This interview with McKay Allen, the Director of Content and Communications at Convirza, is part of our Webinar Series. A transcript of the Q&A portion is available below.
Mike: Thanks, McKay. So actually one of the questions that I had that just popped up when you were talking about lead scoring there, I’ve seen this in the lead generation space where they brought in lead scoring and then it almost became too good, where it actually killed some of the volume and some of the profitability for different publishers and advertisers in the space. Have you guys run into that at all where the advertiser ends up not wanting very many of the calls at all, and so it’s unprofitable for the publisher to run the campaign?
McKay: No. I guess there’s two answers to that. The overarching answer is no. On the whole, the vast majority of people find this beneficial. It is the publishers and the affiliates that are producing garbage that are scared of it, naturally. So we literally were talking to an agent the other day and they said, “Well, I think this gives our clients too much visibility into what we’re doing.”
And I guess that if you’re frightened of your quality, then yeah, I guess it’s probably a negative for you. But if you’re confident . . . And really what we’re seeing is on the buyers’ side, the lead buyers’ side, we’re having advertises say to the people down the pipeline, “You don’t have a choice, we want this data. So if we, the large for-profit EDU or we the large advertiser are going to buy calls from you, you have to provide this data to us, otherwise we’re not going to buy the calls.”
So it’s becoming more of, I don’t want to say mandatory, but it’s been coming from the top down, the people who actually spend the money to make the whole system work, who buy the calls, they’re the ones saying, “We need this data to make better decisions. We’re spending too much money.” So I think you’re probably right to a point, Mike. I think there’s probably some people down the pipeline who are like, “Oh, I’m scared of this.” But as the end of the day, the people that are spending the money are the ones who want it.
Mike: Yup, for sure. Okay, cool. Here’s another one, kind of in the same realm. What if advertisers do not want calls recorded or monitored based on sensitive information that can be exchanged between the consumer and advertiser?
McKay: Good question. So there’s two answers to that, and generally we deal with this in the legal space, and of course, in the medical space, and sometimes in the financial services space. We have a lot of mortgage guys that are our clients and a lot of insurance guys. So we can do two things. Number one is we have a security and roles-based system so we meet all the compliance and federal regulations both in the U.S. and Canada that’s for both call recording as well as security of call recordings. So only certain people can access the information. It’s very gradated in terms of hierarchical access to the calls.
And then the second thing, you can block access to everybody so literally no one could even access the call recordings. You’d just get the raw data out the backend. The other thing that you could do is you can actually programmatically delete calls from the system if you wanted to record them and give people access for a period of time. So there’s several ways to work around that, and I can understand that concern. We get that concern, I don’t want to say frequently, but a decent amount of time.
Mike: Yeah, I can see that coming up.
McKay: But there are ways to easily overcome that.
Mike: Yeah. Okay, another one here is about voicemail. Can you detect for a voicemail?
McKay: Yes, we can. And that is actually another example that I probably should have included on that problems of call duration slide. Yeah, voicemail is something we can detect and we don’t include voicemail in our overall analysis. So like that 5,000 calls that I showed you this data on a moment ago, we don’t include voicemails in that, so like in this data, for example. So yes, we can detect voicemails and we ensure we’ve built our algorithms in such a way that it does not skew the data that you’re seeing within your system.
Mike: Right. And so with something like that, for RingPartner, a lot of times we deal with multiple buyers, so we’ll have a single campaign that multiple buyers are buying those leads. Is there a way to send those calls back or once it’s hit the voicemail, that’s it, it’s gone?
McKay: When you say send them back, tell me what you mean exactly, Mike, because I’m not sure.
Mike: Send them back into the system to go to a different advertiser or go to a different buyer for better monetization, basically.
McKay: You could actually programmatically probably build it. We don’t have anything built like that out of the box, I guess I should say. But you certainly could with that recording or with that data, programmatically send it back into whatever system you want to send it to. But it’s not something that’s out of the box and in the system.
You could create custom indicators, we call them, that would more clearly delineate the voicemails as well, which would allow you to make better decisions. But yeah, you can clearly delineate them now, and you can’t send them back programmatically. It’s just not something that’s out of the box.
Mike: Right. Okay, I got another one here just about spam calls. You mentioned a couple of examples. Do you have any others that you could share? I’m just trying to get an idea of what types of spam calls that are out there.
McKay: Yeah, you would probably know more about spam calls than I do. All I know is this, that we have clients that are calling us. And so with our new platform we actually have developed, because it was such a problem, we were hearing from so many clients, we have actually developed a series of spam call filters.
Everything from blacklisting numbers, caller ID numbers that come in because what we find is that a lot of spam callers just use the same robo-call numbers over and over and over again, so we block those numbers programmatically and you solve a fairly large percentage of the problem. So blocking numbers is a big part, blacklisting those numbers.
But then also with the percentage of silence factor, sometimes you’ll have those calls and they’ll just stay connected and nothing is said. And so the percentage of silence indicator that we track is a big part of that as well. And then the other factor is also lead quality. So simply treating calls that have a lead score below 10 or 20 as a spam call is another way to track it. So there’s a lot of ways to do it there.
Mike: Right, so grouping them as . . . ?
McKay: It’s better if you answer that. Like I said, you probably know more about that than I do, frankly.
Mike: Yeah. Well, the big one is the robo-calls obviously and then, like I mentioned, is the incentivization. I can’t, right now, think of many others, but you end up grouping those as poor quality.
McKay: And when you say incentivization, what do you mean? Or when you say incentivization, you mean somebody trying to game the system and make money they shouldn’t be trying to make?
Mike: Basically, yeah. So saying, “Call up and I’ll give you five bucks if you stay on the line for over two minutes,” which is when the campaign pays out. So that’s a big problem. And in some cases, it works with a soft incentive where you can get so many coins for some app or game or something, but in the end, it doesn’t end up working out well for anybody. So it’s something at RingPartner that we just avoid.
And this is from my own curiosity. You’re scanning for keywords within conversations, correct? I don’t know if you dug into that in much detail, but that’s essentially what you’re doing is listen in to the conversations and pulling out keywords, like did somebody say, “credit card” or “purchase” or that sort of stuff, correct?
McKay: Yeah, that’ one of the things that’s based off of. We look for certain keywords in sequence, and so yeah. The easiest way to explain it is probably the example of someone paying with a credit card. So you hear the phrase, “credit card,” you hear a series of numbers in sequence and then you hear a confirmation number going back to them with a series of numbers in sequence. So we look for certain words in order.
The way I explained the analysis too is let’s say my wife calls me and says, “Hey can you grab some milk on the way home from work?” And then she talks about her day for a few minutes and then we hang up. I won’t be able to give you a word-for-word verbatim transcript of that phone call, but I will get the gist of the call, hopefully, or she’ll be angry at me, that I needed to stop and get milk on the way home.
So that’s the way our system works. We’re accurate well over 90%-93% on a per-call basis in terms of determining the gist of the phone call. What happened on the call –
that’s our goal. We don’t care if we get every single word right, we just want to know what happened. And that’s what we’re really, really good at.
Mike: I think that’s a really interesting analogy. That’s something I can really understand is you might not listen to the whole conversation, but you pull out all the important parts, and you can take the action that’s required, and hopefully you didn’t forget anything. So I can relate to that as well. Okay, I’ve got one more question here and them I’ve just got a couple of notes to bring up. Do you have the ability to detect the same person, the same voice calling multiple times and flag it as fraud or potential fraud?
McKay: Not from a voice perspective, but from the phone numbers perspective, yes. So no, not from the voice, it wouldn’t be able to tell the difference between two people. But it would be able, obviously, to marry the specific phone number to the numbers that you could associate as spam.
And the other thing you could do is you would easily be able to . . . I mean all you’d have to do is pull the data out onto a spread sheet and say, “Okay, we’ve gotten 20 phone calls from this number that have a lead score under 10. Well, obviously that’s a spam caller and you can easily go in and manually blacklist that number, so yeah.
Mike: Yeah, right. So yeah, it’s low-quality, that’s really what it boils down to is whether it’s a spam call or a great call, it’s high-quality or low-quality. So it’s not necessarily whether it’s specifically spam or fraud or incentive, it’s just low-quality and we don’t want those calls so let’s find a way to block them.
Mike: Okay. That sounds pretty interesting. So I had a couple notes as you were presenting and one of the things that we do is set all of our campaigns for a connect duration versus just a raw duration. So there’s some challenges around that, publishers don’t necessarily understand what that means. So basically we’re not starting the clock until they’re through the IVR that’s on our side, so we may ask them a few questions before we actually send them on to the advertiser, versus a raw duration which would start right when the call started.
And so this overlaps as an ongoing issue with Google AdWords. So if someone’s running a campaign and it pays out on a connect-duration, which, say, it pays out after 90 seconds, you’d probably set that up in AdWords as 90 seconds to track the duration of the call but you forgot about the 10 to maybe 20 seconds that’s in the IVR so the actual duration is a little bit longer. So that’s an ongoing issue that there isn’t currently an easy solution with AdWords anyway to correctly track the performance of the campaign right on AdWords.
So It sounds like you guys have a solution with using some of the other tools like Kenshoo and things. But that’s something that’s an ongoing issue. There’s no question there, it just got me thinking about duration and how to do a better job of it, and especially when it comes to buying any type of ad space, any type of paying for clicks or calls or whatever it is.
McKay: Yeah, there’s a ton of it, especially when you’re talking about calls generated via Goggle page search. And there’s a lot of use cases in terms of how you could use this data, absolutely.
Mike: Yeah, for sure. You got me thinking. And then the other thing, right out of the gate, that I started thinking about was when we get a new advertiser that starts on the platform and starts running traffic, they usually are basing it off of some, like if they’re new to pay-per-call, they’re usually basing it off of some other value which is a duration. So they may be getting calls from TV and they know that everything that lasts 30 seconds is worth $50, but this is just a different beast, a different animal, a different set of media. Or they might be doing radio or print or anything else, maybe it’s display traffic versus pay-per-click or something else.
So that’s a real challenge when they come to us and say, “Hey, every call that’s 2 minutes is worth $50,” and then set up a campaign, and they’re actually worth $30 or they’re worth much more or something. So that valuation of duration from the start creates a bit of a challenge so it’d just be better to have a better way of determining the quality of calls that’s not duration. So as an industry, I think that we could all use something better.
McKay: Right. And I think, like you said, it’s so ingrained, it requires a total refocusing or recalibration of what’s important. When you literally have people saying, “Hey, it’s over two minutes, I need my money.” Well, garbage, so sorry. There’s a whole mindset that needs to shift with it, I agree with you.
Mike: For sure. And I think that this is just one of the other things we have to overcome and one of them dealing with just phone numbers have a limitation. And maybe one day, we won’t have that issue, but I think that’s a ways down the road.
Yeah, so I think that wraps it up for me. I don’t know if you have any final notes. But before that, I just want to let everybody know that this recording will be up on our YouTube channel, hopefully today, and watch for an email from us with a link to the recording.
And be sure to follow RingPartner on social media. On Twitter – @ringpartner. Our YouTube channel – YouTube.com/RingPartner. And Facebook – Facebook.com/RingPartner. Feel free to email us at Contact@RingPartner. Or since we are in the call business, feel free to call us at 1-888-656-3726. We’d love to hear from you especially feedback on the webinars or anything in general that you’d like to know about the call space. So I’ll leave it to you, McKay, to plug Convirza as well.
McKay: Yeah, thanks. I think I said most of it during the presentation, the real key that I want to stress is, there’s a lot of data in phone calls and you do have competitors using this data. And so if you’re not, they have a significant leg up. That’s the bottom line. And the other thing I would stress as well is that the people who spend the money in this space – the advertisers, the ones who are buying all these phone calls – they’re starting to demand that this sort of data be available to them and so it’s becoming more and more important.
So if you want to engage with us, just shoot me an email, it’s probably the best way to reach me if you have questions about anything. We host a webinar series of our own. I think RingPartner is going to join us for a presentation later in the summer. And if any of you want to join us as a guest presenter, as a contributor, email me directly. That would be great. So thanks again, everyone. I hope you’re having a wonderful day. And if you’re in the U.S., have a great Fourth. If you’re in Canada, I hope you had a great Canada Day yesterday. So thanks again, Mike.
Mike: All right. Thanks a lot, McKay. And thanks to everyone for joining and we will see you soon.
McKay: Okay. Bye-bye.