EP 22: Building Sourcing at Scale: How Booking.com Turned Pipelines into Market Advantage

[00:00:05] Kanwar: The brief that I had when I first joined was: we need pipelines. We need to hire at scale. We have some profiles and crafts that we hire for all the time, 24/7, 365, and we want to make sure there’s a machine that powers that.

We’re no longer interviewing all the same candidates that our competitors are interviewing. We’re interviewing candidates that we know are working for companies that work at the scale that we do, that have the same problems and solve the same problems we do.

I want more candidates. I want to know where to get more candidates from. I want to know why I should be getting candidates from there. I want to know what the market looks like so I can be confident that who I am hiring is the best in the market, not just the first available in the market.

The business needs to understand what the value is and what you’re driving. They need to sit with your sourcers and do a live sourcing session, understand what value they’re bringing over and above stack-ranking candidates and selecting the ones that look the best.

AI is going to make poor sourcers obsolete and really good sourcers unstoppable.

[00:01:02] Adriaan: All right, ladies and gentlemen, and welcome to another episode of The Leaders in Talent Podcast. I’m excited that Kanwar Kohli is joining me today. Welcome, Kanwar.

[00:01:14] Kanwar: Thank you. Thanks for having me.

[00:01:15] Adriaan: Yes, so Kanwar spent most of his career first in agency and in an RPO before he joined his first real in-house role at Booking.com to build out Booking’s global sourcing capabilities from scratch.

Currently, Kanwar leads a global team of around 20 sourcers. And last year, his team supported close to a thousand hires, with Booking and the different brands that Booking owns.

Kanwar, I’m super excited to welcome you to the pod, especially because there are so many conflicting messages, at least from my vantage point of view, around sourcing. Is sourcing still relevant? What’s going to be happening with AI? I know with sourcing, but also strategically, when should companies be investing in sourcing, when not. There are so many avenues that we can take this.

And why this is also a topic that’s very close and dear to my heart is that, historically, we started as a niche tech sourcing boutique.

[00:02:21] Kanwar: Yeah.

[00:02:22] Adriaan: And through trial and error, we built a sourcing team ourselves that, at its peak, was 25, close to 30 sourcers as well. And there were so many things that we had to learn and go through in order to get that engine running.

So I’m super curious where we are now, seven years later, to hear from you what that looks like at Booking.com and your experience. Let’s dive in.

So tell me a little bit more: why did Booking decide to build their sourcing function? What was it for you that attracted you? Take me on that journey when you started, and where you are now.

[00:03:02] Kanwar: Coming up to half a decade at Booking. It’s interesting you said seven years, similar timeline.

I was in sourcing before I came into Booking. I was leading a team, an RPO and MSP team, that were sourcers supplying into multiple clients. So probably a similar setup to how you started at Matchr. Embedding yourselves in multiple different clients and driving the value of sourcing.

And I think that’s the question, right? What’s the value of sourcing in the future?

There have been so many iterations of what sourcing is, how it impacts, how it’s different to recruiting, how it impacts TA workflows, but also where the value is. Is it just more candidates? Is it just candidates from competitors?

For me, the answer has always been pretty simple: sourcing is the only part of TA where you control the supply, right? And not just selection.

Over time, it became a craft of its own. I know there are opposing views that sourcing is a step up into recruiting. You start with candidates, then you move on to managing other stakeholders like the business, and then you become a recruiter, right?

For me, it’s a very different craft and has a very different value proposition, which has evolved. But the start of that has always been controlling the supply: where it comes from and why it comes from where it comes from.

Going back to why Booking built a sourcing engine, I can imagine there’s some synergy there, because we were one of your first clients way back before I joined.

[00:05:36] Adriaan: No, that’s so true. Booking was our third client. We originally pitched more of an end-to-end recruitment model, but they said, “We actually need pipeline. We need tech sourcing support.”

And I had never worked in recruitment. I had never worked as a sourcer. I was so new to the game.

While I was trying to hire what I thought were sourcers, and we went through many iterations to find the right profile, at night I was educating myself with SourceCon, Sourcing Summit, and people in the sourcing community to really understand how this works. It’s so much more than just pipelining.

What was it like to come into Booking and build your sourcing function from scratch? For us, the most difficult part was that because we didn’t control the full process, it was hard to show our value. Especially when Booking.com said, “We need you to build pipeline so we can hire more candidates without having to expand our recruiters, because you do the heavy lifting on the sourcing side.”

Because we didn’t control the entire pipeline, it was quite hard to prove our worth because we wouldn’t always see it in the hires that would come through.

[00:07:06] Kanwar: Yeah. Look, that resonates so much, because that was the first year and a half, two years of building sourcing at Booking.

We grew the team pretty fast because of that same need: pipeline, especially in tech sourcing, but actually everywhere else too.

Tech sourcing and pipeline building can be looked at in two different ways. One is sheer volume: you put as much throughput as you can into a machine. But if that machine can’t convert the throughput as fast as you put it in, it isn’t adding value. It can actually create more noise.

So the first year and a half at Booking was really demonstrating the value of: how do we influence the process for passive candidates who we are headhunting from a competitor versus candidates who are applying?

Because Booking.com, now and when I joined and before I joined, does not have a super strong employer brand. We get the same applicants that all the other big orgs get.

I joined during COVID. The talent market was still competitive because everyone was in hypergrowth. All of our competitors, both in industry and geography, were doing the same thing: hiring as fast as they could and as many people as they could.

The brief when I joined was: we need pipelines. We need to hire at scale. We have some profiles and crafts that we hire for all the time, 24/7, 365, and we want to make sure there’s a machine that powers that.

That’s usually the starting point for sourcing, whether you’re an RPO, agency, or in-house. The need is usually driven by volume.

What differentiates what sourcing has become, and why we’ve invested so heavily in Booking, is when you translate that into value. We’re now developing competitive advantage. We’re no longer interviewing all the same candidates that our competitors are interviewing. We’re interviewing candidates that we know are working for companies that work at the scale that we do, that have the same problems and solve the same problems we do.

It’s a marginal gain, but it’s a gain nonetheless. We can find candidates, nurture them, and bring them on the journey into a process because we always hire for a certain set of skills. We don’t have to pull them into the funnel today. We nurture and spend time doing that.

[00:09:53] Adriaan: Wait, and this is what I always find so fascinating because this sounds amazing on paper. We need to nurture these talent pools and we’re going to bring them in because one day in the future they might be able to work.

How do you convince the business of that? If you’re speaking to people that are most likely not going to move in the next two to three to four years, and Finance is looking at the P&L, it’s like: why are we spending so much money? We’re not getting any hires from this.

[00:10:23] Kanwar: Yeah, so here’s the thing that’s really important to making sourcing successful at any point in the journey, but definitely at the beginning: you can’t just come to the table with pipeline, even though the ask is just pipeline.

The ask, when you unpick it, is really: I want more candidates. I want to know where to get more candidates from. I want to know why I should be getting candidates from there. I want to know what the market looks like so I can be confident that who I am hiring is the best in the market, not just the first available in the market.

Building sourcing was coupled with a peer of mine building talent intelligence. Building muscle around: what is the total addressable market for backend engineers in the Netherlands, across Europe, and across the world? At the time, we were one of the biggest relocating companies on earth, back in 2021.

So it was understanding and segmenting our markets and being able to demonstrate to the business: here’s everyone you hired through sourcing. Here’s a smaller subset. Yes, a smaller number of candidates hired, but here’s the pipeline of candidates we’re going after that is now different to what you’d get before. Here’s why it’s different: they work for direct competitors. They don’t just work for smaller organizations who would love to work for Booking.com. We’re directly targeting competitors that are doing the same thing and solving similar problems.

The value is difficult to articulate until you’re a couple of years in. The runway you need to really build, I’ve learned at Booking, is two years to demonstrate real value. It couldn’t be: first six months, where are my hires? It had to be: what is the plan, and how are we increasing both the volume of the pipeline and the conversion we’re getting from the pipeline, and who we’re taking candidates from.

We knew where candidates were going when we were losing them. We didn’t really have insight into where we were getting candidates from. How can we get more from there? Is that the right place to bring talent from all the time? Do we bring too many from one industry or set of organizations versus others? Do we end up creating homogeneity in the talent we bring in because people come from the same places and everyone thinks the same and solves problems in the same way?

Long answer short: you can’t prove the value of sourcing in six months. You need a year or two.

[00:13:06] Adriaan: This is interesting. So you really need the commitment of senior leadership that understands the value of what sourcing brings in the long run.

[00:13:17] Kanwar: And you need to sing about that to everybody who doesn’t understand, not just within HR and TA. The business needs to understand what the value is and what you’re driving. They need to sit with your sourcers and do a live sourcing session, understand what value they’re bringing over and above stack-ranking candidates and selecting the ones that look the best. That’s the difference.

[00:13:39] Adriaan: So then what targets and KPIs do you set for your team, and what targets and KPIs do you get measured on?

[00:13:49] Kanwar: There’s short cycle and long cycle.

Short cycle is building pipelines for priority roles, which depends on the business being able to define what a priority role looks like and what the volume of that looks like.

And landing high-caliber candidates. Offer to hire conversion is a shared responsibility between sourcing, recruiting, and the business. Making sure we’re offering the candidate, in the first instance, the best possible and most attractive package that will draw them away from where they are today. And when I say “package,” I don’t just mean comp. What are we offering about the team they’d be working in, in this big, complex Booking, that will really draw them away?

The rest of the KPIs are focused on quality and building long-cycle value. If we map the market six to twelve months ahead, every quarter we refresh our penetration rate. We need to show the business that our team is penetrating at least the top 30% of the market. Because if we’re not, then we’re not necessarily adding value over and above the market.

We know our application volume is closer to 30% of the market, but we don’t know which segment of the market that is, in terms of who our competitors are and where we should be hiring from. So penetration rates is another KPI.

What we’re getting to now, five years in, is closer to my version of nirvana: we’re oriented around shaping demand, not just reacting to demand.

Here’s what we see in the market. Here are the skills that are emerging. Here are the companies that are shedding certain types of skills, or have really concentrated skills for a long time that we think we could target because their people have been there for three, four, five years on average.

For me, the next step in measuring the team is: how are we influencing forward-looking decisions as opposed to just what’s right in front of us?

It’s not just: “I need to hire a thousand backend engineers. How do I get the best talent?” It’s: do you need a thousand backend engineers, or do you need 500 who can work more efficiently because they’re more adept with AI tools than the other 500?

[00:15:59] Adriaan: Tell me a little bit more about this, because this is interesting. Can you be a bit more pragmatic or practical? I’m trying to wrap my head around it. It sounds great on paper, but I want to understand how it works.

[00:16:14] Kanwar: Because it’s the problem statement we’re solving today.

When a company prioritizes iterating a product, the first thing the business usually thinks about is: how many people do we need to do the work?

One of our company priorities this year is called Accelerate AI. It’s about accelerating not just how we use AI internally, but also what our AI products look like in terms of our offering to customers.

So the question for me becomes: if we have to hire a bunch of backend engineers to build a foundation that we can leverage and build new products on top of, how are we assessing how they use AI in their day-to-day job? And how do we know what good looks like in the first place?

Are they using AI tools to help them code? Are we assessing for that? Or are we discounting candidates as a result of their use of AI in assessment processes or coding interviews?

The current problem we have is: do we need to change the bar for how we hire engineers based on AI skills being a core skill? Or is that a nice-to-have?

If we hire someone with great AI skills as an engineer, but we don’t have the tools and infrastructure to use that at Booking, is that going to help us?

My team is actively looking at who else is focusing on this. The answer is everybody, but who else in our space is focusing on this, and how are they assessing candidates?

How are we learning from alumni from those companies? How are we leveraging candidate conversations?

Because the one thing sourcing has that no other area of TA has is constant candidate conversation. We have the most real-time market intelligence that isn’t a number pulled from Gartner or LinkedIn insights. We hear directly from candidates about what they do and how they use AI, and we can use that to shape strategy.

We know some competitors are using AI more aggressively than we are. How do we make sure we have the infrastructure so that when we attract those candidates, we can talk credibly about how they could use that same tool or methodology at Booking?

When I say shaping demand, I mean influencing the business on what they think they want at the point of planning, not the point of an individual hiring manager opening a req and saying, “I want to hire a backend engineer today.”

What type of engineer do you want? Why? What work do you want to do?

That should be driven by business priorities versus individual team priorities.

We can do this in a company our size. I wonder what it’s like in smaller organizations, or for some of the clients you have that are at a different stage of scaling and growth.

[00:19:30] Adriaan: When does a sourcing function make sense for a company to think about?

[00:19:38] Kanwar: It can be very early stage if you’re going into a market where you have established competitors. That’s an ideal place to start thinking about: how do I build my scale-up?

I want to know who is doing this really well somewhere else, and I want to entice them to come and help me do it here, or build it differently, or give them more room to innovate than they have with more guardrails where they are today.

I don’t think there’s one answer. But the one answer I will give to any company of any size thinking of implementing sourcing is: before you hire your third or fourth sourcer, you need to establish your value in the business.

You don’t want to keep growing just in size, because then you’re only ever going to be seen as driving transactional value. More sourcers means more pipeline because you have more hands on deck. But it doesn’t necessarily mean more insight. You can become an order taker: “Here’s the demand, go find it, give me more.”

Candidly, that’s where the Matchrs of the world can show immediate value. If I’m building it in-house, I’m a cost, so I need to demonstrate the value. The cost of a Matchr is not the same as building capability in-house.

Smaller organizations or embedded partners can fill a gap. If an in-house team is optically seen as “filling a gap,” you never demonstrate value to the business.

[00:21:23] Adriaan: Tell me a little bit more about how you structured your team. If I remember correctly, your sourcers do market mapping, talent intelligence, outreach, interview or first screening, and then schedule them with the recruiter. Is that how responsibilities look?

[00:21:50] Kanwar: It’s evolved a lot. We no longer even schedule with a recruiter for our highest volume crafts, in the areas where we hire the most, and the most technical roles: engineering, site reliability engineering, product.

We do sourcing and screening. Sourcing, outreach, screening. Then we move candidates directly into interview processes.

The first two years were really building the muscle and trust: we know the right candidates, we keep doing this, we keep iterating. Now we put them straight into the process.

Passive candidates are less likely to move if we protract the process and make it long and arduous. They don’t want to speak to everybody before they get an offer.

So we move straight past recruiter screening. The recruiter manages the journey after the business interview stage. We hand over to the recruiter and introduce the candidate to them, but the recruiter is there to help the candidate feel prepared for interviews, feel clear on expectations, and get strong and honest feedback at the end, regardless of outcome.

The recruiter takes the candidate from business interview all the way through to offer or no offer.

That wasn’t the case in the first two years. Then we handed over to a recruiter to do a recruiter phone screen. We looked at conversion from sourcer phone screen to recruiter phone screen to business interview, and we started to say: this is one-to-one. Why are we spending time? Why is the recruiter talking to the candidate before they speak to the hiring team?

For me, that was: now we’re just ticking a box. We’re doing it because it’s part of the process.

Influencing the process was a big win. It also meant our sourcers feel their value. If sourcers don’t feel trusted to speak to the business or deliver to the business, they either veer off into other directions or move into recruiting, even when their best skill is sourcing.

[00:24:06] Adriaan: Talk about skill. What are some of the skills you look for in a sourcer?

[00:24:10] Kanwar: It’s evolved and it continues to, but the core skills I really value are:

Data fluency and storytelling. Being able to take insight, not just macro talent insights, but also your own funnel. You have a hundred candidates in your pipeline. You’re queueing 40, submitting 20. Being able to say more than the numbers, and tell a story that influences a decision or changes strategy with a hiring team or recruiter.

High-conversion outreach. Copywriting is a real skill. Response rates directly impact outcomes.

Deep search logic. It will always be at the heart of it. You have to do more than search a job title or build a huge search with terms you already know. You have to understand adjacent skills, and AI is helping with that.

Consulting and stakeholder management. Setting expectations without putting up blockers. Helping teams feel like you’re solving problems, not just highlighting new ones.

And what’s becoming more important this year is augmenting workflows with AI. Where is the efficiency in elements that don’t require storytelling, personalization, or influencing? Where can we stop spending time so we can spend more time on the higher-value skills?

Embracing AI as part of the skillset is going to be core for sourcers going forward.

[00:26:21] Adriaan: Let’s talk about the forbidden fruit: AI. What’s your philosophy with AI and sourcing, and what do you see happening?

[00:26:35] Kanwar: I see a lot from vendors, from our current tech stack, and from the market. You can’t log into LinkedIn without a TA leader sharing a hot take on AI.

I’m not going to share a hot take. I’ll share my take: there’s no way to look at AI other than as a force multiplier. You can decide how that looks.

For me, AI is going to make poor sourcers obsolete and really good sourcers unstoppable.

It strips away low-value work: data entry, heavy research, batching and sequencing outreach. Humans can spend more time understanding context, understanding strategy, and influencing.

But there’s one thing I really don’t think AI can replace, and I’m happy to be proven wrong: judgment.

We can train it to find red flags and green flags, but there’s a human instinct AI won’t replicate. Being able to use judgment, build trust with a stakeholder, build trust with a candidate, and move something forward beyond binary flags.

For me, all tools come down to: where does this make my team’s job easier, and where does it allow them to spend more time speaking to candidates, speaking to hiring managers, influencing process, keeping conversion high, and keeping candidate and hiring manager experience paramount?

That’s where I see some really good tools and use cases.

[00:28:42] Adriaan: Anything you want to share about what your team is doing?

[00:28:48] Kanwar: Our stack is HireEZ and LinkedIn. Both have introduced AI agents recently, and we’re seeing value.

There isn’t one tool that does everything. It’s fragmented right now. But we’re shifting from tools that just store data to tools that help us figure out what to do next and automate repetitive tasks that are useful but time-consuming.

The agent functionality in LinkedIn and HireEZ has had marginal but meaningful impact from day one. For example, responding to messages from candidates where we know the answers. If we feed it Booking’s hybrid working policy, and a candidate replies with that question, I don’t want a sourcer spending time thinking about that. The agent could do it overnight.

Same for refining a search or opening it up. You can research adjacent skills manually, but the tool can help by learning from the candidates we hire.

I can’t quantify it yet. We’re talking three months of use. But I see impact. It’ll be interesting to revisit in a year.

There are lots of other tools out there. Jury’s still out on some. I spend a lot of time talking to other TA leaders to get insight into what’s working for them, and I’m always happy to share what’s working for me.

[00:31:03] Adriaan: How do you see the role of a sourcer evolve? I had Balazs on the podcast six months ago. You probably know him.

[00:31:18] Kanwar: Yeah.

[00:31:19] Adriaan: He said sourcing as we know it is over. One, the market has changed. Macroeconomic conditions are not huge expansion, so there’s less volume. Two, with AI, market intelligence is so much easier to pull out. With one prompt you can do a lot of work. And with AI being able to do outreach, what is still the role of sourcing?

He shut his sourcing business down and is now full-time in-house at Michael Page. It was interesting to hear his thoughts, because he’s been in sourcing for 20 years.

Do roles in your team change? Are you hiring less or differently? What do you see?

[00:32:19] Kanwar: We’re hiring differently. My team is smaller than it was two years ago, through natural attrition and people moving into different roles. I haven’t continued to grow the team.

One reason is because I believe AI can amplify the value we have as a sourcing function. There is still value in the human element of sourcing.

If you go back to what sourcing was: how do I find an email address? How do I reach a candidate I see on GitHub or Stack Overflow, not LinkedIn, because every recruiter in the world is in that inbox?

Then it was outreach: how do I speak to this candidate so they reply to me, not somebody else?

And today it’s still understanding markets. Knowing where and how to segment your talent market and target the right candidates. You can find a massive list of candidates easily. AI will automate a lot of the searching and sequencing, but that raises the bar.

The new value is: how do you interpret intelligence in the context of your business? How do you use that to influence leaders and move faster? How do you get leaders to start hiring for roles that aren’t open yet because candidates exist who are about to be laid off, or just got laid off, at a company doing great things with AI?

I think sourcers will still exist in five years. They won’t be measured by pipeline volume. There won’t be as many of them. They’ll be better at what they do, and measured by how well they can predict where to get talent from next, because AI will enable that.

AI will give us information quickly. People who use it best will move from reacting to today’s demand to thinking about next year’s demand.

Sourcing will be smaller as a discipline, but so will TA and any role with large elements of automation in the workflow.

I don’t think sourcing will be gone in five years, or three, or ten. It will become different. The difference is: you can find candidates faster, and know the market faster. What are you doing with that, and how are you helping candidates move into roles faster?

These are passive candidates. You still need to spend time helping them understand your EVP and why it would be great to come work here, not just because you have a cool office in the center of Amsterdam.

[00:35:42] Adriaan: It’s so good. So good.

[00:35:42] Kanwar: So true.

[00:35:44] Adriaan: That’s also what I tell people. People still want to talk to people. Of course there are a lot of things AI can do, but you and I don’t want to talk to agents to hear about your experience in sourcing. I want to talk to you to hear what your experience is leading a sourcing function at Booking.com.

[00:36:00] Kanwar: Look, I get the takes from some people, but I also think it’s an easy take to have: all of this is here, it’s taking away the most difficult parts of our job, so why are we valuable?

I believe the human element is the one thing that can’t be replaced, and also the most important to the customers we have, whether it’s candidates or the business.

[00:36:28] Adriaan: That being said, thank you so much for being a guest on the podcast. What’s the best way for our audience to reach out to you? LinkedIn?

[00:36:40] Kanwar: Yeah, you can find me on LinkedIn: Kanwar Kohlind. Happy to connect and always happy to talk and evangelize about sourcing, but also anything else in TA.

[00:36:49] Adriaan: Awesome. Thank you so much, Kanwar.

[00:36:52] Kanwar: Thanks so much, Adriaan. Take care.