How DHL harnessed genAI to unify 200 career sites into a single platform

Like many global organizations, German logistics company DHL had career sites for almost every country in which the company operates — nearly 200 of them. Most of the sites, however, lacked brand consistency and none of them were leveraged to attract new hires. They were merely a place to fill out a job application.

DHL’s country-centric career sites also had limited integrations across a multitude of applicant tracking systems — this at a time when skills-based hiring had been growing to meet digitization and other business transformation efforts.

That prompted DHL to turn to generative artificial intelligence (genAI) to fix its talent recruitment problem by consolidating its 200 disparate sites into one site that parsed job seekers into ranked lists to match the company’s needed skillsets, and to which recruiters could quickly respond.

To create its unified site, DHL put together its own internal AI team and partnered with Phenom, a Philadelphia-based HR tech startup that sells a SaaS-based platform offering genAI and automation tools.

Though the company now has just one career site, each of its national locations can maintain their own identities with real-time customization capabilities. For example, in Prague, DHL created a site specifically to attract frontline workers for the peak holiday shopping season. (The low unemployment rate in the Czech Republic had made finding talent a challenge.)

Since the revamped site went live, the number of job seekers per job increased by 25%, 800,000 job searches found a relevant job within 5 milliseconds using AI-enabled features, and more than 200,000 job seekers became applicants in less than 14 minutes of the site’s launch.

DHL

Meredith Wellard, vice president of Group Talent Acquisition, Learning and Growth at DHL, oversaw the AI project from the HR side. DHL’s Express division had recently launched the Phenom Intelligent Talent Experience platform and saw early wins, piquing Wellard’s interest. So, she decided to expand it to the entire company. The results were almost immediate. Wellard said the new AI-based career site produced a spike in candidate traffic, greater overall engagement, and faster hires. The shift to a more engaging career site was particularly good at attracting frontline workers, especially truck drivers, delivery people, and warehouse workers.

Wellard spoke with Computerworld about the genAI HR project, and offered insights into how the project rolled out and what other companies should look out for when executing their own genAI-based tech plans. The following are excerpts from that interview:

Tell me about your job at DHL. “I’ve been with DHL for 18 years. Over the last five years, I’ve been working in our corporate center, looking after the group topics when it comes to everything from employer, brand, the way we position ourselves in the markets, attracting and recruiting team members, building their skills, building them into internal talents and then, in principle, also offboarding them in a respectful way. Although we don’t do a whole lot of that.”

When did you start the project and what hurdles were you facing? “We started the project around 2019. That’s when all of this tech was quite new. We were only just starting to hear about this cool thing called Machine Learning or AI. They were trying to get their heads around what it meant and whether it would really be useful. So probably for the first 12 months, it was about us learning.

“I think that’s an important part for any team venturing out on new technology — learn before you get distracted by the shiny things, you know, that great sales pitch that many companies can do. So, we did a lot of learning, and we were cleaning up our applicant tracking system landscape at that same time — that’s the sort of back end of the recruiting process. Then at one of our divisions — our Express Division — they had actually done quite a lot of work and found Phenom and they had decided to try using Phenom and see if it worked for them. That was quite successful. So, off the back of their success, we thought, ‘You know what? I think this could work for the whole globe.’

“AI is not going to get rid of our jobs. Our jobs are not going to be taken by AI. Our jobs are going to be taken by people who know how to use AI.”

Meredith Wellard

“We set up what we called our Skydiver project. The reason we called it Skydiver was because it was fast, and it was like a jump out of a plane — kind of risky. It worked. We partnered with Phenom and over a six-month period we got the career site up and running. After six months, I think we had like maybe five languages on the site and 70 career sites [consolidated on it]. After about 18 months, we were down to about 11 career sites [that weren’t on it] and up to 21 languages. As of today, which is three years on, we’re on about 40 languages and we’ve got one career site.”

DHL had 200 career sites. First of all, can you explain that? Why did you have 200 career sites? “So, as you know, we have [about] 600,000 employees, and we’re in basically every country in the world — so, 220 countries. We do operate in a decentralized model, so the power is in the hands of the countries. They need to be really relevant to the local markets. We’re in a service industry, so it’s super important that you know the local environment is reflecting who they are very well when it comes to our recruiting activities.

“We also have a ‘global as necessary, as local as possible’ mindset, and we probably we missed a bit of the global part when we allowed whichever country that wanted to to set up a career site. And a lot of them did. By the time we started to think about partnering with some [of] these great career site tech companies or talent experience platform companies, we realized that we had quite some work to harmonize the landscape and bring together all of those career sites. Some of those sites had a lot of love and care put into creating them on a local level, but they had no [click-throughs] because they were turning up on page 17 of a Google search.

“And so it was quite important for us at the beginning to take those 200 or so career sites and figure out how to bring those together so that the local environments don’t lose what they need in terms of the localization. And it’s way more than language. It’s also things like the way the visuals appear, the way the jobs are presented, how users can click on them.

“We did a lot of research in the early days, looking at what would someone in Denmark expect from a career site versus someone in Japan, and from a user experience point of view those two are so different…culture-wise — even the visuals like the real estate usage expectations in a country like Japan versus Denmark, where it’s rather more sparse, and you know, spacious. It’s a cultural preference.

“So, we had to take care of all of those things and at the same time really take care to communicate and educate our business around how it wouldn’t be a natural thing for an HR professional — who knows everything about HR, legal aspects, recruiting, what a good interview looks like — it wouldn’t naturally be that they should know what it means to create a good SEO. Or how they could become more transparent on Google search. Or what would it take to have the right words, or the right visuals, or the right number of interactions in order to bring yourself up higher on a search result. This is not something that HR people know necessarily. Some do for sure, but not all of them do.”

What other issues were you facing? “I think the biggest one was definitely, because of those disconnected sites, it was definitely our visibility. The second one is around our brand. We have an amazing consumer brand. Everyone knows DHL — the red and the yellow. It’s bright. It’s engaging. It means a lot to every employee, and…our customers and the partners that we work with.

“Unfortunately, whilst the consumer brand is still is incredibly powerful, our employer brand was somewhat ad hoc. You know, people were creating things that they thought look nice that they thought represented. But there was no standardization and if you looked at one brand in in Germany, it might not look the same as one in South America. It just didn’t work. So that was the second big thing. We needed to align our brand to be more visible. As an employer, we want to be this great place to work for everyone. We’re this great company to work for, but how can you ensure everyone knows that if you can’t even tell a coherent story out into the market?

“A third thing is we wanted to take care that we minimized the likelihood of things like the fraudulent use of our brands. You hear these things in the market where companies had fraudsters say, ‘Pay me $50 and I’ll get you a job interview here or there.’ We just wanted to make sure that we weren’t exposed to that, because when you’ve got so many different sites, it does increase the risk of that type of activity. So, with that sort of compliance topic, as a company that really values good business practices like compliance, it is important to us. So that was the other thing that we had to address.

“The fourth thing we had to address was just bringing the recruiting audience together. When we have all of us looking together at our talents and saying, ‘OK, well you know if I’m recruiting out of our supply chain division and I’ve picked my gold medalist, but geez, there was three really strong silver medalists.’ We were just letting them go back into the market. We needed to be directing them go into other potential jobs in our company, not back into the market — especially nowadays.”

What are some of the talent challenges you’re facing, particularly with some of the digital technologies that you’re rolling out. “I don’t think we’re unique in this. There is obviously a huge demand for tech talent across the globe at the moment. And, the demand changes, I feel like it’s changing every six months. We started with everyone wanting blockchain specialists. Then there’s this whole push on project managers and some cloud architects. And they have moved on to the cybersecurity space. Now, most recently, we’ve got the whole generative AI and AI technicians and data analysts and so forth. It’s a constant churn of changing demand, and it’s sort of pulsing.

“I think the initial mode, because it’s what our go-to always has been, is we go to market and say we need this type of person. Of course, because we’re DHL and we’re a logistics company, you [as a job applicant] may wonder why would we need technical people. We realized that our reputation as an employer of technical people — IT-type people — wasn’t that it was a bad reputation. There was literally no reputation. Technical people were not thinking, ‘I can’t wait to get my degree and work for DHL.’

“So we had to really get the story out there why we are a good match for someone who wants to work in that space. We really have that type of technology that is cutting edge in terms of what you do as a technical contributor, it manifests in our business immediately. Whether it’s a proof of concept or a pilot, whether you’re working on Google glasses or Internet of Things, you can actually see how your work impacts customers and the business — and quite quickly. And, I think there are a lot of technical people who want to see the impact of their coding or their design or their project management or whatever on the customer or the business.”

You partnered with Phenom to help you roll out the genAI for your career site, but did you also build an internal team to roll out this technology? What kind of people were on that? “We have a really good internal team. Phenom became our partner, and we learned a lot from them. They had a lot of expertise about this very new technology. Internally, though, we have a really great coherent team working in a sort of agile working mode. That was also new to us, a working mode being in a scrum-based environment. That team’s made up of a set of HR professionals who know recruiting and who know what it means to bring and attract people onto our business but who also had the ability to translate those HR needs into technical language, if you like. They could talk then to another group of people who we worked with internally, IT professionals who were willing to talk to HR and ask them, ‘What does it mean to transform HR into this technical world?’

HR was one of the late adopters of technology. And I think that we found a really great group of IT professionals who were fascinated at how technology manifests to enable our people in a way very different from, say, the operations of finance.”

Did you also include folks from legal, finance, and compliance and others? “We call them our product owners. They’re the ones from the HR and from the business side. They describe what’s needed. They work with our actual operators out in the field, the recruiters and so forth. And they understand what’s required. They talk to IT in the language of IT, and IT says, ‘Here’s what we can provide as a solution.’ Then we say that’s an interesting solution — now let’s talk to compliance about how this will work from a compliance perspective? You know, we can’t be a company that says you can trust us completely unless that trust extends to candidates, applicants and so forth.

“We also talked to the compliance people. Then we have our procurement partners. They’re  talking about how our relationships are set up and how we can get the most out of these really powerful contracts that we have in place across the globe. And then of course there’s finance, which is looking after all those financial things. And then the managers. And the employees — everyone providing some insights from a personal perspective on how it should work for them. What would they want from your product like this? It’s really the whole company coming together.”

So how far down did you go on the management rung to get input? “All the way to our frontline supervisors in terms of testing it, trying it out, and giving us some feedback on how it looks and feels. We involved everyone…who is actually someone who needs to click the buttons to say, ‘Yes, I approve of this.’

“At the same time…, we spent a lot of time on our employer brand. Looking at what our brand looks like if we want consistency. It’s not just having the technology to do it, but you’ve got to have the right attributes. It’s a reflection of who we are internally. So, we talked, interviewed, surveyed, and ran workshops with literally 3,000 or 4,000 employees — maybe even more — to understand what their experience was working for us. What do they love about us? What should we talk about and not talk about? And through that we created this story around where we are as a company, and about how we’re a great team. And, once you’re in here, you bleed red and yellow. That’s the truth.”

Can you explain how your new, consolidated jobs site works to direct job seekers to the right open positions? “The nice thing about some of these new technologies is that AI is part of the experience, but it doesn’t feel somehow tricky. [Employees] will say I use ChatGPT to do this or that, and I’m like, it’s OK that you used it for that, just as long as you didn’t present it as your own work. On the career side, AI makes the whole thing super easy. It’s much simpler as part of the search experience. If I want to search for a job in Vietnam, I start to write ‘Vietnam,’ and the AI immediately recognizes where you’re located. It then triggers a set of jobs that start to come to you in the language that you’re writing in. This is very simple algorithm in the end, but it feels very personalized, very relevant.

“The second thing that happens is that you get the job recommendations. We have a Phenom-branded chatbot called Larry that allows customers to personalize it. He pops up and says, ‘Do you want more information?’ You start to enter in things and Larry will say things like, ‘Would you like to apply for a job or are you just browsing?’ He’ll tackle some of the first compliance questions like ‘Would you give permission for us to record some of your data?’ And as it does this, it starts to understand what sort of jobs the applicant or interested person might be a match for through one answer-type questions.

“Then once you get into the platform and start looking at the jobs, of course, it gets a little more intense — not so much for the candidate themselves, but more so for the recruiter who will immediately start to see people recommended for specific roles if they’ve given their consent for that. The [job seeker] didn’t apply for that, but let’s see if he’s interested, because this could be suitable for him.

“This type of matching AI is very useful for helping us bring to the surface the right people for the right jobs. In the past, it would have required recruiters to go through lists and lists and lists of [resumes] to try and find the right people for the right jobs. So, for me, that’s a super powerful use of the machine learning technology that we just didn’t have access to before.

“And then there’s about a million other things that it does. For example, once you find someone, you can send them an e-mail and the recruiter can click on a little AI helper in that email and choose to send an e-mail to the [job seeker] asking if they’re interested in talking about another opportunity we might have. So, it really makes the job easier.”

So give me an example like before and after, since rolling out the new AI-powered site. So back when you had 200 sites as opposed to now. “Let’s say I’m in South Africa on DHL’s career site. I’m hoping like crazy that someone’s going to find the career site and click on it. Maybe they go there and maybe they see some nice pictures and a statement from some of our employees on how they like the company. Then they might click on a job, and then that might get to a recruiter that information along with all the other people that have clicked on that job. But mostly what happened is that people would put [resumes] in there and probably not hear back, if I’m being honest.

“Now what happens on our career site is South Africa is in bold, and so they still have a very localized site, but it’s fully integrated with the global site. It’s the same global URL; it’s just that the page would have some local references, probably some local visuals, maybe some of the local employees talking about their experience working at some in South Africa with our company. The real difference is that as a candidate, now I go in and I get guided to the right sort of jobs based on my [resume]. The system parses the [resume], finds my skills and tells me there’s some jobs that might suit me. And then the recruiter who’s looking for someone with those skills doesn’t have to search through hundreds and hundreds of [resumes]. They get presented the ones that have the matching skills to the top of the list.

“So recruiters still have the list of 100 [applicants] if they want to look at them, but the ones that are matching what they’re looking for are at the top of the list. And that list [of applicants] could come not just from my unit…, but also includes DHL’s Express supply chain and our Freight Forwarding units in different parts of the country. They get to see it all now, not just that singular entry point.”

What’s the most unexpected benefit from the genAI-assisted, unified career site? “The unexpected things? This last couple of weeks, we’ve discovered that we are the most visible career platform in Germany. That means if you go searching for a job in Germany, you will find DHL before any other site, and we’re very close to being the number one job site in a lot of other countries as well. That is because it takes a good [career] site platform and it takes a good understanding of how you use that platform. We’ve reached the point where those two things have become really congruent. So we have some great people doing wonderful things with the content, continually updating our tags and the language alignment.

“But we also, of course, have a platform that allows us to do that fast and efficiently and we’ve set up a really nice process where we have people who are responsible for administering the content. In Costa Rica, in Malaysia, and in the Netherlands, it’s 24 hours a day those updates are made. So, for example, if the administrator in a particular company comes to us today and says something like, ‘We’ve just heard that XYZ company is closing down and we immediately want to get something onto the career site so that we can bring their employees to our site,’ we have someone somewhere in the world that can get that content on to the system. We didn’t realize that this would even be possible when we brought the product on.

“The expected things? Our reputation is improving, particularly with technical talent and young talent. Our ability to customize and really target content to appeal to certain [talent] audiences…has been quite rewarding.”

In what other ways has DHL used generative AI technology other than the career site? “We have what we call our Career Marketplace, which is an internal marketplace recommending people not only for jobs, but also learning networks and mentoring and these types of experiences. We’re in the middle of rolling that out at the moment. That will also include some very nice AI capabilities.

“We use a generative AI chatbot specifically for creating some of our internal marketing content, so, short videos and…training videos and things like that. The chatbot is called GAIA, which stands for Generative AI & Intelligent Automation.

“Internally, it has been super for those things also, because it doesn’t have to be done by a specialist, it can be done by people who have the right tools now. The company has been very wise and very fast, but at the same time cautious in setting up an internal generative AI tool that our employees can use, and that’s safe, and that’s compliant, and that’s able to be accessed by…everyone. Anyone who is the least bit interested in accessing it can. That has had two benefits: one, it gives people an awareness of what generative AI is and that it’s an actually something valuable if you use it well. And two, it also helps those who may have some fear or concerns understand that it’s not a monster. It is just a tool.

In what ways did DHL train its full-time employees in the use of AI? “We have a group within our company that took a more proactive role in informing and educating [employees], running workshops and webinars and inviting people to information sessions that were a bit more formal than our previous sort of soft rollout. And, there also people who are responsible for using AI or creating algorithms or machine learning or data analysts, who have had very formal education, and they’re using some of our learning partners to provide them with proper education on AI.”

Who are your learning partners? Are they universities in the various countries you’re located or online course providers? “We use Coursera a lot, for example. Also, Skillsoft Percipio. We also do have partnerships with universities…and other places that provide us with super advanced levels of education.”

Can you offer any tips for other organizations who are considering rolling out AI? What should they avoid? “I think they should avoid hysteria. I think what our company has done super well is I have never felt scared of AI. It was explained to us that this technology was just something that’s coming. It’s going to help us. It’s not going to cause problems if you get on board. There’s a quote that we’ve kind of thrown around in our business a bit. I can’t remember who it comes from, but it says something like, ‘AI is not going to get rid of our jobs. Our jobs are not going to be taken by AI. Our jobs are going to be taken by people who know how to use AI.’

“It’s not about replacing your job or being something you should be afraid of. It’s about getting your arms around it and learning how it’s going to help you be better at what you do.”