We talked about the state of AI in recruitment and how Talkpush is constantly re-evaluating the human/bot interface
George LaRocque is the Principal Analyst and Founder of #hrwins reports on HR technology innovation. He has amassed more than 25 years in the field in Talent Acquisition, Talent Management, and HR as a practitioner, HR Technology executive, analyst and consultant.
I have always enjoyed George’s Marketwatch podcast, with a thoughtful analysis of the industry and was delighted to be invited on his show. Below is the podcast transcription, enjoy!
Intro: “From HRWins, I am George Larocque. When it comes to automation, artificial intelligence, and machine learning, things are happening quickly in the world of Talent Acquisition. Maybe faster in that category of HR tech than most other areas. Recruiting is a highly transactional process. We have long considered it inefficient, and at the same time humans have not proven to be consistently good at hiring other humans. So it’s no wonder we have seen a lot of venture capital investment in this area. And a lot of early adoption by employers, a lot of testing and we have done a lot of learning, too, about where this tech can deliver results. Most of what we have seen has been narrowly focused on application of this technology. Chatbots and messaging — that initial candidate engagement and preliminary screening — that screening is very binary. It’s more like programmed logic trees. Do you have this kind of experience? Yes or no. Leading to one path or the other. What data is available? Leading to one action or another. Well, this is an area we have seen the focus in the world of HR. It’s an equivalent of implementing a point solution. It solves a very specific problem. And it’s a very narrow application.
Now when I step back outside of human resources and recruitment, in areas where we are always pushing the boundaries for innovation, places that tied directly to strategic business goals revenue like sales and marketing, I see a lot of effort going into conversational platforms. And conversational platforms allow people to communicate with applications, websites, devices, and natural language — and everyday-human-like natural language. And this takes place via voice, touch, text or gestures. For now in recruitment, we are mainly focused on text and recruiting, but voice is coming out very quickly. These conversational platforms strike common key points that should resonate with recruiting leaders.
- They improve experience, so they enable the person interacting with them to move forward in the process. They handle a non-linear process to figure our world of recruiting; different candidates pursuing the same position may not only start the process from different points, maybe one from a job board, one from referral, another directly to your career site, they go in different directions when they get there. They consume different content, reflect different behaviors. Some may quickly express interest based on their behaviors, others are in more of an exploratory mode. And that leads me to the next point.
- The systems can predict when to engage the company’s resources — in our case that would be a recruiter. I think about having your CRM (Candidate Relationship Management) system and your messaging interface tied together — prompting your recruiters when to engage, and giving them the right method to do so, and the right message: is the email, is it messaging, is it text, is it picking up the phone (something as archaic as that). But doing those things versus putting a wall up between the front end of the process and the actual hiring process. So extending throughout the front end of the recruiting process — enabling both the candidate and the recruiter.
With that said, today’s guest and CEO of a company that has been on its journey of enabling customers with technology, now a conversational platform combining messaging and CRM solution. Listen in — I hope you enjoy the conversation as much as I do!
Talkpush is the conversational candidate relationship management system or CRM. Max Armbruster is the founder and CEO. Now this is not Max’s first HR technology company. It’s his second recruitment software shop. He is also a practitioner — which means he really understands recruiting and its challenges. He has done thousands of interviews and it’s always a pleasure to talk to someone who is on both sides of that relationship/ dynamic.”
George: Welcome, Max!
Max: Thanks, George, for the intro!
George: First off, I mentioned that Talkpush is a conversational CRM but why don’t we start with how Talkpush was created? What brought you to start Talkpush?
Max: It was 5 years ago. At the time I was in the process of selling my first recruitment software company. It was a traditional ATS. And I met a lot of big organizations like banks, telecom operators and companies, I noticed they were keeping the entrance to their career websites a little bit closed. Many of them did not have the ATS and I was selling to them. But they were not interested. When I was digging in to why they were not interested in buying an ATS — it turned out that they didn’t know how to handle the huge amount of volume because we were working in Asia. They were just worried the huge flow of candidates coming would be more than they can handle. And then there were other employers that were more eager to get candidates who were looking for sales and customer service jobs. So, for example, in Asia we have a lot of call centres — they were looking for solutions but not ATSs. They were interested in finding ways in which they could engage candidates — quick and easy — and do a quick first engagement without having the candidates to do too much. The ideal solution for that group of hires would be something like a video-interviewing solution (HireVue). But they couldn’t roll this out because the candidates to do a video interview was asking them too much. Many people did not have the broadband connection or the mobile connection to allow them to do that and they did not download apps. So that was the problem and I was trying to figure out a way where what kind of solution can we build that can handle high-volume, that can do quick engagements, that can be easy for candidates, and that can help to capture their voice and find out who they are in just a few minutes and a few interactions. And the initial prototype we rolled out was the system that would call candidates over the phone and ask them a few questions, record their voice and send that back to the employer — it’s known as an IVR. That was the first product that we built. And the rest is history!
George: I didn’t know about the IVR- sort of like an automated outbound dialer. So there were no humans involved; it was just processing the responses. Is that product still used today?
Max: Some people still are. It’s like 1% of our volume. I was happy when we rolled this out and customers were happy. The total addressable market size was not massive but it was a good business on its own. The question that came coming back was — ‘Oh, you guys work on the phone. That’s not the way you communicate these days. In fact people text. So what do you have for that?’ And I would have my own objection handling. I would say, ‘Well, text is nice, but voice is important because you need to hear people, find out about their communication skills and so on.’ And so we were doing just that until it became possible to build this chatbot interface. So the technology became available for developers like us to do some cool stuff. And that sort of changed in 2016 when Facebook opened its API.
George: Okay, so you’ve been working on these Facebook and chatbot interfaces. And you have seen the adoption by candidates and employers over these years. You told us a little bit about your customers that are high-volume shops. Is that pretty much who this product is best for? Or who would you frame your typical customers?
Max: Our customers are high-volume talent acquisition operations — meaning it’s both employers and staffing firms. We work with big names like Accenture, Walmart, Adecco, Manpower. And we have strong foothold in the call centre industry as well. Typically if there is a situation where we can expect a few thousand candidates to come in and where conversion rates are important — conversion from who get contacted to who gets scheduled for an interview, who gets screened, who gets to meet a candidate face to face. All of these critical steps in the front of the funnel — how can we optimize around that? And then spending a few thousand dollars on software is a quick decision because the ROIs are immediate, if you look at it from an advertising standpoint, how much money am I putting in to get all those leads or from a productivity standpoint — how many recruiters do I need to handle this volume. So these are the two ways we explain our value to the customer and build a business case.
George: So where does the chatbot live? You mentioned Facebook messenger. Where does the product live and what role does the chatbot play in the process?
Max: Chatbots these days, like you said they are 3 years old, they are everywhere. They are probably sitting in your HR. You talk to a bank, you have a chatbot, so it’s important to make those distinctions. And for us we spend a ton of time building these chatbots for different environments. So the way to visualize this is to imagine our product, which is the conversational CRM that sits in the middle to collect conversations from all the different channels. So a high-volume recruitment operation typically rely on job boards, employee referrals, career fairs, campus hiring, and social media. And for each of those platforms we would have an interface which would engage with candidates in a conversational way. So you would have a chatbot that would live on a website, one that would live on your Facebook page, one that would live on SMS potentially. And you would have custom bots that would be specifically designed to handle incoming traffic from the universities. We have built all these experiences (interfaces) for different audiences in a way that is no different than building websites. Each chatbot can be perceived as as exactly that. That said, there is a lot of movement that happens from one channel to the next, so candidates would go from your career websites to Facebook to somewhere else. Our CRM allows you to track that activity.
George: Where does the recruiter come into play? You’ve got the chatbots on all of your properties or your channels as you call them. So at what point does the human jump into the process?
Max: Human always jumps in at some point. One battle that I had to fight in the beginning to reassure people that the user (recruiter) is in control. So we always reinforce that message: recruiters are incharge of the bots, not the other way around. The exact point at which the bot to human interaction moves to human to human interaction will vary. There are different ways people do it. For some employers, they use the bot to ask open-ended questions, like, ‘Tell me about your career aspirations’. So rather than set up some rules to automatically shortlist candidates based on that, they actually take the time to listen to / read every answer. And that’s their prerogative — some people would like to listen to everybody, and some people would automate the whole thing. So we support both. I was mentioning video interviewing before, I have seen large organizations relying on HireVue, for example, that spend dozens or hundreds of hours listening to the recording, simply because you never know what you are going to find. There is a needle in the haystack or a diamond in the rough, really great candidate out thousands of candidates did not fit but somehow shines with his/ her personality comes through — and you decide you want to hire such a person. These types of environment when you have an open mind, open policy on who you might be hiring, usually the human comes in a little bit sooner for environments where you need to have this black and white decisions on who’s in and who’s out. You can automate the screening and you can immediately tell the candidates, ‘Congratulations, you have been shortlisted. Please schedule a time to come to our office or a phone interview’. Then the only time the recruiter jumps in is at the time of person to person interview. There are exceptions to those two cases. For example, if a candidate complains or uses profanity, if you insult the chatbot (the chatbot is very sensitive) and it will shut up. We have an alert system to stop the conversation in such a case. And if you continue the journey of the candidate past the first interview, things like background checking, onboarding, you may have full candidate journey with some segments that are assisted (with the bot) and some segments that are unassisted where you need a human interaction. And you can forsee a world where a greater part of the process will be assisted as the technology gets better. But there will always be segments that would be unassisted where the humans handle everything.
George: It sounds like this technology is really augmenting recruiting resources. So not really about replacement; it’s about shifting them into that critical engagement transaction — right? We all want our recruiters to be engaging with hiring managers and decision makers, coaching them through the hiring process and doing a better job of doing the intake and consulting process in creating openings and requisitions. And we want them to be engaging with candidates when it’s most meaningful. Sounds like that’s the kind of environment you are describing — and I right on that?
Max: The name of our blog is augmentedrecruiter.com. Augmentation is definitely the key driver. I read a lot of science fiction and I love working in recruitment because I feel like this is one area where robots will be the slowest take hold in a way. I have more time to prepare and get better compared to fields like high finance where the majority of work is already automated. Recruitment is still a human segment as long as we are around!
George: Thinking about your CRM aspect of what you are doing, and the data you must be capturing about conversions on these different channels and the kind of messaging and tests you could be able to run on those channels, you must be able to show an employer an incredible picture of how and when and why candidates are converting based on channels? And the kind of communication… whether a human jumps at what point, sooner or later in the process? I am getting ahead of the product is where today?
Max: I think we are a month or two ahead of you.
George: Oh, wow!
Max: In the sense that we just started publishing these bot stats. I’m a bit of a sports fan and I read a lot about the NBA and NFL news. For some reason, I was always drawn to the numbers — the stats — and we have a ranking of the best bots. And have more than 50 of those bots living out there — presenting different brands: each brand with different personality, each bot with a different ecosystem, different channels. And we rank them on stuff like the hot sentiment, how likely is it for the candidate to give a high-rating experience, or how often they say thank you, how many engagements are there for each bot, how long is the conversation. Of course, the first thing you look at is the completion rates, how many could finish the process. I have over 4 or 5 metrics. I am looking at the dashboard as I am speaking to you. And we publish these reports now on a weekly basis. We share them with our customers so that they get an idea of how well is my bot performing as compared to other bots. That’s relatively new for us, but it’s fun stuff!
George: So you’re benchmarking across your customer base with all of the bot data?
George: That’s really cool. That’s really powerful. I do these round-tables and I’ve got these top talent and recruiting leaders, they’re normally large/ middle-sized global enterprise shops attendees. You know they have varying opinions and comfort-level with this technology. While needles are moving quickly, but everybody is getting their heads around as they are not coming from a place that is optimized. They are coming from a place where pieces are optimized — and we still have a lot of work to do. So what I’m getting at is that having that benchmark data would empower me as a leader to make an argument to move to a new channel with a bot or to change the way we are engaging with our talent. It’s hard to argue with the data point and everything else is an emotional argument around the comfort level or what we perceive as the candidate’s comfort level with the bot. I have seen some data speaking of candidates around candidate engagement, and everything that I have seen says folks are not only comfortable with bots, but the one of the biggest issues in candidate experience is the feeling that the loop is never closed and it’s a real black hole. And the bot addresses that? Do you have any insights on candidate sentiment on the bot? What’s the candidate’s mindset around this technology?
Max: Well, of course, they love the fact that they get immediate response, that they can frequently get their questions answered. The quality of the feedback you give to candidates about the status of their application is very much dependent on the quality and the usage of the CRM. If an organization doesn’t know what they are going to do with the candidates, or if the candidates have been put on hold, the bots are not going to be wise for it. So you can only expect so much. We still need excellent human behavior to deliver excellent service. And so, the improvement is a faster processing time and the way we measure the sentiment back to this data point is looking at the words that the candidates are using. And also if they spend a moment to rate the experience — so that this allows us to gradually month over month improve our bot for our every customer. It’s going to be a long journey, but we have a team of conversational designers that is every month incrementally making those improvements. And it kind of feels like traveling back in time 20 years ago when people were making websites for the first time. We have hired designers who have never worked on a chatbot before and we put them on this problem of how do we improve the candidate experience. And every month we come up with new stuff. So yeah, it’s fun! For example: for Facebook messenger, you can have a lot more visuals, GIFs, videos and how you can immediately measure the impact of adding these elements to the conversation, completion rates and engagement scores. And that’s the kind of experimentation is going on — which is making the bots win the war against more boring, static form interfaces.
George: That’s interesting! I have a couple of questions for you before we wrap up. First off — what’s next for Talkpush? Where do you see your product and platform going?
Max: Last week we released our first mobile app to allow the conversation to continue in the hands of the recruiter. This mobile app allows the recruiter to just jump in at any point of time, take over from the conversations happening with the bot, and message the candidates directly. This means that imagine the workflow where you have a candidate who did all the interaction with the bot, comes to your office to do a face-to-face interview with the hiring manager/ recruiter and is 5 minutes late, for example, they can connect that way. The app will also allow conversations to happen through voice. So a live conversation — meaning that recruiters can use the app to make phone calls the way they make with a normal phone call. With all the conversations and recordings being transcribed, allowing for this information to be restored centrally for the recruiters to have an extension of their memory where they can record all their conversations and all the facts. We are continuing our journey of building around the recruiter and giving them the tools so that they can focus on what’s important. And that’s what we are excited about right now! We are helping recruiters enrich the conversations not just on chat but also voice.
George: That’s really interesting because your latest release (your newest product) is pulling recruiters in to the loop, not keeping them out. Your website says you’re getting candidates and recruiters talk to each other — that’s a great intent for your platform and your chatbot! I think a lot of folks get hung up on the chatbot replacing the recruiter and it’s good to hear you talking about those conversions and making those interactions happen.
Max: It just brought to my memory the experience of a candidate, I haven’t been a job seeker for a long time. And I remember when I sent out those resumes, it was frustrating. I spent a lot of time on resumes and sending it to a lot of people. And oftentimes, I just felt like, if only I could get my message out to the right person. I feel I could convince them that this job is for me. And it’s the same problem we are trying to solve but with a different set of technology and tools. We’re trying to make it possible for companies who have normally closed their doors to 90% of candidates, and say open it to everybody. The AI and tech will make sure your recruiters are not wasting their time. They can still potentially talk to that one candidate that stands out that they would have missed on otherwise.
George: That’s good work to basically widen your funnel with a larger net. There is this one business problem where it’s tougher and tougher to attract and retain good people. It’s a numbers problem, record-level global unemployment, and that keeps marching on. And then another business problem focused on diversity and inclusion. One of the first things to address that is to consider more/ all candidates. There are other tools you need to bring in to play during this conversation.
Max: There’s a lot of good work done by Facebook. The jobs platform is really helping to widen/ open the talent pool. One piece of data that talent acquisition professionals are still stuck on and I think is a bit dated is the hire rate. Look at a channel like Facebook, for example, social media in general, and say ‘Well, it doesn’t convert as well as a job board’. And so, I only hire one out of the hundred candidates, for example, just to throw a number. But actually, those numbers don’t really matter. What really matters is the marketing cost per hire — and whether you have a higher quality of hires on the end. The effort to go from a hundred candidates to one hire that can be automated to a very large degree. So I understand why it was an important metric 10 years ago, when you physically had to call every candidate, do the background checking and everything. But because the technology allows you to automate most of that and this shouldn’t be the main driver to decide what your channel mix is. You shouldn’t be scared by big numbers and use the technology to reach out to this new talent.
George: Right! That’s a great point. The cost for that marketing conversion and how that translates…it’s something that is lost when you are focused on that hire data point. Umm Max, we are running out of time, I would love for you to tell everybody where folks can take a closer look at Talkpush and learn more.
Max: Talkpush.com. We’ve got a chatbot running on our career page. We’ve have a got a Facebook page with a chatbot. We’ve got a WhatsApp number with a chatbot, so you can test us out on any of those places. I’ve been showing our new CRM and mobile app on a couple of video podcast shows. So yes, looking forward to talk to your listeners!
George: Cool! Well, thank you so much for all of your time, and all of the transparency about Talkpush and technology. I really appreciate it and I learned a lot!
Max: Thank you! And thanks to HRWins — it’s a great show and I always get inspiration from leading edge technologists on your show. And thank you for your contributions as well!
George: Thank you!