246: Fireside with Ethan Walfish

Charlotte Ward: 0:13

Hello and welcome to episode two hundred and forty-six of the Customer Support Leaders Podcast. I'm Charlotte Ward. Today, welcome Ethan Wolfish for a fireside chat. I'd like to welcome to the podcast today, Ethan Wolfish. Ethan, lovely to have you back. It's been a little while, uh, not least because it's been my bad for not arranging this sooner. But thank you for reaching out because we had a lovely little chat on Slack about a topic that's dear to my heart. And uh you said, you know what, we should just hit record. And I don't, I think it was a brilliant idea. So here we are. Hello, and we're recording.

Ethan Walfish: 0:57

Hello. Um, nice to uh nice to catch up with you again, Charlotte. Um, yeah, I'm excited. So I was thinking the other day about um case tagging. And I don't know how this came up, but I started thinking about the model that I use. And I think someone else talked about the model that they used, and I thought it was crazy. And then I thought to myself, you know, I feel kind of opinionated about this. I think that maybe Charlotte and I could talk about um our thoughts on case tagging and talk a little bit about the the differences and and why why we feel so strongly about ours.

Charlotte Ward: 1:29

Yeah, I hope it wasn't my case tagging model that you thought in your head was crazy, right? But it might be. I'm willing, I'm willing to just go on that journey with you. Should we shouldn't we talk about this? So you have so you have a method, and I have a method. Um do you want to go first and tell me like structurally, how do you think about it? What what have you put in place?

Ethan Walfish: 1:52

Yeah, sure. So I've implemented this method a couple of times now. And um, someone on my team at one point actually pointed out um that we were trying to measure three things with two fields. And we use custom fields as much as possible. So there's there are drop-downs, right? There's a limited selected list. Um, and there's basically three areas we measure right now is uh product area, what part of the product is this case um related to? So that's like, you know, if it's with our dashboard or with an export, et cetera. Um what is the cause of this, right? So why did someone actually open this up? Is it because it was a you know how-to question? Is it because it was a configuration request? Is it because um they ran into a UX problem, right? And something that's super confusing to them. And the last thing that I think is kind of unique is how did we resolve this? So is the resolution of this particular case? So did I tell them how to? Did I send them documentation? Did I kick them over to their customer success manager? Or in some cases, and and I think one of the most valuable ones for us is did I send them to engineering? And those are kind of the way that we think about it across these three different tags, which allows us to do a lot of slicing and dicing of our data.

Charlotte Ward: 3:00

That's interesting. Uh, I have a different method, which I'll describe to you, and then we can pick each other's apart.

Ethan Walfish: 3:07

Go for it.

Charlotte Ward: 3:09

Uh like because I I I like some of the things that you're trying to keep track of. Uh, but I think there's other ways to do that. And I'm sure you'll think there's a different way to do what I'm trying to do. Here's here's what I do. I'm only interested in answering one question. Interestingly, you said you were trying to answer three things with two fields, but I'm trying to answer one question with two fields. Uh and but also like you, I do make use of a custom field where it's a limited drop-down. It's very structured. I I'm not a fan of the free form tagging. And you know, that's like you. I've got a well-honed kind of it becomes absolute chaos.

Ethan Walfish: 3:46

I know some people use use tags, and someone came up, someone I heard someone ask me the other day about like, what do you use true true tags for, right? Freeform tags. And I said, like only things that are ephemeral, right? Only if it's something that I'm using only today and never again, or that I can do it with automation, right? So in Zendesk or in any other platform that you're using, right? Like Help Scout or or you know something like that. But those are the only uses I think for those freeform tags, is if it's either super temporary or entirely automated, one or the other.

Charlotte Ward: 4:12

Yeah, yeah, absolutely.

Ethan Walfish: 4:14

That's a whole separate topic. But but tell me, Charlotte, what are you measuring? I I'm I'm in suspense.

Charlotte Ward: 4:19

Oh, I'm keeping you in suspense. Uh I I only want to know one thing, which is why did the customer come to us? What is the what is the cause of that ticket? That's all I'm answering. Um and the way I do the way I answer that question is I break it down into two parts. One is the customer had a job to be done. That's my first field. And the second thing is why couldn't they do the thing that they wanted to do? What's the cause, the actual cause of them reaching out? So it's job to so my two fields are job to be done and cause. Um and uh yeah, I I just find that works really well because I I think that, and I know this will vary a little bit product by product, org by org, but I think that there are other ways to capture some of the things that you talked about. I think product area can for us, where we are now, be inferred by other things. So we have product analytics that can tell us things about what the customer is doing at certain times and how they're using the product. So what product, what the product area is, what the screen is, what the feature is, right? And then um the resolution type, I um it's an interesting one. But again, for me, I think I would always be able to infer that by looking at the path of a ticket if it was following a certain process. So for me, it's just why did the customer come to us?

Ethan Walfish: 5:46

For us, I find that resolution to be so incredibly valuable because you know, I don't get questions from people like product and engineering that's saying, like, well, um, you know, how many cases went over to engineering? Right. And how often did things, how many, you know, enhancement requests are you filing? And there's other ways to derive that, right? Like maybe I look at Jira count or something like that. But but that's not always, since there's that many to one relationship in some scenarios, right? That doesn't necessarily represent how many things actually went over that way. It is interesting to not consider it. I'm trying to think about what other some of our other resolutions are and how valuable some of these are. But it's been really helpful for us to say, oh, in particular, one of the ones that we're looking at now is we have a uh gift operations team here at Alice. And um, so our gift operations team is um is sort of helping to manage the flow of gifts as we're purchasing and getting them shipped out to people. And so we want as much as possible to manage the interactions of the customer on in support because we're there, we're in the moment, right? We want to be able to like actually bring that full loop. And um, so we're looking now at what percentage of our cases are are going over to the operations team and and can we instead manage that in within our team in order to make it more effective for our customers and and our gift recipients, as well as keep that burden away from the operations team. So it's there's uh a number of interesting insights that come out of that data that we have.

Charlotte Ward: 7:10

Yeah, yeah, I can see that. I can see that. Um I I think for for the way uh like it it's so specific, isn't it? It's so it's so uh specific to the organization and how you how your like your whole ecosystem runs. Because what you said there about deriving that data from other sources is absolutely the the thing that we do. So I wouldn't expect or want my team to invest any manual effort in saying this is the thing that I did, because uh exactly for us, exactly like if a ticket follows a certain process, I know it's been escalated. And uh uh, you know, the the response also we time track, we time track every ticket and we assign a task quite often if it's out something out of the ordinary. So I have other ways of inferring that. And you know, I I guess that that's that's ultimately the thing that you need to ask yourself, isn't it? Is is how valuable is the thing that I'm trying to measure and is it valuable enough that I need to invest time in catching it manually.

Ethan Walfish: 8:22

And and I mean, one of the other uses that I have for it right now is that it's actually informing my capacity model. So I basically have the resolutions tagged with a level of you know complexity associated with them. And based upon that, and some you know, some terrible math in an Excel sheet that's overly complex, right? It comes out and sort of tells me like basically here's the number of hours that we spent on these various tasks across the month because a how-to you know problem is going to be you know more expensive potentially than just sending something over to a customer success manager, which is going to be less complex. So I guess you're probably doing your modeling. Are you just using um handle time? Is that your primary metric for capacity?

Charlotte Ward: 9:02

It is, yeah, yeah. Which which which absolutely comes out of time tracking. So my team time track every single ticket, every single activity, because um my environment's also quite complex in the only about a third of their time, maybe a half on a good month, is spent in tickets. The rest is spent in infrastructure support, or you know, oper like internal operations and things like that. So it's a really complex uh ecosystem, which is why I derive so much other stuff from time tracking. Um, yeah, yeah, it's an interesting one. I um the the other the other thing that you talked about was escalations into product and and sort of the metrics that you can apply to that um from the tagging. What are you looking what are you looking for there?

Ethan Walfish: 9:53

So one of the things that I I like to look at is looking at the tags nested within a tag, basically. So if I look and I filter and I say, you know, within my my data and I say, show me things that were sent to engineering, I want to understand what the product area and what the cause was of things that were sent to engineering. So I can look and say things like, you know, of cases that were, you know, sent to engineering, 30% of them were this product area, or you know, maybe even you know, 30% of them were caused by, you know, UX or performance or something like that that might be under that cause um field. And so by even just filtering into the individual, you know, tags there on the resolution, I can look back at the other data and have a better understanding of what is actually driving volume for those reasons. So um sent to, you know, referral to engineering being one of the big ones, referral to product being another one of the big ones. Um looking at other things like, you know, again, for us, our gift operations team, looking at when we send to our gift operations team, why are we sending to our gift operations team? Um and the answer is that most of the time it's it's an issue with the gift that's come up. But um there are other trends that are within there that are interesting. And some of the the other side too is I can look and say, like, if it is the product areas related to gift cards, right? Understanding why are people opening that up. And one of the things that came out for us was that there was some really the predominant percentage of our cases related to gift cards were because of user experience. And so that was something that we provided a lot of feedback over to our product team, saying, like, hey, like our like this is something that we really need to make some changes to because people don't understand how to redeem the gift cards correctly. And so we made a whole bunch of changes, some updates for mobile, which I thought was really wise. That was something that we kind of teased out along the way that we're like, you know, I think these people are on their phones, and that's part of the problem here. And um, and that was something we were able to provide some really actionable feedback for product in aggregate without going too far. I I will say though that one thing that I like about there are two things I really like about the derived data model that you have is that one, there's less of this, um, there's less decision overhead for the team. But I really hate having to ask for the team to make more decisions on their cases and and to have to think more. It's like having to ask to go through that drop-down and pick between 20 things is another thing that I don't really like, but I don't have a really other great way to do it.

Charlotte Ward: 12:27

Like how robust your systems are compared to how robust your human beings are. I mean, all things are fallible fallible, aren't they? Right? Uh uh any system is only as good as as what you code into it, you know, garbage in, garbage out, and all that. But um I think the thing that I will say about systems is they're easy, you don't get it right first time, but they're easier to tweak than people. Because old habits die hard. If somebody's like stuck on picking that thing instead of that thing in a drop down, then it's a hard habit to break. I mean, it can be, you know, but uh yeah, yeah, yeah. If you trust the systems and you're very careful in like how you how you operationalise against it and calibrate it, then generally I'm with you. Like I I just I want fewer things for my team to do. That decision overload is is the thing because what support environment isn't stressful and isn't already full of 20 things you have to think about with every case, right? So uh I just I just want as few, uh I really just want to take as much off their hands as possible. Um, and in some ways you could say I'm forcing them to do other things like time tracking, but that's that's a that's not a decision, that's just a start stop, you know. It's in it's in the ticket or it's in a, you know, we've got that, we've got Clockify plug-in everywhere, basically. So if they're logged in, it's I'm doing this now, I'm doing this, I'm doing this. So it's it's there's it's it's no overhead. It's time tracking is like the equ like the the uh you know, and is like the data uh creation, if you will, equivalent of like Steve Jobs wardrobe. You just you just open the doors and you pick out the next black pullover, you know.

Ethan Walfish: 14:23

I was so bad at time tracking myself, right? Like when I was working on the front lines, I just you know, the the clock either wasn't ticking or was ticking for an hour too long.

Charlotte Ward: 14:33

It's a habit, yeah. You just gotta start. It's it it can take months to really, really uh embed in a team, but once it's there, it really just takes so much of that decision overload that you're talking about away. Um the uh the thing that um I think is interesting is what you do with the tagging though. Um and for me, and I I take from what you're saying as well that we're always looking for the upstream fix. Like we're always looking for what do I do that this doesn't have to happen again? Right? Yeah, for sure. Um and and that that is not always a product thing, but um what what's the uh what would you say is the if you can think of like a really great example of where tagging got you a big win, like it just called out to you from the spreadsheet or from the dashboard or wherever and said, you know what, this is a huge problem we did not know we had, and we know how to fix it.

Ethan Walfish: 15:42

So we had it was the end of last year, and I was looking at our tags, and um I think it was maybe a year, year plus ago. And what are the years anymore? And I was looking at our our tags, and the product area in particular was huge for gift cards. That was by far our number one case type in in support. And so this sort of started what has been, frankly, a journey of iteration on our gift card uh user experience. But what happened when someone redeemed the gift card within our platform was that you would get an email afterward with the gift card redemption information. And um, everyone lost this email. Nobody ever saw this email, everyone would lose it. They would say, I didn't get it. And so what would end up happening is we would have to go to the back end, pull out the redemption info, send it somewhere, try to make sure we were sending it as securely as we could and make sure because it was effectively us sending, you know, some amount of money over email funds to someone. And so it was this whole big process problem because it was expensive to fix. And it was something that, you know, like it seemed like, why do we send this via email? Can we do this differently? Can we have this instantly happen? So we provided and said, like, hey, I think it was like almost 20% of our volume was related to gift cards. And so we were able to say, hey, here's the number of cases that we have. And in some way, we reached a point where I was like, had another one today, had another one today, had another one today, had another one today, and popped this into a JIRA continually until a product manager was like, okay, okay, I've got it. I've got it.

Ethan Walfish: 17:15

And we made a change so that the gift cards are redeemed now almost immediately in the browser. So when you claim a gift card, you get a redemption code right there in the browser instead of having to receive this email and waiting and this whole sort of like delay process. And it made a huge difference and reduced that volume to, I think it's less than 5% of our volume now in gift cards. And the ones that we do get, we're still tweaking a little bit. Like I said, one of the things that we did, and this was a couple of months ago, was we made an update to um to the information that's displayed on the page that you get when you get that card redeemed. So previously it would be like I used to say it was sort of like Russian nesting dolls, would be like you get to a page, you have to follow to another page, you have to follow another page. Right now, you get to one page and you have to go to another page. Right. And that's like a great improvement over having to make you know, sort of three visits through different pages. And by the way, you have to copy the code from this page and the number from that page and paste them both into this final page. And you can speak of like overhead of memory stuff. And um, so it's been a huge improvement for us. But I think that was the biggest one that really jumped out was that we were able to say gift cards are number one. And if we look at gift cards, it's UX. And right now we're fixing it by it's a task, right? We're manually sending this stuff along. And that was like the biggest sort of thread that we were able to pull through all of those tags that we had.

Charlotte Ward: 18:29

Data's super powerful, isn't it? I mean, there's a broad statement for you, but uh who knew. But uh yeah, um, I it it's just when you when you think about that kind of volume, and you know, somebody lost an email, you have to resend them something or regenerate something and get it through a process. It's probably not the longest, most complex task in the world, but when you get in that sort of volume, it's noise, isn't it? Like it just it just clutters up the rest of the day. So you're not as effective. Your team isn't as effective on everything else they should be doing, right?

Ethan Walfish: 19:05

And it's like I think, you know, for it for managing a team, you know, the you have this horrible, unpleasant repetition that exists. And it's like to me, support is interesting because it's always changing and evolving. And if part of your support becomes like this human automation process, yeah, right, that's something you really need to take us take a second to look at and understand. I went, like I said, I I think it was almost 20% of my volume was these gift card things. So it was really something, it was a moment for us to say, like, we can do this for a period of time, but something needs to change at some point in the future because we need to take on other things. And I think about like I tell people that support sometimes is, you know, the guy that's holding the piece of sheetrock against the wall. But I think maybe maybe even you and I talked about this before. It's like the you know, support can sometimes be the guy holding the piece of sheetrock up against the wall. Someone else has to come through, put screws through at some point, right? To actually hold that onto the wall.

Charlotte Ward: 19:53

Yeah, damn right. Yeah, yeah. I I think it's a really great analogy. Yeah. Um, I we have a Couple of uh a couple of ways of putting that human in the loop is a big phrase. Yeah, how do we get the humans out of the loop? Uh and I I dive onto that a little bit further. So that's our generic term for where humans are in something they shouldn't be in. How do we get the humans out of this and do more valuable work, right? Um, but human in the process is a like is it an operating? Are you really just picking up that thing and moving it over there every single time? We have to do that, you know, and then human in the product, that's the guy with holding the you know, the sheet rock up against the wall, right? Yeah.

Ethan Walfish: 20:38

I've seen a lot of that. I've seen a lot of that. And I mean, even frankly, I mean, that's a business system. Sometimes you think about like, are we going to do something to get this human out of this process? And I've seen the answer come back is no, right? Like, and that was a whole thing I've seen a couple of times. And they're like, okay, if that's the case, we like that's a we have to pay a human cost and we have to build up this machine in order to do that. The thing that I one of the things that I hate most about that is that that's not fulfilling work, right? To the point of what we talked about a minute ago, is it's really not, you know, it becomes like click here, copy, paste here, click here, copy, paste here, hit return, move on to the next. And it's like because when you end up with those processes, you start looking at can we make it fewer clicks, can we make it fewer, fewer, you know. Um and you end up with what remains is this super terrible process. It's never, it's never fulfilling work.

Charlotte Ward: 21:33

Exactly. I couldn't agree more. And I think what's interesting about those situations is that in startup land, they often exist and they exist for good reasons, and that is you're evolving and fixing on the fly very quickly. And sometimes you just have to get somebody to carry that thing from there to there for a few months or maybe a year or two, right? While you're still sorting out the big shit, frankly.

Ethan Walfish: 21:57

While you're still building the rest of the house, someone's gonna hold a piece of sheet rack up.

Charlotte Ward: 22:00

Exactly, yeah, yeah. Um, but but you know, ultimately, what I think is uh uh what I think is sort of funny is that it is unfulfilling work, but not every human in the loop. It's it's not like there's a sudden strategic decision, right? Today we've reached a certain size, size, we've got enough money in the bank, we're gonna invest all of our money in just removing humans from the loop. It's like you get these weird kind of asymmetries, don't you? Where and uh it's all about ROI, of course, and and how cost averse you are on as an organization. Um, but yeah, you can end up with kind of some things you can take away really quickly because they are product evolutions, and other things like there is still a guy doing carrying the thing from the sometimes you find people that really like those jobs, right?

Ethan Walfish: 22:59

There's no no like I think that like I think about role players on on support and success these days a lot more than I used to. And you know, I remember someone on the support team I worked with turned to my day and goes, You like house music? No, not not really. And what I found out, this this guy was just jamming the house music all day. And meanwhile, he was just like flipping feature flags in our back end for customers that were signing up deals. And I mean, I just it was the strangest picture, like headphones on all day, just like jamming along and just click, click, click. You know, someone says this feature, that feature, that feature salesperson sends an email, and he's like, Yep, uh-huh. Click, click, click, all set, like moving on. Next one, right?

Charlotte Ward: 23:39

Yeah. Mind blown. Yeah. Yeah. Some people are happy with that, but but yeah, some people are happy with that. But I think what was also interesting is people can also, you know, support folks can also accept their lot, you know. This is my job, so I'll have to keep doing this because this is how I add value. And and actually, I think it's a really interesting mind mind um shift to take someone through. And it can take a really long time sometimes. That if you've been doing human in the loop for a very long time because your your product, company, state of growth, whatever demands it of you, uh, but nonetheless, you have a bunch of other like super amazing skills that are very valuable to your organization. How much value as an individual, like a support person on the front line, are you placing in that component of your work that is human in the loop? You know, uh I think it's really interesting. Uh, and I've had some very interesting discussions around whether that's valuable work or not. Um yeah, it's valuable, it's valuable because a whole other topic. Yeah, uh, it's valuable if you're doing it, but because it needs to be done. So it has a value, of course it does. But also if we took it away, I could think of 10 other, honestly, more valuable things that you could do. How do we take it away? We get our tagging model perfect, don't we?

Ethan Walfish: 25:02

I I think I think that's sort of where we have to come back to. So we have to start, I think that makes the most sense too, is we if we understand what are these things that we're doing, and I did a deep dive recently with my team on like what are things we're marking as a task? What are the resolutions that we say are a task? And understanding what are we actually doing there and are there opportunities for automation? The sad truth, there weren't a lot, but we had to do that exercise anyways because it was something that was really important for us to report out and report up to say, like, here's something we're spending on.

Charlotte Ward: 25:32

Yeah, yeah, absolutely. A good amount of it is certainly not going to be in your agency, but but some of it will be. And you might as well take easy wins if you can if you can take some of that work away, right? Uh we came back to tagging. Did you see what I did there?

Ethan Walfish: 25:46

I thought it was perfect perfect.

Charlotte Ward: 25:48

Pretty smooth, huh? Pretty smooth. Um listen, Ethan. Thank you so much for joining me. What a super interesting chat. We we did go around the houses a little bit, but uh, I just think such a lot of that there's such a lot of complexity in potential tagging models and you know, so much value to be had if you get them right. So thanks for exploring that with me.

Ethan Walfish: 26:24

Awesome, man. This was fun. Thank you so much, Charlotte, and um hope to uh chat again soon.

Charlotte Ward: 26:28

Absolutely, come back soon. Please do, please do. That's it for today. Go to customersupportleaders.com forward slash two four six for the show notes. I'll see you next time.