tl;dr

1. Prioritizing customer success and finding ways to manage customer success teams efficiently can lead to significant cost savings, improved customer retention, and, ultimately, profitability.


2. Automation can be a powerful tool for reducing costs and improving efficiency, particularly for customer success teams. Overall, the company benefits from projects like automation of data aggregation, analytics, and client reporting processes.


3. Balancing the demands of revenue growth with the need for efficient operations is a common challenge that companies face. Thoughtful planning, execution, and investment in technology and resources can help companies achieve this balance and align their focus on profitability with their growth strategy.

Listen to the podcast from an engineering leader and CTO, Dave Peterson, who recently achieved success by automating customer success use cases:

Full script of the podcast on Customer Success Automation

Shikhin Agarwal - 00:05

Hi everyone. Today I have Dave Peterson on this podcast. I am Shikhin Agrawal, founder and CEO of Stats Lateral. Hi Dave. 

Dave Peterson - 00:16

Hey Shikhin. 

Shikhin Agarwal - 00:17

So Dave, you and I have had a great conversation about some of your recent work. Let's start with a quick introduction. Tell us about you and your background. Maybe focus on the last two to three years. 

Dave Peterson - 00:35

Oh sure. So I am a CTO that creates a product engineering culture around solving business problems. Basically the last few years have been working in the Martech space so basically helping brands to market their products to customers in order to get consumers to buy those products. 

I typically run a lightweight agile process within my engineering team, which basically means that there are fast iterations. The engineering team talks directly to customers in business language. 

We also strongly believe in automated testing for quality as we are often working in startups that require 24 x7x365 uptime. 

Dave Peterson - 01:37

Typically the functional composition of the teams that I manage are big data engineering to make sense of massively large data sets, data science and analytics teams using machine learning/AI to improve performance and allow for data driven decisions to drive our business strategy.  We usually have a full stack team that creates web applications for the workflow. We have a DevOps team to create global infrastructure for 24x7x365 uptime. Also we have an IT team that manages the office systems and helps with security compliance. Things like SOC2, ISO 27001, and things along those lines. Finally, sometimes Product is part of the engineering function but not always. 

Shikhin Agarwal  - 02:08

Understood and this is great. I've also had quite an in depth experience in Martech and Martech is an ever expanding field. So just as a quick follow up question on that.  Any particular most interesting sort of areas of Martech that you have focused on or other recent experience? 

Dave Peterson 02:27

So for Martech I recently worked for a company called PebblePost which was programmatic direct mail. Essentially the business is a new channel by comprising two legacy channels. It basically combines the best of ad tech with the best of direct mail. There are definitely a lot of good things about digital ad tech and a lot of good things about legacy direct mail. But there are also some problems with each of them. So for example, ad tech has fraud, bots, pop up lockers and things like that. The experience for the consumer from a creative standpoint is not always great. 

Dave Peterson 03:16

And the same thing, there are problems with legacy direct mail in the sense that it's typically carpet bombing of zip codes. By identifying the demographics of your brand's customer audience and then those zip codes that skew more heavily to that demographic, you end up carpet bombing them which is not very cost effective. So basically what Pebblepost was doing was basically marrying together the best of digital, which is the interest and intent signals. So what I mean by that is when you're typically looking for a new product, say I'm looking for something like a new cell phone. I'll basically do research out there looking for what the options are, the cost, the plans and all that kind of stuff. All that happens on the Internet, typically. But the creative execution on the Internet is not that great. 

Whereas Direct mail has got great creative execution.  The US postal Service, the Canadian Postal Service, the UK postal Service have all done studies that your brain operates differently when you have a tangible piece of media in your hands. So direct mail performs extremely well. The typical conversion rate for digital ad tech is 0.02%, which means you have to issue a lot of them to get people to notice them. Whereas in direct mail it's like 2% to 5% for traditional direct mail and in programmatic direct mail it's like 4% to 15%, which is quite a bit better. 

Shikhin Agarwal - 04:39

Right? Absolutely. It's a very interesting use case. You have a very deep CTO experience. I know, through our conversation before. Tell us a little more about your leadership style, given the complex and cross functional teams that you have managed. 

Dave Peterson - 04:59

Sure. Like most things, my leadership style has evolved over time. But I'm a big fan of self organizing teams. I read Jurgen Appelo's Management 3.0 Managing Agile Leaders and there are a lot of good concepts in there that I've tuned to my own style. So my leadership style essentially is what I call autonomy within guardrails. What I mean by that is everyone has skills that they learn as part of their career path, but there are obviously new things that they have to learn that they have not yet mastered. So those new things they haven't learned yet are basically bound by guardrails, which means essentially ask an expert, potentially myself, potentially an architect or whatever, before you actually release your work or decision just to review and stuff. 

However, once they've learned the skill, then they have the autonomy to handle similar situations themselves without needing to check in with the expert. That mentoring and autonomous culture is spread throughout the engineering team via the leadership structure. And so it's basically a culture of mentoring everyone. 

Shikhin Agarwal - 06:04

Right. Sorry, were you adding something else? 

Dave Peterson - 06:08

Yeah, sure. So an example is that, say, engineers are generally shy and say an engineer is working their way up into management and has never done a company-wide presentation before. So we want to set them up for success. So what we do is basically say, look, I want to give you some exposure to the company, create a presentation to talk about, say, how we did in the last quarter, whatever. So they would create the presentation, we review it with them, give them feedback, maybe say less words, more pictures, whatever it might be, and then they present it and we give them feedback on the presentation. At that point they've mastered the skill. Like, okay, you can basically create your own presentation from now on. If you want my feedback, great, but you don't require it. That's an example of this.

Shikhin Agarwal - 06:50

Right? Well, that's great. And it won't do a full justice if I kind of ignore in your leadership style the relationship with product management. I'm a product manager for life. So how would you define your ideal relationship as part of your leadership style and the success examples that we are going to discuss in today's podcast? 

Dave Peterson - 07:15

Sure. So I'm a big fan of product management engineering working together in partnership. There have been cases in the past where product management has been kind of like hey, you dumb engineers, build this thing that I tell you to build because I know the business in the marketplace.  That stuff has kind of gone out of style in many companies. Because of the advent of Agile, one of the primary tenets of Agile is that the engineer should get as close to the customer as possible to hear the customer's voice firsthand. This game of telephone where the customer says something, they tell it to the salesperson, salesforce tells the product manager and then the product manager tells the engineer. It gets kind of skewed along the way. So I really much prefer to work with product managers who work in direct partnership with engineering. 

So basically product management brings to engineering the customer’s desire, the problem, whatever, that they're hearing from the customers and they discuss together different solutions, alternatives. We can basically say solve it this way which will give you a short term outcome, but it's probably not the best way. If we did this another way, it'll take longer, but it'll give you these extra things or maybe we do both where we get something to market quickly, learn and then that'll give us better ideas for the long term solution as well. So that kind of collaboration, I think, is the real way of the future for product management and engineering. 

Shikhin Agarwal - 08:42

Yeah, and I totally agree with that. I think both the product management and engineering are in my mind like a two in a box.  They’re joined at the hip model, where the success for a product depends on how good that collaboration works between product managers and the engineering teams. So we have discussed the specific cost, efficiency, and success that you recently achieved and that's one of the key topics that we want to address today. So tell us more about what was the situation and the key business challenges you faced for our audience to understand better. 

Dave Peterson - 09:26

Sure. So the situation at my last company was that due to COVID lockdowns, many of our brand customers just stopped marketing. Their spending went to zero. For example, 20% of our business was in the travel vertical. And with the lockdowns, nobody was traveling. So obviously nobody was marketing. Similar situation with gym membership and many other verticals as well. So due to that business downturn, we had to lay off some of the company, which included engineering.  Essentially engineers were down to a skeleton crew and were tasked with just keeping systems running and don't bother doing anything else. However, that's not a great situation for engineers to be in. Engineers love to work on interesting things and through our previous work with automated testing, the systems were quite reliable. So this left time for engineers to work on something. 

So the only question is what that something would be. We had just released two new products within the last few months, so the sales team really had plenty of shiny new tools to sell. They didn't need more products. So I didn't think working on any new product was the right approach. However, there were business challenges that I think aligned to a different approach. So those challenges were that we had a small staff, so we needed automation in order to scale up because once the lockdowns were lifted and the business ramped up, we'd have this lag to growing the business since we have to rehire in order to scale up that business. We also had higher costs since we were running a managed service business with the Customer Success and BI teams manually creating brand performance reports and campaigns which were also sometimes inaccurate. 

Dave Peterson - 11:01

There was also a higher churn rate than we desired from the brands due to these problems with performance reporting and things. So these were areas that we could consider for potential alternative approaches. 

Shikhin Agarwal - 11:18

Right. And over the last ten years, since Capital has been so cheap, I think very few tech companies have talked about higher costs, especially in manual process driven teams like Customer Success. So it was pretty smart for your team to start addressing those higher costs in Customer Success teams, if I heard you correctly. 

Dave Peterson - 11:42

That's correct. 

Shikhin Agarwal - 11:42

Who were the primary stakeholders for something like addressing efficiencies in Customer Success teams? If that's the objective of the project?

Dave Peterson - 11:52

Yes.  So obviously the head of Customer Success would be a primary stakeholder there. However, the head of Customer Success in this particular instance was on maternity leave. So reality meant that the stakeholders were the executive team and the sales team. So we needed to figure out a plan and create the automation changes without any disruption to the campaigns being run for the brands or any impact to the revenue or the CFO would essentially kill us. So in order to get the buy-in from the stakeholders, I first wrote up a plan.  I reviewed and got buy-in from the engineering team Leadership to make sure that they also felt that we could accomplish this. 

Dave Peterson - 12:34

I also felt that giving them something interesting to work on would raise morale since everyone kind of suffered the trauma of seeing their teammates laid off and so giving something exciting to do would motivate them. The next step is I reviewed the plan and got buy-in from the head of product. She and I then took the plan and got buy-in from the sales team and finally reviewed the plan of the CEO to get his buy-in and he took it to the board and got their approval. 

Shikhin Agarwal - 12:58

Yeah. That seems like a pretty strong collaborative process. Along this process, how did you define the key objectives and strategies which I'm sure were part of this plan that was presented all the way up to the board? 

Dave Peterson - 13:17

Yeah, so the objectives were to reduce costs, automate much of the manual labor being done by BI and CS and thereby reduce our brand churn rate. 

Shikhin Agarwal - 13:32

Got it. So any of these objectives, the way it was set up at PebblePost, were they jointly owned by you and customer success leaders or who are the owners of such objectives? 

Dave Peterson - 13:49

Well, it's kind of the executive business strategy that most of the stuff we've been doing up until then had been around new products, new revenues, and so this was kind of a new way of thinking about the problem. And so it was pretty much a collaboration between the engineering team and the product team to try to create this plan. 

Shikhin Agarwal - 14:18

And I think this is the key point here. This is highly relevant. As we look forward to the next two to three years, there's no reason to doubt that we are in a contraction cycle. From a macroeconomic standpoint, the available capital is not going to be as cheap as it was in the last ten years. And almost everybody in the tech industry or been in the industry with experience and the right expertise are saying that there has to be higher efficiency, lower costs with manual labor associated or the processes associated with manual labor reducing the churn rate. All of these are going to be probably even more important for business over the next two years. So I think this is great. Tell us more about how you thought through the strategy, some of the hypotheses and assumptions that you had in place. 

Dave Peterson - 15:18

Yes, you're absolutely correct in this downturn. I think it's going to be a big focus for everyone to reduce their costs because there's only so much revenue that you can capture in a down economy. So you really need to hunker down and reduce your costs. 

So yeah, pertaining to the hypothesis and assumptions, the first hypothesis was that automated performance reports to deliver faster and more accurately (without human error) would reduce the brand churn rate and increase renewal speed, which would lead to more revenue and faster growth for the company.. 

Dave Peterson - 16:01

The second hypothesis was that automating the manual operations by CS and BI would allow us to quickly recover post COVID once the lockdowns were released and business scaled up and returned to fast growth  Instead of having this lag where we have to rehire staff to facilitate the scale up, we should be able to hit the ground running, which would be awesome. 

And then the assumption we made was that we could also reduce our AWS costs if we figured out where the costs were coming from and then eliminated those that we didn't need. Cost reduction in AWS was never really a focus for us. It was always our strategy to continuously build more products. And so this is an opportunistic point to reduce some cost. 

Shikhin Agarwal - 16:28

Right.  And maybe you mentioned this before, I believe the acronym CS is Customer Success and BI is what? 

Dave Peterson - 16:37

Business Intelligence. 

Shikhin Agarwal - 16:38

Right. And that was part of customer success?

Dave Peterson - 16:41

No, they are separate functions. BI is essentially analytics. They created performance reports for brands to say here's how your campaign performed. 

Shikhin Agarwal - 16:49

Right. But this BI is not just internal company BI, but it's basically Customer facing BI. 

Dave Peterson - 16:55

That is correct. Customer facing BI. 

Shikhin Agarwal - 16:57

Awesome. And so in this process, in my head there are strategies, these hypothesis assumptions that you just talked us through. And then the next step is what kind of capabilities are required for the teams to execute these projects more confidently. What were those capabilities required and which of those were in house available? And you had to outsource as in source them from outside just walk us through some of that process. 

Dave Peterson - 17:29

Yeah, so we were under a very strict cross cost structure. So the board basically gave us guidance that you can't have over this headcount so we couldn't really add any new staff. But what I did was propose that we actually remove our BI function and repurpose that money that's freed up from that to basically rehire much of the data science team that we had just laid off. The data science team would run the reports manually, which is essentially what BI was doing for the brands while simultaneously building automation to generate those performance reports. So that was very much an internal switch where we basically freed up some money to spend on an automation solution. 

Dave Peterson - 18:12

As far as the Amazon costs were concerned initially because we wanted to get those costs down really fast, we actually did an internal analysis of it. Basically kind of all hands on deck on the engineering team, look at everything that's in our Amazon costs and clean up a lot of the stuff that has just been built over time and we just hadn't really paid attention to. And so we managed to get our costs down significantly through that process. However, we realized that was really a high cost for the engineering team.  Essentially the engineering team spent a couple of weeks just staring at Amazon dashboards and trying to figure out where to reduce all the costs. We did manage to get the cost down by 80%, which was pretty awesome. But we realized we didn't want to do this level of effort again. 

And so we actually spun up a sub project from that where we essentially evaluated some companies that would help us to understand our AWS reports better. Amazon does not make it easy for you, especially in the case of using EMR, because those EMR costs get hidden in other things like EC2 and S3 costs, and it makes it very hard to distinguish where exactly your cost is coming from. So you can use other vendors who will actually help you with that a lot better. And so we evaluated, chose, implemented, and integrated one of those vendors. So basically it would be sort of an ongoing process as opposed to a once every couple of years process. 

Shikhin Agarwal - 19:35

Right. So to summarize, it's leveraging the API economy at a greater depth, finding those integrations which can deliver the efficiency along with using any other innovative analytics and similar tools available on AWS and cloud computing. And there are so many new products and startups that offer such capabilities. It's a combination of these two or three areas which help you achieve the goals of cost reduction, correct? 

Dave Peterson - 20:09

That's correct. Yes. As you noted, there are a bunch of companies out there. There’s an offer in my  LinkedIn inbox basically every other week. It says something like hey, let us help you. There are a ton of choices out there that the trick is figuring out which is the right one for your needs. 

Shikhin Agarwal - 20:25

Right. So maybe we'll address this in a greater detail in a couple of minutes later. I want to go back to another comment about replacing the BI analysts with data science. I think there might be something interesting there. I feel like there is something interesting there in the sense that there is this traditional thinking of obviously doing business intelligence within mid size and larger enterprises and then offering some of that to customers for value or not. But then clearly there is a totally different value when it comes to data science. Do you think this is something that most of the technology and product leaders should think about in terms of how do we replace the manual BI processes with data science or do you think this was a one off situation? What's your opinion on that? 

Dave Peterson - 21:21

I definitely don't think it's a one-off situation. And you're right, this is very tricky to pull off because there was definitely a lot of skepticism from the executive and the sales team that the data scientists who were typically kind of like back office people sitting in their cubicles toiling away working on things and they didn't really have a lot of front of the house interaction with brands and clients. And there was skepticism that they would be able to handle that. This required a lot of reassurance from us. 

We had them talk to a bunch of the data scientists so they realized that these are actually really quite articulate very smart people and that they could handle this and that through the automation there are so many benefits that even if, say, one of the data scientists had a struggle or whatever, we would just basically not put them in that front facing role. But much of the team were clearly presentable, articulate, friendly faces to help the brands understand the reports. 

Shikhin Agarwal - 22:23

Right, that makes sense. So without getting into any proprietary information, I want to also get a little more specific from your experience, which seems like very successful and obviously applicable in the near future for a host of other companies which are going to look for cost efficiencies like automating, manual processes and so on. What were these the proxy for? And we do not have to have the specific metrics from your previous employer, but what are the specific metrics that people should think about in terms of managing or measuring the automation of manual labor? Was there a couple of metrics that you were going to recommend? 

Dave Peterson - 23:16

A typical one that I use is basically how many CS people does it take.. Sorry. How many brands can you add to the platform before you have to hire another CS person? And first of all, CS is kind of an umbrella term. There's actually three people there. There's basically the ad operations person creating the campaigns. And then there's the technical account manager, TAM, who basically helps with the technical integration between the two companies. Then there's the optimization manager as well who doesn't set up the campaigns but helps the campaigns reach full delivery and reach the performance goals of the brand. 

And so each of those roles has a different scaling factor, but all together can result in say for ten new brands, we have to hire at least one CS person. And if you can scale it up to say, 20 brands before needing to hire a CS person, then you're doing much better, right? 

Shikhin Agarwal - 24:20

​​Absolutely. And then the other specific metric regarding automating, the BI reports, and maybe before I ask that a follow up question was that helpful in creating any kind of net revenue? Like were you able to charge for these reports from the brands or your customers or were they more of an efficiency play? As in it made the brands happier and the NPS went up or some other proxy metric for that was a leading indicator for achieving success? 

Dave Peterson - 24:56

Yes, this initiative, it was really just an in place replacement for the BI function. But yes, you're right, over time this kind of led to a level of service discussion. How big did the brand have to be before they could get a custom report and how do we measure big? It's kind of a vague term. Is it one campaign spend, repeat campaign spends and such? A lot of subsequent debate about that as we got more sophisticated further on. But this initial cost reduction project was really just measured on replacing that function and were able to successfully do that very well. 

Shikhin Agarwal - 25:40

Right, that makes sense. So I know situations like this are unique to businesses but there are always learnings for other business managers and leaders in their situations. How would you summarize some of the key learnings for us? As we have both commented on such initiatives being quite interesting and useful for the near future, next two to three years for sure. 

Dave Peterson - 26:17

Definitely some interesting learnings here. We already talked about the Amazon cost learning that it shouldn't be done periodically, instead should be done on an ongoing basis. So get a vendor to help you with that. I won't go into more detail about that one. But CS automation, I'll put a little more color around that. Essentially, we were basically getting SFTP files from the brand and that process resulted in bad data, incomplete data and other things, because essentially the brand would hire some intern in IT to basically send the files to us. This is unreliable and the data was often dirty.  By basically creating API automation to get the data directly from the CRM and the ecommerce system that got the data much cleaner and much faster. We basically pull every day as opposed to when the intern was getting around to sending it, which allowed us to basically operate as a business much more effectively and have the campaign run much more effectively.

 So for example, if we're basically trying to filter out customers from a prospecting campaign and the data doesn't come from the CRM system on a daily basis to filter customers out then we may actually mail customers accidentally. That's just the nature of this sort of manual process. But now that it's automated we're basically pulling it every day. Then we don't have that sort of leakage anymore. 

Dave Peterson - 28:02

Then for performance reporting, this is a really interesting one is that for digital ad tech typical attribution Windows are about two weeks. For direct mail it's more like 90 days. So what this means is that if you run a quarterly campaign it doesn't reach full Attribution until 90 days after the last send. So you're talking now the campaign runs in Q1. Then you wait 90 days for attribution which is all through Q2 and then there's a 30 day backlog from the BI team to send the report. By the time you're sending that report it's so much later and the brand is like who are you? Why are you sending this report? They don't even remember you. 

Shikhin Agarwal - 28:18

Right

Dave Peterson - 28:19

That's the point in time where you're trying to ask the brand for a renewal. Now what we did as part of this automation is that we could produce reports with essentially a push of the button and we also added in predictions from a partial Attribution standpoint to full Attribution with a confidence interval on that. So for example, what I mean by that is say two weeks into the campaign. So let's say it's a Q1 campaign, say mid January to end of January, we can actually have the conversation with the brand at that point to say hey, we are predicting with an 80% confidence that at full attribution your campaign will reach your performance goal. So let's talk about renewal right now. 

So that sets them up for actually planning, renewing and starting another campaign in Q2, as opposed to running Q1, getting a performance report in Q3, and then running a campaign in Q4. So this is one of the most impactful changes that we made to our business to create success. And the results of this were just enormous. We cut brand churn rate down by 60%. We also reduced 2 million in annual costs. By reducing the variability in the BI function, we were able to achieve 100% revenue growth during COVID and profitability without adding new CS team members. 

Shikhin Agarwal - 29:40

That's awesome. That's really good. Just to summarize one of the important points like you said about the performance reporting automation, it seems like you're able to fast track that campaign planning and launch process within a quarter as opposed to waiting a full quarter and then do replanning for the brand to launch a new campaign. 

Dave Peterson - 30:04

Yes. This is powerful for the sales team because they were previously basically visiting the brand after they got the full performance attribution Report. And it was a very hard sell at that point because the brand has basically not planned for you anymore but now they have the conversation during the current campaign. Then it was easy for brands to plan for including us in subsequent quarters. 

Shikhin Agarwal - 30:26

Yeah, absolutely. You worked on a certain scale with this initiative. What are your thoughts on repeating such a process at a different scale, maybe at a larger scale where these customer success teams might be 3X assuming or 5X, and so there will be complexities associated with that different level of processes and so on and so forth. How would you recommend scaling such an initiative where in terms of success, the cost savings could be even larger? 

Dave Peterson - 31:02

Yeah, absolutely true. I guess I would basically try to start this process earlier. I mean, COVID was kind of an opportunistic point for us to do this, but what happens a lot in the startup space is that you want to focus the engineering team on just new products and new revenue. And I don't think there's a really good understanding from a business standpoint of how the engineering team works or what they're capable of doing. And so quite often you end up just spinning up a managed service team without necessarily looking at alternatives and that can quickly spin out of control and as you know, get to be 3X larger and can create this profitability problem where you're not able to reach profitability because for every ten brands you're adding more people in. 

And then those people need managers, and then those people need more HR people and everything else. So you're on this treadmill of just being barely profitable or not profitable. And typically this sort of decisioning is handled by a COO. But many times in a startup environment. There isn't a COO. The CEO is handling the COO function and that doesn't necessarily get the right focus that it needs and so you can quickly end up getting these team sizes out of control. When I talked about saving 2 million in BI costs, that was at the levels at the time. If we talked about a year later, as we got 100% growth, that team would have cost $4 million. You need to get out in front of this stuff earlier if you want to reach profitability. 

Shikhin Agarwal - 32:47

Absolutely. And I want to double down on the results that you have shared. And again, we don't need to get into specific, even if we can put them in a context of a range. There's an 80% saving on the AWS cost, $2 million savings on the annual cost from performance reporting and that's in the range of how much SaaS revenue? 

Dave Peterson - 33:31

Well, we doubled our revenue, let's put it that way. I don't want to identify specific revenue numbers, but we basically doubled our significant revenue during the COVID period and we achieved profitability as well. 

Shikhin Agarwal - 33:22

Right, so that's huge and I think that sets the right tone for the company, for startups and enterprises of any size I would say, which are able to prioritize some of these initiatives in order to achieve any efficiency cost savings. And clearly for startups, being profitable in such an economic environment should be key so that they are not dependent on external cash or investments. So this makes a lot of sense. Any other closing thoughts? I know we are almost out of time but we'd love to hear any other closing thoughts from your end. 

Dave Peterson - 34:07

No, I think you're absolutely right that, you know, it is very difficult to get any new money in this environment and so reaching profitability puts the control back in the hands of the company and that is very powerful. And I would encourage anyone in this space, especially somebody running a managed service team, to look at their cost and focus on reaching profitability because it really empowers the company to shape their vision. As we reached profitability, the company was suddenly able to now hire people we had put off. We didn't have a marketing function for the longest time and suddenly now we freed up money for that. And a head of HR and all these things that we wanted to do, we now had the money to be able to do. 

Shikhin Agarwal - 34:53

Absolutely. And I'll say as guilty as a product manager it's not so sexy to talk about efficiencies as compared to launching new innovative products. But as a business manager, I would say it totally makes sense and I'm really happy that we are able to tap into your experience and I really appreciate you sharing these learnings with us. Thank you so much and I hope we are able to have such discussions again.

Dave Peterson - 25:25

I really enjoyed this conversation with you and our earlier conversations.  I hope we can do this again.  And I hope this is helpful for people out there struggling with these problems. 

Shikhin Agarwal  - 35:33

Awesome.