When recruitment, staffing, and workforce provider TrueBlue embarked on its cloud journey in 2018, it approached the migration as a way to create value for the entire enterprise, rather than just an IT enhancement. Having migrated the bulk of its primary data centers by 2020, the company is now innovating at speed, operating at far greater capital efficiency, and releasing a steady stream of new products and services. Jeff Dirks, TrueBlue’s chief technology and information officer, discusses his company’s cloud journey with McKinsey’s Chhavi Arora, Stavroula Papadopoulou, Venkatesh Lakshminarayanan, and Brendan Campbell. What follows is an edited version of that conversation.
Putting technology in service to the business
McKinsey: Could you give us a little bit of background on your cloud program and where you are in your journey?
Jeff Dirks: When I joined TrueBlue nearly four years ago, we had nothing in the cloud and were debating whether to go there and, if so, which provider to go with. My vision was to become a technology organization in service to the business, allowing our business leaders to realize value through technology innovation—rather than being an IT-centric organization. What I mean is how we think about our products and services across our three brands in the staffing industry, and then working backward from the benefits and competitive advantage we could create. From there, it’s about leveraging technology to deliver those products and services, and the corresponding business value.
The cornerstone for all that was getting into the cloud, and that journey began with a lift-and-shift approach, where we decided to get out of our data centers. We managed to get our primary production data center migrated at the end of 2020. Now we’re working on residual, lift and shift for back-office data centers. For us, the cloud was also about building continuous integration and continuous delivery (CI/CD) pipelines and focusing on improving our velocity and acceleration metrics.
Lastly, from a business perspective, it’s been transformative in terms of capital efficiency relative to innovation. By using a cloud-consumption-based economic model, we can test and learn and form hypotheses in a capital-efficient fashion. For the ideas that don’t work, we shut them down without having to spend materially. And for the ones that we believe represent tangible business benefit at scale, we can continue to invest and grow our spending proportionally to benefit realization, versus spending well ahead of any benefit realization. We spend 11 to 12 percent of our department budget on our cloud service provider.
Managing parallel work streams
McKinsey: While you had an ongoing lift-and-shift effort, did you have a team doing new deployments in the cloud?
Jeff Dirks: The trajectory with the lift and shift was very much supported by our cloud service provider and our technical account manager. We had very close guidance from them in terms of the sequencing of systems to move.
We didn’t start with the most complex systems. We began with simpler systems, which provided a significant learning experience that allowed us to build up muscle memory that benefited the rest of the lift-and-shift effort.
In the past four years, we’ve also been releasing products to market. For example, our PeopleScout business launched Affinix, a platform for automating the recruiting and outsourcing processes that our recruiters execute on behalf of sourcing, recruiting, and hiring candidates for major clients. We also have a significant amount of innovative work we are undertaking in our PeopleReady “on-demand” staffing business—simplifying how contingent workers can be matched with work opportunities. Within each area, we began to automate our pipelines to support our CI/CD approach. Building our data pipelines to support the training and development of our machine learning and inference pipelines became much simpler because we didn’t have to buy a lot of infrastructure up front. We could consume it on an on-demand basis, which made us much more efficient.
So as we’re migrating applications, we’re also rapidly developing a very deep level of sophistication on how to use and leverage our cloud platform. That’s particularly true in regard to data science and machine learning, where we’ve made significant investments and are seeing some early breakthroughs that we’re very excited about.
Incentivizing cloud training from the top down
McKinsey: What sort of training have you done so your people can work with your cloud service provider?
Jeff Dirks: We made significant investments beginning in 2019 on upskilling the technology organization. We started with basically zero people on the team with cloud certifications, and today roughly 40 percent of the team worldwide has been certified at some level. It was a major investment, and we incentivized people to get those accreditations and continue to be very intentional in our promotion of a technology culture of continuous learning.
That helped us gain more momentum as more people—including myself and my leadership team—all earned accreditations early on to set an example. And the team really took advantage of these training opportunities, which accelerated our expertise in working with our provider tremendously.
As we continued to migrate and our monthly spend increased, our provider was also good at identifying beneficial programs along our journey and gave us access to pro bono consultative services at key times.
The business benefits of cloud
McKinsey: You mentioned looking for business benefits from the start of the journey. How did you approach this?
Jeff Dirks: I put a lot of energy and thought into how to move us past this just being an “IT thing.” And if you look at my background, I’ve spent far more time on the business side leading technology companies than I have as a CTO.
I really approached our migration to the cloud not as an exercise in getting to the cloud but as an exercise to demonstrate the tangible business benefits as a result of being in the cloud. One example was eliminating the need to build a completely redundant disaster-recovery data center in a different region, with all the accompanying expenses of physical allocation, procurement of equipment, licensing, and so on.
Second was pointing out our ability to innovate with far more capital efficiency. We can test hypotheses for thousands of dollars versus tens or hundreds of thousands of dollars by consuming cloud-based services on an on-demand/consumption basis versus having to predict demand and usage up front for a hypothesis that may or may not pan out at full scale.
Those innovations are material. We have a very vibrant patent pipeline. I think we’ve filed eight or nine general patents in the past year, and there will be another 15 or 20 that will come out of our work. And we’ve been able to do this because it doesn’t take a lot of money to go fast and do very innovative things.
Overcoming inertia and resistance to new ways of thinking
McKinsey: What were some of the challenges you faced on this journey?
Jeff Dirks: An initial challenge was moving from capital expenditures to operating expenditures. The economics of capital expenditures, being able to capitalize work and depreciate it over three years, is much more friendly than operating expenditures, which goes dollar for dollar against your earnings. And so we had to put a lot of energy into making those early economic arguments around why it made sense to move to the cloud.
Secondly, I would say the inertia of old ways. We had the classic back-office IT, where everything was ticketing and silos. We needed to move to a “run what you build” mindset, with small “two-pizza” teams creating pipelines that reflected the ability to deploy. We also had to overcome the old approaches to what I call the “star chamber,” our weekly change advisory board, which was a place where great ideas and work would go to die by committee.
We just systematically killed all that and automated all of our policies as code. So now we can be compliant at a far higher velocity. We also brought in new blood to help us go fast but remain compliant as a public company.
How remote work helped to attract nationwide talent
McKinsey: What was your experience with hiring talent and creating a team that would drive this transformation?
Jeff Dirks: I’m fortunate in that I’ve been here in Seattle for 30-plus years working in technology. I’m a University of Washington computer science alum, so I have very deep ties in the community, which helped attract data science and application developer talent here in the Seattle area.
The shift to remote work during the pandemic ended up helping us open doors to more talent as well. We had tech centers in the Seattle and Chicago metropolitan markets, with a smattering of people around the rest of the country. And as we moved through March and April of 2020, I was seeing the productivity gains and how the organization became much better at working remotely.
So we changed our geographic hiring focus from predominantly Seattle and Chicago to all of North America. And while we still have a concentration of engineers in the Seattle and Chicago areas, we’ve become very geographically diverse. And that’s given us the ability to cast a much broader net to attract and hire phenomenal talent. We plan to remain highly remote, apart from periodic in-person meetings and strategy sessions, because this geographical diversity has helped us tremendously.