How to Unlock the Power of your Data with Andrew McKay
You collect talent data all day every day. You’re sitting on a completely untapped resource. Besides simply tracking your business, what could you be doing with that data? Today we hear from Andrew McKay to learn about what’s possible.
transcript auto generated from episode above
Brad Owens
Welcome back, everybody to another episode of the transform recruiting podcast as always I'm your host Brad Owens and with me today I really enjoy getting in the weeds for recruiting tech I really enjoy. Talking to those that understand how it ticks how it could be improved I understand I I enjoy talking with people that truly understand kind of the behind the scenes the in-depth how you tweak it how you make it better how you can actually use this thing to make some more money and I've got the perfect person for you all to learn from today. So I want to introduce. Mr. Andrew Mckay Andrew welcome to the show. Sir.
00:48.29
Andrew McKay
Well thank you very much brad and with that austicious introduction of me I better actually say some smart things during the next half an hour but yeah
00:55.20
Brad Owens
Well let's just have a fun conversation here. So before we dig in I think it's going to be helpful for everyone to kind of know who you are where you come from your background Can you just hit us with the highlights.
01:05.17
Andrew McKay
Sure I always tell people that I'm a musician who you know grew up in the software industry. My grandmother always told me that you know take your first love and make it your hobby and your second love and make it your career which is exactly what I did otherwise I'd be the same you know. Drunk person playing in the same awful bar that I played in when I was in college but to be honest with you I fell in love with software very early in life and it was the early stages of the industry and I was a predilection towards startups anyway, just because I would overstay my welcome in the larger companies that purchased us. But. I got into relational databases. Um and data warehousing and business intelligence and somewhere along the line around 2001 I was introduced to the idea of a search engine when I say search engine I mean like Google not like search engines and recruiting and I realized that this is actually the technology that should be used.
01:51.62
Brad Owens
Sure.
02:00.68
Andrew McKay
For the data Warehousing world and it wasn't so I said what? ah we we do if we brought the 2 of them together being able to search through structured information. You know records and databases and unstructured information which is just Files and documents. Great example in our context being able to search the candidate records with the resumes the job records with the job description. So that was um, the impetus for me to join fast search and transfer to start the idea there then a bunch of us started a tivio in 2007 and sold that to service Now. Based on that premise and during that process I discovered Salesforce brought it into the company I said I like these guys they're in the Cloud and I think they're going to be huge so I actually want to take my idea and take it to the Salesforce community because I Said. Salesforce is my trial by Fire. It's a it's a traditional relational construct unfortunately predominantly populated with text and that's actually how I built Kona search and during the course of building Kona search in various markets that we got sold in the largest Market. We picked up from the beginning. Was recruiting staffing and executive search as all 3 like to call each other differently and ah, that's been our main market ever since.
03:21.18
Brad Owens
Ah, that's incredible. So ah for that industry the primary use of your tool would be what.
03:30.12
Andrew McKay
So we started out of course providing a search engine. You know I look at a job order I go in and I say okay I'm going to create a query in my search engine and find me the best candidates and we did well and then we said why do you keep doing that. Why don't you just sit in the job order and. Click a button that says find me the best candidates and we introduced our matching engine. You know so we could build that query automatically based on the job and then we said all right? Well um, we're entering the industry and everybody kind of likes what we're offering, but your competitors have 1 thing. You don't have that's a parser. So I said oh that makes sense. We'll create a parser because we were already searching. This is the unique thing about us. We were able to search at a level a multi-oc level above the individual objects. So candidates job history and education history and everything else goes along with it. But we were also able to search the resumes along with that and cover letters and everything else to provide a better relevancy score and the matching engine did the same when we brought the parser out we said well we're going to create one. Ah, it just so happens when we created our parser l lens appeared je a I appeared. So rather than generating haikus written by sixteen year olds about the inside of a ping-pong ball which is what everybody likes to do at chat gpt my 16 year old son I don't think he starts with a blank piece of paper anymore with any of his homework but we said.
04:54.96
Andrew McKay
We're going to use a specialized lm going to stick it entirely behind the firewall because security is a very important thing with pii which is essentially what you have with resumes client data and we're going to use that in order to do fuzzy matching between job descriptions and resumes match them up and see what matches the best. And that's what we started to do when we introduced the l la and we said you know what? this is smart enough to be part of a parser so we founded our parser based on this llm um technology and then we realized that our ip is building edge cases all around that model which is what we've been doing ever since and we started developing the parser which in fact. We released the first version of this week so there you go thank you very much. Well thank you.
05:36.65
Brad Owens
That's so amazing. Yeah congratulations on that one I'm really excited for it for sure. Ah, so you are seeing how the recruiting industry essentially Matches candidates at least how they try to match candidates so in your opinion. What do you feel like is really.
05:48.90
Andrew McKay
Yeah.
05:54.58
Brad Owens
Ah, big challenge in the industry currently like how is it being negatively affected by you know those types of tools.
06:02.51
Andrew McKay
Well one of the things I ah noticed from the grid report I think everybody I think your listeners know the grid report from bullhorn. Yeah, it was interesting. Um, they said that firms that were winning business in 2023 were 3 times likely to use. Ah.
06:07.43
Brad Owens
You should.
06:17.51
Andrew McKay
Self-service technology tools which meant that they still believe that I think that technology is a way for them to differentiate themselves from the competitors and do better business but the other 2 graphs that were intriguing was one that said the top challenge for firms today. Is a ah that they have tight talent pools and they have skill shortages and gaps and that makes sense post covid. We understand that in the industry but the other graph said that of the revenue of the firms that had positive revenue revenue and the firms that had negative revenue they were equal in their. Ah. Implementation and investment in search and match which says to me they believe search and match is the technology they need but it's not really giving them the ah response that they need to fix their talk problems and I think they're right I think search and match. And and technology in many ways. Ah suffers in this industry from a little bit of legacy. It's been around for a while. There's a lot of technical debt and we just happen to be young enough to be able to you know, not carry that around and build on what the latest technologies offer. In terms of what we've learned from the other industries that we've implemented our product in so the challenge today I think is what is the next generation and why is that a challenge today. It isn't because of that gap or the tightening pool so much as your candidate database is no longer your ip.
07:44.51
Andrew McKay
Just isn't It's democratized now everybody has access to everybody all the time. Yeah, there are clever ways to get reach out and get and create new contacts. Um, you can monitor github and you can find out all the engineers and the languages they use and you can extract from the summaries that they write their. Areas of expertise and you can categorize them by topics and all sorts of things but in the end everybody has access to that. So What is it? That's your I P going to be and our argument is that. And I look at it a Salesforce I think start start off with the idea that Salesforce is an excellent foundational platform for any applicate tracking system moving forward in this World. You know what is it 70% of it is already there but it's not even so much about the tech. It's about the delivery and the maintenance and the flexibility and the operations you have.
08:26.29
Brad Owens
Agreed.
08:39.77
Andrew McKay
Ats is now that aren't just hard packaged apps. Oh I'd like feature number 46 well, you'll get it in six months or a year it's very definitely a template oriented more malleable type of solution that I think people are looking for but having said that salesforce is still predominantly. But we would call a clerically-centric data data entry form-based type of report-based system but an Ats is now these next generation of atses are more about flow about actions you know, putting together processes that are malleable and optimized for your. Um, business how you see matching or making money in terms of ah, putting candidates in jobs or jobs to candidates and I think that flow based process the next best action if you will I guess has to be integrated with. Knowledge that is distilled from your data and all the data you connect to and I think that's where the IPHappens it's the interplay of the knowledge and the actions actions are context for distilling the knowledge and the distillation of that knowledge determines what the next action is that interplay.
09:56.46
Brad Owens
Could not agree more could not agree more and that I actually did a speaking engagement on this. That's I'll at least bring up here and when we can dig more into your ideas the the conversation that I was having was around hey your spreadsheets.
09:57.16
Andrew McKay
Is your I p now.
10:14.25
Brad Owens
Your fancy spreadsheet that you keep track of all of your applicants and your warm applicants and everything else. Um and all of your amazing candidates that you've gotten in your database and everything that you built over the past five to 10 years. It's absolute meaningless garbage because all of the.
10:32.69
Brad Owens
Transactions that are occurring that are meaningful that Ai can actually learn from that machine learning can actually learn from they they all occur outside of that system. They are you picking up your phone and calling someone they are you texting someone off of that list. They are you.
10:43.83
Andrew McKay
Um, yeah, yeah.
10:49.90
Brad Owens
Copying that email address put in a Gmail and then sending it out all of those transactions that actually show you not just who is a good fuzzy match from your job description to your resume. But it now also shows you hey this is a person that was a good enough match that we took action on it and we emailed them or we texted them and here's how fast they responded. And here's how many times you've placed them like all of that data. That's what's going to be Someone's I p so completely agree with you.
11:14.58
Andrew McKay
Yeah, there you go, You could think of it this way. It's not the data. That's the I p It's the recruiter's interaction with the data the knowledge created and distilled in context and the action that results from it is the I p and that's what's different between you and your competitor right.
11:21.40
Brad Owens
Yes.
11:30.90
Brad Owens
That's exactly it I see that's exactly I love that you said that see I bring you all people I bring you the people that understand the future of recruiting. So thank you so much even if it's just to agree with me in my own echo chamber I love it. Andrew thank you.
11:34.37
Andrew McKay
Yeah, ah that.
11:45.70
Andrew McKay
Well honestly, it's like you know ah could be fools seldom differ. But I think great minds also think alike so you knows here's the thing we should bring up Ai right? because Ai is is is the big thing now and everybody's talking about where is Ai going and I think.
11:51.70
Brad Owens
Chew sure I'm there. Let's do it.
12:02.61
Andrew McKay
Especially Gen Ai I mean somebody says you know where does Ai fit in recruiting I'd say you're asking the wrong question because Ai is very fundamentally 2 different things. There is learning whether it's deep learning or machine learning and whether that learning is based on ll lens or not um and then there is.
12:21.14
Andrew McKay
Llms as ah, you know for text and analytics or distilling information. So look at it two ways we're using Llms our cells. First of all, they have to be behind your firewall entirely if they're not then your resume somebody's resume just ended up in the public space and this is going to continue until somebody gets sued.
12:38.73
Brad Owens
And.
12:41.59
Andrew McKay
That's going to happen. There's going to be an event horizon for that one? Um, like there has been previously and in other industries but the learning 1 is actually the the more dangerous one long-term I think you think about how data is trained you start with the success criteria. And 6 criterias what showed up in the search results. What got on the short list or hot list. What got it who got interviewed and who won the job so you have control of the first 2 you can make it so that the ah whether you're matching or just straight up searching. The results that came back in the order of relevancy scoring is neutral. It's fair. It's not demographically biased. It's not ageist. It's not sexist racist or anything but the final 2 which the data will train on is who got interviewed and who got the job that isn't you making that decision. That's your. Client making the decision and you don't know but inherent biases are in them but that will trade in your system and if your system keeps coming back with oh just hire a ah white male between the ages of 24 and 34 and you're done. That's not the right answer as we all know.
13:52.80
Brad Owens
Letters.
13:54.16
Andrew McKay
But it's not even the best answer and quite frankly as we move forward in this world. It's not even the economically the best answer either. So you know you've got to watch out for that. So one of the things that we're looking into is what we call Ai on ai. Ah, this gets back to my old thesis on. You know, subjective satisfaction of video games if you're watching the Ai train the system as going down a bias like lane. The other ai is watching it to correct it and push it back to the middle you need both to be honest with you to be safe. And I think that's one of the things that we should all be cognitive of as we bring training into the system now if you're using lllms for not for training. But you're using it for a nice way to be able to really pull information meaning out of blobs of text you can argue that there's some bias in there. But it's really small compared to the one. We just talked about.
14:47.93
Brad Owens
Right? It's just taking the thing like that's a great example of how I've seen people using this in the industry at least the llm um part the generative Ai part is here is a resume here is the job description I'm putting them up against tell me what makes them a good fit for that.
14:50.40
Andrew McKay
Moment ago.
15:04.95
Brad Owens
All that's doing is looking at those 2 sources and coming up with that generative script that generative idea of here's the paragraph that makes them of good fit that I agree. Not so much that you can do to buy that one unless it comes up with fake things in the person's background to show that it's a good fit. There's really not a lot of risk there.
15:05.34
Andrew McKay
Yeah.
15:24.80
Andrew McKay
You'll you'll find people that will get very a particular about this and say even the language itself is biased. Okay, existentially I suppose that's true, but there's nothing we could do about that one? Yeah yeah.
15:24.23
Brad Owens
But.
15:30.63
Brad Owens
Yeah I mean you can get that deeper right? exactly like so controlling what we can control here. Um yeah I like that and I agree I think the having Ai watch Ai is goods I mean there's always that level of There's going to be something that we didn't predict that probably will come up then at least for now we can start doing things to notice it to recognize that any Ai tool we use for a matching thing is going to be risky if it's not also being pleased by something. Yeah.
16:02.80
Andrew McKay
Right? right? exactly but getting back to the original thing that you asked which was you know what is the underlying future. What is the underlying problem and the future I want to get back to this idea of knowledge and action in its interplay. You know to get this is a very abstract idea but to get very pragmatic about it. We can see examples of this I mean imagine putting a flow that has if done else paths in it in your Ats ah, a candidate reacts this way. Um or a conversation between the recruiter and the candidate goes this way or. Ah, get an update for the interview they had over here. Well, where do we put that candidate. What do we do is the next action all of that really depends on the knowledge. We distilled from all of this conversational data. That's the data that counts now not just the resume I mean there are some arguments out there that say we should just just.
16:56.69
Brad Owens
Yes, please Yes, please.
16:57.24
Andrew McKay
Done with resumes altogether you know and I think we all know that that's actually probably a good thing and you know to be honest with you not to be controversial but I would put job boards in that same category with 1 exception approves the rule linkedin because nobody knows what Linkedin is really. Strategy I don't think Microsoft and even knows what it is but it's too good to pass up. There's something there. We haven't quite grabbed on to it yet. But in general I think you get the same ideas that all this information that's out there to present everybody in the current format is already a little speculative to begin with but is what we have. If we can get the conversations in there to also bias that the interactions as you had mentioned earlier between the recruiter and the candidate or the company and and the recruiter or the company and the candidate then we have a real system for the next generation.
17:50.32
Brad Owens
Yeah I agree with that I'd like that a lot I would very much like to solve that one? Um, but that's it I'm about to get on my Ais so box I'm going to leave that one for the next conversation we can always have a part to um so thinking about that future of what the.
17:52.91
Andrew McKay
Yeah.
17:56.81
Andrew McKay
Yeah.
18:09.68
Brad Owens
System could be and what it could be doing if you were talking to a recruiting firm owner right now that's like man I want to make sure that whenever that stuff comes out I am ready to use it I am my system is just revved up. It'll be the perfect location to have some of this ai. What are kind of the first 1 or 2 steps that you feel like someone could take to start setting them up for that inevitable. Um, being able to use Ai on their data.
18:34.97
Andrew McKay
Oh that's a great question. Well the first thing that I think they need to do is that they have to understand what it means to have an applicant tracking system right? I think they need to move towards the more malleable template based type of systems that are out there and I'm going to be biased I'll tell you my business. Been centered around the salesforce community. But I think I can adequately defend the idea that an ats based on salesforce is probably overall the best thing for you to have that's number 1 number 2 is that bias towards an Ats that is focusing more it focuses more on the ah if then else actions. Then it does on just data and matching and I say this as a company that provides a matching engine. But I think you need to do that because then the matching and the search and all the other parts um have far greater value as they interplay and affect the actions. It makes the process more efficient. It makes it more intelligent and I think that's where your ip comes in so we said that before. So I think you need to do is is get your ats ready for that next generation get the right Ats for it or that's number 1 number 2 I think there's not a whole lot. You have to do on the data side look. But it comes to ai. There are a lot of systems out there that think that since ai is bright and shiny and new. It should be in everybody's face and everybody won't mind spending two months training and onloading and offloading data and uploading downloading data and all that sort of onboarding process that frank frankly makes no sense at all.
20:09.68
Andrew McKay
You should be able to get any system that you pick up install it configure it ah and build what you need to make it your own and get it up and running on day one or day two or day eight not day forty and certainly not because ah. I can't use any of the stuff until my data is trained this reminds me and of ah plumbing in the late nineteenth century when it was brought indoors. Ah when it was brought indoors in the city people would say I've made it I've got indoor plumbing so they would put the plumbing on the outside of the walls. So you could sit there and round your living room and show your guests. How successful you are because the plumbing is on the outside of the walls of course who would do that today. Well the plumbing on the outside of the walls is where Ai is in front of the eyeballs of mirror mirror mortals and it shouldn't be right. So that's the other thing too is demand that you still have a system that you know more or less when you get it. There's time to tailor it to your particular needs. But you know the technology underneath like the Ai and everything else is is hidden. It just works I think that's the second thing they do. But other than that.
21:18.46
Brad Owens
Tip.
21:24.48
Andrew McKay
I would say um, there's not a whole lot more. You can do Yeah, we can talk about reaching out to you know the corners of the earth where you can find new candidate information like going to Github. For example, if you if you put out engineers but to be honest with you. You're going to do that anyway and. System itself still needs to be good. So That's what I would say is that those 2 things are probably what you need to prepare for what I would call the next generation of where this business is going I will bring up one last thing. Um.
21:53.90
Brad Owens
Sure. Okay.
21:59.59
Andrew McKay
I always noticed this industry That's a very intray outtray model a job comes up. You find somebody to match it. You got the deal. You win you collect the money to move on to the next one but I take a look at the arc of how investment companies personal investment companies work and they said I want to own you for your whole life. So I'm going to put together a plans that are tactical but they manage you through school. They manage you through your career. They manage your retirement funds. They manage you how you retire what your next ah stage in life is and they manage all the finances all the way through what if you could do that to candidates, especially high value ones. But if you could do it for the job. The company looking for the job. You're sort of doing that now but imagine carrying a candidate all all the way through their career and filling gaps through educational um ah discoveries I'll give you an exact example of a client of ours and I won't mention who they are.
22:55.45
Andrew McKay
They were placing candidates for so long into this one huge company in Europe that they had a long history sitting in Kona about all the different places these candidates went and I wish we had made money on this. We didn't make any money on this part of it and what they did was they said um. Client went to them and said you know what we're getting really high scores of satisfaction working for us. But we have high churn can you help us with that so they ran a whole bunch of clever searches about what candidates were doing over time and they felt that the candidates were leaving once they worked for that company. They would not only leave the company. They would leave the entire industry that they were in and there were other indicators they found as well that gave them the conclusion that they love working there but you're burning them out. They're working too many hours. They reduce the amount of pressure on the on the ah employees. And their churn went down this is looking over time at changes and trends that are going on with that same person from job to job and then looking at the aggregate of it to see what they can do is a high value. Ah service. So that's another one I've noticed.
24:02.91
Brad Owens
That's the value of the data. That's the value of the actual transactions and interactions happening on that data and how they relate to others. It's not people's candidate profiles and their resume like there's so much more to it than that.
24:08.37
Andrew McKay
Exactly yeah.
24:14.37
Andrew McKay
Yeah, the hidden data the data that's in your face. It's it's hidden in plain view because it's there every time you get on the phone with somebody. You just don't use it for anything. Yeah, yeah, exactly.
24:22.80
Brad Owens
Right? Yeah, it's what they do all day every day and that's not what they're putting value on but it should be it Absolutely should be all right? Well this is incredible I Really appreciate you coming on and having this kind of discussion with us because I feel like there's.
24:32.69
Andrew McKay
And.
24:41.49
Brad Owens
Yes, there's a lot happening currently in the market. But oh my gosh. It's not even scratching the surface and I think the one the firms that are going to truly own the future are the ones that can start to think about this now and start to prepare themselves and get their system set up in such a way they could take advantage of that so that they're not trying to catch up to those that really did it. So.
24:45.60
Andrew McKay
Yeah.
25:01.14
Brad Owens
I'm really glad that we at least had this conversation now. Um, let's put together your dream solution then unlimited budget limited resources. What would you create or change for our industry today.
25:15.23
Andrew McKay
Wow you know and I read I think you there was a question you posted to me I think earlier about that one and I looked at that one and I just sort of half of me salivated and the other half of me got worried because you know what could you actually do to build that perfect system. So there's some obvious things in there.
25:26.84
Brad Owens
A few.
25:34.60
Andrew McKay
I start with the idea is that your database is the world's database and the world's database is your database don't ever think. Otherwise you've got to get your company around the idea that your intellectual property is no longer your your contact database and that means because it's all democratized and it's out there. You don't need to start throwing money left right and center at job boards you can if you want if there's good access to get that some of that information as a source but what it is is a source that's all it is the next thing you need to do is to um and I think this is also the other thing I would introduce. Search match parse connect to you know the heterogeneous environments including the job boards out there democratizes the data and gets you that world's database but using that technology in order to automate learn and anti-bias ah the matching that goes between the two sides. And integrating that with the flows and including and using and supporting the data from the conversations that are going on during the process when they're working with your organization all of that together creates a system that will find. Not only the best candidates and the jobs together which is the the thing that makes your money but will also then promote the idea of carrying that person's career and that company's job needs and much in a much longer trail than the transactional one and done type of model that we tend to see today and you'll.
27:05.60
Andrew McKay
Be able to automate that process at a level which doesn't require so much cost per transaction and that's kind of like Nirvana we get it. But um I think that's where we are you know and I think that'll transition the industry to what we used to call headhunters.
27:13.20
Brad Owens
Her her.
27:24.76
Andrew McKay
Something that is actually um I would call them career investors you know and they're working both sides and they're there for the long haul and they make a lot more money doing that.
27:28.49
Brad Owens
Sure sure Yeah, they would Oh it's amazing. Okay, well now you've got everyone super excited and they're gonna want to find out a whole lot more about you and your products and things that you're doing so where can they find you.
27:44.17
Andrew McKay
So I of course I have my Linkedin page and I put out these crazy little caffeine articles every once in a while if you want to sign up for those half the time I'm talking about drinks you can make and and and one hit wonder.
27:55.96
Brad Owens
Perfect.
28:01.84
Andrew McKay
Ah, songs from the 1970 s but most of the time it's talking about Ai and related to the industry but I'd like you guys to all to come to a webinar that it's actually happening March Twentieth Eleven am M Eastern standard time sorry that I'm plugging this but I think you guys will find that that webinar really digs a lot more in detail around what we've been discussing here.
28:19.93
Brad Owens
Yeah, and I'll make sure that no I'll make sure that that webinar is linked below and your Linkedin page is also linked below so we get lots of people there because I think this kind of conversation is absolutely needed for the industry. So Andrew thank you so much for spending time with us. Thanks for your knowledge. Thanks for your expertise and.
28:21.71
Andrew McKay
So If you're interested at all in what we're saying please be. Yeah.. Thanks.
28:37.91
Brad Owens
I will point people in your direction for more all right? Oh absolutely no, please come back. Please come back all right? Everyone that's it for another episode of the transform recruiting podcast if you'd like this and'd like to hear some more. It's pretty simple. It's at transform recruiting dot com. So we'll see you there and hope to catch you on the next one
28:39.40
Andrew McKay
Thanks Brad and look forward to talking to you again continuing the conversation. Thanks well.