Talent Cloud has been an unusual initiative for the Government of Canada.
The team took a unique approach to testing ideas on optimizing the hiring fit between talent and team, increasing inclusion and diversity in recruitment, and reducing time to staff in the Government of Canada. Talent Cloud went so far as to build a full staffing platform where new ideas could be tested from concept, through hiring, and all the way to on-the-job performance.
The platform is for externally advertised, competitive processes and is open to the public. The project itself is open by default and strives to embrace the GC Digital Standards in all aspects of decision making, design, and development. The initiative was built through extensive user testing, and engagement with partner departments, international experts, and equity-seeking groups, including Indigenous communities.
Talent Cloud has been built from the margins in. Every assumption, every process, every value was turned around and viewed from other angles… many other angles. The idea was to build for the edge cases and then scale towards jobs and groups that are already well supported by the government system, rather than to build for what is easy and known, and then try to scale a standard model to include new types of work and underrepresented groups. Talent Cloud is a platform intentionally optimized to recruit, for example, a black or Indigenous developer with leading edge skills and unconventional education, and provide opportunities in emerging and hard-to-fill areas of need. The platform was built to attract a new group of applicants to a new type of role in the Government of Canada. Inclusion by design, not by accident.
The platform was also engineered to support a talent model optimized for the digital age, including actively enabling project-based work (Talent Cloud recruits for term positions, but the platform could theoretically be used more broadly). Significant effort went into behavioural interventions and process redesign related to the applications themselves. To be more specific, Talent Cloud aimed to reduce the overall volume of applications per process to save time and energy, while increasing both diversity and the percentage of high-performing applicants in the pool. Fewer people, better outcomes. This meant targeting a shift in application behaviour patterns - not through recruitment drives, but through interventions on the platform itself.
In the digital age, a project-based position can’t take longer to staff than the position is needed for. A responsive and agile government must be able to secure rapid access to high-performing talent, and attract those with in-demand and emerging digital skills. To be competitive with industry, Talent Cloud set the target of building a platform that would produce a highly ambitious 30 day time to staff (plus security clearance time, which was outside Talent Cloud’s scope).
In the past three and a half years, the Talent Cloud team and its partner departments have crafted new behavioural and business processes, built and launched a live staffing platform, and run more than 50 staffing processes to test ideas and measure outcomes. In the end, the results were impressive.
While Talent Cloud’s initial staffing processes took approximately as long as the Government of Canada average, two years after the platform launch the average time for an externally advertised process (from job advertisement to verbal offer) is down to ~40 days (plus the standard 1-2 months for security clearance and HR finalization). Several processes in 2020 reached a verbal offer in ~20 days. Including security clearance, this means the time to staff on the Talent Cloud platform is now down to ~3 months… a savings over the Government of Canada average time to staff by almost half a year.
On average, ~9% of all applicants in Talent Cloud processes have been deemed fully qualified (qualified to receive an offer) at the time a job process closed, which is significantly higher than the industry average of ~2% (Jobvite, Harvard Business Review, Knowledge@Wharton). In processes conducted in 2020, this number has been over 10%, as the platform continues to release improvements in optimizing the hiring outcome. Managers interviewed 1-2 years after their hire was made indicate overwhelming satisfaction with the quality of the hire and the fit-to-team. ~95% of these hires (according to managers interviewed) remain in government, even after some having required term renewal, and ~80% are still with their original team. Qualitative research shows that the platform is attracting strong applicant diversity, and that this is transferring through into the final hiring result. Research with applicant groups shows that the deliberate efforts Talent Cloud has made are working: the platform is reaching new audiences, including those in equity-seeking and underrepresented groups, and user feedback on the experience is positive.
In addition to producing hires, Talent Cloud also generated research findings. For example, the team learned that factoring work environment, team culture, and management leadership style into the design of the job advertisement significantly improved the talent-to-team match. The team also found that making positions remote work accessible substantially improved the chance of a successful hire. Talent Cloud tested 20 different points of intervention related to reducing time to staff, and identified the 5 that were the most influential. The team ran experiments on the usefulness of applicant self-assessment, optimizing the number of selection criteria, and ways to reset the defaults to promote diversity and inclusion.
It’s important to remember that Talent Cloud is only just completing the experimental stage - it hasn’t been resourced as a full scale solution. As a result, the majority of its features have been released as minimal viable products. This means that the platform has a few bugs and glitches, and has never had the force behind it for adoption that an enterprise solution would have. Yet despite having been built from a lean startup approach, with minimal influence, the experiment worked. However, confirming whether or not the model would work at a larger scale would require a new approach.
The question now is what will the Government of Canada do with the insights gained and the lessons learned?
