Streamlining HR for Startups with Peter van Kersen

Thomas Kohler:
We have Peter here again. Peter and I had a conversation on just how we’re both doing. And you also now fully started running your own business. Right. And you also scaled it a bit. And we talked about one problem that we both still see and face, that HR systems are not talking to each other properly, are not communicating to each other properly. And there are a lot of implications out of it. One, you cannot really rely on the data you have, you cannot really get efficient. You also cannot really use AI at some point maybe if you want to implement it as a business, if the foundation is not there. Right. So there are a lot of implications. We will talk through them and how to anticipate them. But maybe first, Peter, give us an update on, on what’s happening on your side personally or career wise. And then let’s go a bit into the topic.

Peter van Kersen:
Yeah, sure. So my company, WorkingCapital, we partner with generally startups to take over their HR team or to be their HR team. That usually involves some recruiting, it involves some admin, and it involves an HR manager or an HR business partner. And the one thing we see with every company out there that’s just getting started, that probably raised some money and is ready to scale is that they struggle with the HR tools. They don’t have the time to invest in researching what the best tool is for them. So they generally go with an ATS that a quick Google search has shown them is the best in class or best in the market and an H R I S that will fit their, their scale and growth plans. And the issues that we then see is that these systems don’t talk to each other. And once they start using the H R I S and it turns out that the engagement module is lacking, it’s not quite what they need.

They get an engagement tool and they want to add performance management, so they add a performance management tool and before you know it you have five or six or seven different tools that all don’t talk to each other or do so on a very limited basis. So I think we’re uniquely positioned to recognize these issues. And now we’ve also tried to start to fix them, but the reality is that there’s still nothing out there that really fits the bill to make this the ecosystem that we all want.

Thomas Kohler:
And what are maybe some, what is the company stage? You usually see this problem happening.

Peter van Kersen:
So we generally work with companies that are pre series A or just post series a, so around 30 employees, up to 300 or even larger. So about to go on the journey of establishing a people function. But not quite established yet.

Thomas Kohler:
And what systems do they usually already have in place or want to implement, and what’s the order?

Peter van Kersen:
Yeah, it’s always the same. It generally starts with an ATS. Once a company reaches 30 employees, they’re still doing most of their HR in Google sheets or in Excel, or they have an accountant that also manages payroll and has most of the employee data. So then on the minds of most CEOs is hiring. So they get a recruiter and an ATS, and then later on they figure out that it will be very useful to have all the employee data in one place and they invest in an HR ATS.

Thomas Kohler:
And when is the performance management tools or the engagement tools, when are they kicking in?

Peter van Kersen:
Generally later. Which is strange to me because it’s not like after you’ve reached 100 employees or 200 employees, and all of a sudden performance or engagement become important. I think they’re important regardless of what stage the company is in, because you still have employees and they still need to know themselves and they still need to grow. But generally we see it as a next phase in the progression. So let’s say around 100 employees, they’ll start thinking about sending out an engagement survey or doing performance reviews.

Thomas Kohler:
Okay. And then you have four systems.

Peter van Kersen:
Yeah. Well, let’s imagine that we’re a tech company and we also want to test some of the developers that we interview. Now we have a test tool that we hook on to the ATS or. Well, we want to send out contracts that we want an E signature for. Now we have an E Signature tool, and you can see how this quickly starts to add up.

Thomas Kohler:
Or you need to source candidates, and then you have a sourcing tool, or you need to schedule interviews and then you have a scheduling tool.

Peter van Kersen:
Exactly.

Thomas Kohler:
Okay. And what limitations do you see when you have different tools decentralized, that actually contribute to one process or to one experience, maybe.

Peter van Kersen:
Yeah. So the main limitation we see is the way data moves around this ecosystem and what the source of truth is. So I remember the last time we spoke, we spoke about workforce planning and where roles originate. We can go back to that example. So if you have a workforce planning tool or if you have the need for a position, you can, for example. Yeah. In an Excel, there’s certain characteristics about that position that then get transferred into the ATS if you’re lucky, and will also need to reach the HRS at some point.

Thomas Kohler:
What could that be? Let’s say we are coming from a budget. Yeah. Because usually the headcount Plan is done in a budget. And then the budget says, okay, a certain number of positions need to start at a certain period of time at a certain amount of costs. What’s often not getting transferred into the ATS is what should be the start date, what should then be the hiring date. So offer signed date, for instance. And then when do we actually need to kick off the role? And what activities do we need to plan from a recruiting perspective in order to get the hire done and the start date filled? Right. I think these are the basic things that are in the budget that are not transferred to the recruiting system. Just we got approval for this role, hire for it. That’s it, right?

Peter van Kersen:
Yeah.

Thomas Kohler:
And then what’s happening then? Let’s say the hire is done usually maybe delayed, a few were maybe not delayed, but let’s say 2 3rd always delayed because planning is not accurate or not aligned. And then it’s the hire is done. What’s happening then? Usually it’s a limitation. You send out the DocuSign contract or whatever tool you use, and then the hire is marked in Ashby, Greenhouse, Lever, Smart Recruiters, whatever you want to use. And what’s happening then?

Peter van Kersen:
Yeah, yeah, exactly. So now we’ve reached this point where there are. I always try to think of this as packages of data. So the first package of data comes from an Excel sheet that we call Workforce Planning. And it has the team, the department, the division, the salary range, the name of the role. Maybe the name of the role is different as the job title. So these are kind of objective data that originated somewhere. At some point, a person is going to fill this role, and they have a name and a date of birth and an address and all things that are unique to this person.

But we need this in order to send out an offer letter. Most of this information we already get from the applicant once they’ve applied or when the recruiter has put it into the system. But after the offer has been accepted, we also need to create a contract. They need to have hardware. So we need to know what kind of hardware they want. We need the actual start date, because it’s probably different. And then we need a bunch of information like in Germany, Social Security number, the tax id, how many kids you have, the birth dates of your kids. And this, let’s call it package number three, is all the stuff that we need to not just create a contract, but also enroll someone with Social Security.

And in Germany, this is quite complex, but you should see how it is in Brazil, for example, where we also need to know if Someone is a military veteran, what their blood type is, etc. So this package number three gets extremely complex and this is something that ATSS and HRSS have not figured out yet. So how to gather this in a structured way so that it takes the minimum amount of effort and manual work for an HR manager to gather this and process it. So this is just one of the frustrations of how information doesn’t move around in a smooth way for them to be used by the HR operations team.

Thomas Kohler:
In a not even big company.

Peter van Kersen:
Exactly. Yeah. It doesn’t matter how big you are as a company, as a startup in Germany, we still need to gather this information. And I’ve seen so many teams do this via email and then copy paste all the information into a system.

Thomas Kohler:
And how do you solve this now?

Peter van Kersen:
So, unfortunately, through a workaround. One of the latest clients I work with had the Greenhouse bamboohr setup, which is very common. Bamboo and Greenhouse have a native integration, but it doesn’t cover for any of the things that we just discussed in package number three. So what we do is we’ve written a script that gathers the, the data from Greenhouse, which should remain constant because it’s, it’s, it’s. Let’s call it package number one. It’s everything that was in the workforce planning sheet, the team, the department, et cetera. Then we gather package number two, which is all the candidate information that they’ve already given when they applied, and then we send them a form for package number three. So all of the information that we need to get them a contract and get them enrolled with Social Security.

And once. So we have that all scripted together in a Google sheet and from there we can start doing things with Zapier or make.com or N8N to transform that data into the documents that we actually need and to get it into the system. So from a common database like Airtable, it’s relatively simple now with these types of tools to create a package of data that consists of everything we just discussed and to then upload it into your HRIs.

Thomas Kohler:
Okay. And then you build integrations or custom, how do you call it, Like API calls or push pull requests that take data, gather data, and then write it into your sheet, and then from there it transfers it to the right tool with the right information. And this is all automated.

Peter van Kersen:
Exactly. Yeah. There’s a few different ways to do it. One is to create a central database that has all of the information. Another is to create webhooks in one of your systems that then trigger the other system to Go and grab whatever information they need, which is a little cleaner with, but either works. The benefit of having a central database that has all this information is that you can then also do other stuff with it. So for example, in Germany we need not just a contract but also a few other documents that need to be created and signed. When someone joins a company. This can all then be created outside of the HRIs or if the HRIS doesn’t have this capability.

Thomas Kohler:
Yeah, interesting. And you could then also do a lot of analysis with it. Right. Without just processing work steps from I would say a default perspective. So what are the things that are necessary that you have to do in order to hire somebody or prepare everything for payroll? What else can you do with the data? I think you can also do a lot of reports insights. Right. Because I think the whole people analytics function or idea is also based on data literacy.

Peter van Kersen:
It is. I’m just so I think that all the insights that you could. Yeah. All the useful insights are probably, you probably wouldn’t want to use this data set for. Because it’s for two reasons. One is it’s literally the stuff that’s that you wouldn’t be able to draw much information from what’s your name, address, location, et cetera.

Thomas Kohler:
Yeah.

Peter van Kersen:
And the second is it’s going to end up in the HRIS anyway. So if you’re going to get any type of insights, I would rather do it from the HRIs than from the raw data set that serves as an in between to match everything. I think what is very interesting is something that you could do later on and it’s part of the problem. And let me elaborate a little bit. So where insights become useful is if we start to get to know you as an employee a little bit further down the line. Actually that’s not true because we also have information from the recruitment system which we almost never use. Right. It’s often gathered.

We spend hours interviewing you, making notes, filling scorecards and then we kind of forget about you at the three month mark or the six month mark or after your first performance review. Right. It’s. It’s. Most companies will not look at the data that we’ve gathered during the recruitment process when they do your first performance review. Which is crazy. And I. So how we could do that or how we could make use of all of those sets of data is to gather it in yet a third data source.

Ideally that would be the hrs, but I don’t think there’s an HRS out there that now gathers all this information about Your current performance, how many sales you make, what your engagement is, et cetera. So I know a lot of companies do that either in an analytics tool or in a data warehouse and then build something on top with a tableau or power bi et cetera.

Thomas Kohler:
That’s interesting because I also experiment a lot with this, especially in the recruiting side. So there are several aspects what I look at. First, what is the macro information and then I would say also the more detailed information. So macro information for me would be, let’s say a company wants to enter Germany, then they start building up a sales function and then we need to first understand, okay, what’s maybe their revenues, what is their ideal customer profile, what’s the annual contract value they’re selling, what is the sales cycle in other regions, what maybe customers do they already have, what infrastructure do they already have, in what period or phase are there in terms of product market fit? Is it something they test new or is it just something they try to scale and already have experience with? This is the macro meta information, right? And then we can filter based on candidates we already interviewed, what are candidates expecting and what is necessary in order of recruiter input. So how many reach outs, how many TA screenings calls are necessary in order to hire somebody from a 50 million ARR company that has sales targets of 800 to 1.2 million on annual sales per year for a seller where the product is easy to sell? Just from a data perspective, in terms of the recruit, the sellers we interview say it’s 4 out of 5 easy to sell this product versus maybe an early stage company would an a seller would answer it’s one out of five because the company doesn’t, the product doesn’t exist yet, right? So you cannot compare this. Plus also what is the average target attainment within the team within the company if 100% of the sellers reach a target. Wow, great. But if we interview sellers from companies that just reach 60% of the targets, we might can also understand, okay, why is it maybe company maturity? Is it just that they are not set up with a good sales management? Do they have maybe too much sellers and too small territories and so on? And these are all information what I I try to gather and mix up from a macro and micro perspective, right, that we just interview.

And you can now have transcripts or not transcripts. You can have note takers with a certain template of questions you ask and then when you get the answer, you get the transcript and the transcript can be transferred with an API towards a database and then you can filter for what is actually, what are the recruiters talking to? Actually, what data are they generating? Right. And then how can you filter that? Because I cannot just filter transcripts. I need to manipulate the data in a way that it’s filterable. Right.

And that you can do requests from it. So this is super interesting. And then the second piece is the hires we made, for instance, for companies. We tracked them down and then ultimately said, okay, how many interviews on each step were needed in order to make this higher? After 6 to 12 months, is the person still there? Did they maybe get a promotion? Then also trying to check in and say, hey, do you reach your targets? How is it. And we are still in the beginning of doing this because this just takes time, right? You need to do. We probably made around 1,500 hires for customers over the past four years. And of course, you don’t have to remember everything, because I just started roughly a year ago doing this and so on. And then it just takes time building up this data. But once we have this data, you can enrich it with so much matter information that this is so valuable. But the foundation is that the data is correct and that you just can make sure that you also are gathering data in a reliable way all the time. Right?

Peter van Kersen:
Yeah, yeah, yeah. I think this is fascinating, and I think this is going to be one of the big breakthroughs in recruitment and in HR in general in the next four or five years. And I think what. So you’ve said a few things that are extremely interesting to me. And one is it’s very simple. You cannot remember everything. So by transcribing every single interview and having an AI go through it and synthesize it and find common themes.

Thomas Kohler:
And put it in a database format.

Peter van Kersen:
Put it in a database and have it analyzed already saves probably thousands of hours of a human going through conversations and finding patterns. The AI is fantastic at finding patterns, much better than any human could. And now that we have those things, both on a meta level, like you said, what kind of trends do we see? But also on an individual level, how do we get to know ourselves and the people that we work with? How do we get to know our sales people, and how do we make sure that they develop throughout their life cycle? And better yet, so let’s imagine that you could call this an agent. You have something transcribe all of your interviews, put it in a database, give it context, and then deliver an output. Imagine that that agent could talk to the agent that goes through all the performance reviews six months from now. So Rather than having a quality of higher metric that is, is this person still here? Yes or no. After 12 months you could have the actual sales data, what type of how they do conversations, how they do sales, and what makes them better at this than the next person. And can we correlate that with the information that we got from the hiring agent? And I think then we have an incredibly powerful system that not just assesses who your best salespeople are, but why and where they may need some training or guidance or just need to be put on a different product to sell because their style is, is not what we need in this company.

Thomas Kohler:
Peter, you actually just described what I’m trying to build, right? So I show you the concept here. So you have the pre hiring hiring and post hiring stage and then you get maybe data from headcount planning and budgeting, you get data from kickoff meetings. It’s a document, whatever it’s notion, it’s word, it’s stocks, it doesn’t matter. Then you reach out, it’s LinkedIn, it’s GitHub, it’s email, it’s even WhatsApp, it’s sometimes cold calls, whatever it is. But you can somehow now track and connect the systems that then streamline the data. Then you have the ats. Then you have the let’s say interview data from. I just now put some few examples in metaview.

There are also other tools, right, where you can just record calls and transcribe. Then you have calendars data. How many interviews does a recruiter or a hiring manager have in their calendar? What other meetings do they have and do they prioritize hiring? Right? This could be displayed by this because it’s often a problem. Then also what’s then in the offer letter is and, and what, what did the candidate say in the interview? What is then in the ATS and what is then the offer letter? Ultimately, right? How many candidates are gambling salary expectations? This is what you could read from this. And then also when the hire is done, satisfaction metrics like pecan or also performance metric from Caltramp or whatever it is, right? And then what I leave left blank is the quality of hire. And I just put in already some mock up data. So this is not real data. What you could display, for instance you click on AI software engineer or a sales manager and then you have one company I put in tux fix there because I just showed it to them how this could look like and I worked with them and then they could see for instance what are their current metrics because you could Just plug in the technology with all the tools, connect the systems, and then we could then also say, okay, what other customers that are similar, for instance to this company, Taxfix, let’s say SiriusD, 500 employees, certain amount of revenue, a certain amount of markets.

And then you benchmark this with also maybe other fintechs or with other companies at SiriusD or whatever you want to benchmark it against. And then you get a benchmark and then you could see, okay, is it better or worse than benchmark? And this is all just fictional data, right? So we did not use this yet, but this is what I’m experimenting with. And I’m starting step by step with using interview data and then generating actually benchmarks that showcase, I would say, more sales management data. What maybe a VP sales or a CRO or a CEO want to see, Right. What is the collaboration between sales performance and candidates we hire from? Right, Because I think that’s currently the most relevant and the easiest to do because we own a lot of this data, because we generate a lot of interviews and we do a lot of things on our systems. Right. I don’t have access to the HR systems of our customers and so on and so forth and to the budget plans. So step five, right. But the concept and the idea is actually also similar to what you’re thinking. I’m just tackling it from a recruiting perspective.

Peter van Kersen:
Yeah, yeah, yeah. And it makes complete sense. And I can imagine why that’s, that’s the conclusion that you would like to draw, right? Use this model because the, I think you called it, the total value of a placement is much higher, total lifetime placement than you would doing it yourself. And I think this is a fascinating way of thinking about it. But what interests me much more as a, as a, as a CPO or someone working with setting up people department is how can we make, how can we get to know people and make sure that they’re in the right place? And one of the issues I see is that we still do things. So the biggest issue is the missing data. So when you hire someone, they’ve already gone through a selection funnel and we already have certain expectations of them, otherwise we wouldn’t hire them. So if we want to do a proper A B test, we should just hire a, anyone, everyone who applies and then see what the performance metrics are.

So what I guess I’m trying to say is everyone we gather performance data on at some point during their career is already skewed because we already had certain expectations. So we experimented a few years ago with giving people a personality test, personality and aptitude tests during the interview process. And what happened is we only started hiring people who scored highly on aptitude for a certain role and who we thought whose personalities would match best with a sales or customer success role. But the issue was that we only selected for people who already scored highly in those things. So the spread that we had measuring performance was extremely small. And we couldn’t really, we couldn’t really get any useful data out of this. And I think this is where an AI that can go through millions of data points, or at least thousands of data points, will be able to help and predict a lot better which people will be successful in a role and ultimately will be happier in a role. Because let’s face it, it’s difficult to be happy in a role if you’re not successful. And yeah, and I think thinking about the problem like the way you just showed is going to be a massive step in the right direction.

Thomas Kohler:
So how can startups that don’t have the right people function or infrastructure yet reach out to you? From C to serious C. How do they reach you best?

Peter van Kersen:
LinkedIn is easy. Or send me an email peter@workingcapitalou.com but if you connect with me on LinkedIn and shoot me a message, that’s the easiest. Yes.

Thomas Kohler:
Then we also link your LinkedIn in the show notes and you will also be tagged on a LinkedIn video. So perfect. Great. Thanks.

Peter van Kersen:
No, thank you. Always a pleasure to talk to.