in instances where the client is trying to fill a number of roles.
3. Augmented Search Intelligence. Our talent intelligence and talent streams capabilities can be used to make a traditional executive search faster, more accurate, and yield a more diverse field of candidates. Rather than relying solely on traditional network-based candidate sourcing, we can augment this tried-and-true method with candidates sourced from our database, which allows us to consider people who might otherwise slip through the networking cracks. It also helps us find the proverbial unicorn candidates—completing searches that no one else can do.
-
How are Talent Streams different to other name-generation services?
Unlike basic name-generation services, which often generate hundreds or even thousands of names and leave it to the client to parse through all the information, we use an iterative, multi-touch process.
This allows the client to refine their needs as they learn more about the talent market and allows us to narrow our search down to a curated shortlist of candidates. In other words, we curate our findings and deliver 5-10 qualified, ready-to-hire candidates.
-
What kinds of conclusions can the AI platform make about candidates?
As a machine-learning program, our platform can be taught to make any number of inferential assumptions about candidates. For instance, we can identify high performers from a wide pool of people who have similar qualifications and backgrounds by comparing things like time to promotion across wide cohorts of people. If the average time to promotion from associate to vice president at Company A is 3.2 years but your candidate did it in 2.7, this is a possible indicator of high performance. Add to that the fact that your candidate graduated from college after seven semesters rather than eight, and you’ve got another indicator.
OBDynamics can make these kinds of inferences across thousands of candidates simultaneously, helping us to sort talent cohorts by performance.
OBDynamics’ AI inferences can be individual or population-spanning in nature. On the individual side, it can do things like identify people from diverse backgrounds even if their resume and bios don’t explicitly note diversity, or identify people who are likely to have high-level U.S. government security clearance even if they do not actively publicize it.
On the population side, we can combine data about geographical variations in compensation, taxes, school systems, population gains, and demographic changes to infer things like the desirability of a specific location for a specific type of employee. -
How does the machine learning inference capability work?
Though we use numerous machine-learning models, the skill inference model was trained on hundreds of millions of professional, publicly indexed profiles. In essence, it considers the job title, location, and company to infer skills that others who have had that role have reported.
For example, by looking at 100 data scientists at Spotify, the model may infer that if 15 of them have experience working with BigQuery, then the other 85 likely do as well. A similar model infers the candidate’s tenure in their current role in order to identify their likelihood to consider new opportunities. Other models apply company and financial performance attributes to persons associated with the company.
-
Is there evidence that shows OBDynamics can speed up delivery time?
Yes. When we implemented OBDynamics, Odgers got 42% faster.
But the best example is a story: When we first developed OBDynamics, we engaged with a leading Boston-based private equity firm to back test the technology based on a search that another search firm had recently completed for them. The search had taken the other search firm a full year of effort. Odgers was not given any of the names or biographies identified by the other firm, but we were provided specific criteria for the role. In just a few weeks we provided the private equity firm a list of candidates that included every member of the other search firm’s shortlist. -
Our Privacy Policy
All of our data is ethically resourced and GDP and CCPA Complaint—meaning that it adheres to the high standards of privacy required by European and Californian privacy laws.
-
What are our data sources?
We subscribe to numerous market, people, and business intelligence databases, and receive license data from various people data enrichment services. We also have developed proprietary applications that harvest public data from publically available indexed sources, to further contextualize matched records.
-
What is our policy for handling Personally Identifiable Information (PII)
To ensure compliance with all global privacy laws, we have partnered with Manetu, a leading-edge consent-management platform, and consulted with some of the world’s leading counsels on global privacy compliance at the Am 100 law firm K&L Gates.
With Manetu, for example, we have architected our data lake to warehouse EU data subjects’ data on EU-based servers; developed policies for rigorous CCPA and GDPR compliance; and manage all data subjects’ consent through a zero-knowledge blockchain-based platform, where the user can edit, delete, and request their data. Manetu integrates directly with our data lake and assumes responsibility for privacy compliance. -
What kind of information is in the data lake?
It includes resumes and information gathered from social media platforms—but it also includes data sets gathered from a variety of sources. Our philosophy of data use is that an individual’s details don’t mean much until put into the context. What our data lake’s extensiveness allows us to do is provide that context, allowing us to make quantitatively comparisons and qualitative inferences about individuals, their environments, and the relationship between the two.