Dana Therrien 0:00:03.0: 

My name is Dana Therrien, I've been with Anaplan for about six years and I work in the product organization, but I'm really a career sales and revenue operations professional, former Anaplan customer, former analyst. I've covered this space for a very long period of time and today I am joined by Hari Rajagopalan, from Stripe. Hari and I have had a couple of conversations, I know quite a bit about what he's doing with Stripe, also where the company's come and what some of your background is. So thank you for joining me, I really appreciate it. Why don't you give yourself ap roper introduction so I can share some of the great things you've done in the past? 

 

Hari Rajagopalan 0:00:42.8: 

Great, thanks Dana, it's great to be here. A lot of known faces, it's great to catch up with all of you, and for those of you I have not had the pleasure of meeting with, I'm Hari. I've been in the Anaplan ecosystem for over ten years. As part of my Anaplan journey, implementing territory and quarter planning solutions, instead of compensation management, pricing management use cases as Anaplan partner. Then for the last four years I'm with Stripe, as a finance systems architect. At Stripe my role is in the intersection of [?GTM 0:01:12.9] and finance. I would like to say my role as someone who builds engines that drives our revenue forecast, finance forecast and in turn defines our sales strategy, and also influences how we create model territories. I see it as a key link that connects between corporate strategy and operational reality. So that is where I've been at so far and about Stripe, Stripe is a fintech company, it's for the tomorrow. It's a fintech company with the mission to grow the GDP of the Internet. 

 

Hari Rajagopalan 0:01:46.7: 

Stripe offers products and financial tools that equip companies to operate online. Over the years Stripe has become a company that offers its customers the ability to accept payments from anyone, anywhere in the world. Stripe is not just about payments, over the years it has diversified and Stripe is more about now offering the basic financial tools like billing software, the ability to manage companies' finances, fraud prevention, etc. Much more than payments. The recent buzz about Stripe is our ability to accept stablecoins as a payment and our investments in AI-driven space, which is [unclear word 0:02:28.8] available to our customers, putting them ahead of the curve. 

 

Dana Therrien 0:02:32.7: 

So it's a company we probably all use, we just don't know it? 

 

Hari Rajagopalan 0:02:36.2: 

Exactly right. So some of the facts is that nine in ten of the consumers in the US already have interacted with Stripe in one way or another, through our customers, and last year we processed more than $1.4 trillion of payment volume. That's almost around 1.5 per cent of global GDP in e-commerce. So we are small, but yet making a significant mark. We're one of the fastest-growing fintech companies in the world. 

 

Dana Therrien 0:03:04.6: 

Well, I know it's part of the process when we're preparing for these types of events, I have an opportunity to speak to people like Hari and preparing and discussing some of the things we're going to get to talk about. While Hari and I were having conversations we uncovered some pretty cool nuggets about some of the things that he's done and word got out at Anaplan. So if anyone was in the main stage presentation this morning, saw Hari up on the main stage, it was because of a result of that. Hey, would you mind just getting up and sharing some of this? We'll probably talk a little bit about Polaris again today and maybe bring it down to a level that the practitioners are going to think about. So I know that you've been with Anaplan, in the Anaplan ecosystem, for over ten years. What do you think some of the most interesting problems that you have solved, let's say prior to your time at Anaplan, some of the bigger problems and some of the things that you've seen, and the differences that were made. 

 

Hari Rajagopalan 0:03:55.2: 

True, so before Anaplan what I have seen in most of the companies that I've worked with is collaboration begins in effort. Collaboration doesn't happen automatically, it becomes a conscious of where to gather data from multiple sources, stitch them together, bring nice little teams together to form a consolidated or wide view. One of the key benefits we had with Anaplan in multiple companies that I work in the past is this collaboration becomes must easier and faster. So that's one key benefit I see and then also along with it comes the power of automation. We'll be able to automate a lot of tasks with new age add-on technologies that Anaplan releases, like Workflows and CloudWorks, which has only made it easier and made it possible for us to automate a lot of our tasks. Which, in turn, resulted in operational efficiency. 

 

Dana Therrien 0:04:46.4: 

Yes, I think people think of Anaplan as sort of being - although it's a connected planning solution, as we all know, but they think of it as some other software solution. It really enables cultural transformation in the way that people work with one another and the way that decisions get made. 

 

Hari Rajagopalan 0:05:00.2: 

That's true. For example, earlier there are different teams that work in silos and these are teams that are not supposed to work in silos, like one example is a GDM team, like sales team. They will interact with our customers on a day-to-day basis and they'll be the first team to know anything that's happening in the market. If a big deal is signed up, they'll be the first ones to know. That has to go into our finance forecast. In the old world when people are operating in silos this is not going to happen automatically. The sales team have to interact with multiple other teams, regional teams, product teams, and they have to incorporate - interpret and then incorporate what the sales team has signed as a new deal, right? All of these require quite a lot of man hours, effort and a lot of coordination. It took time, it was slow and complex, that's one thing that we have seen. With the connected platform, if you're able to connect all of these teams together, that operates that brings in a lot of efficiency. I have seen that as a first [?bit 0:05:57.4]. 

 

Dana Therrien 0:05:57.4: 

There's so many cultural changes too that occur within a sales organization, because prior to all this transparency that we were able to get, not just with Anaplan, but other tools, the sales organization would try to keep this information close to their vest. Not necessarily share it outside of the organization because they didn't want that type of visibility, especially something - if they set a certain level of expectations and then they didn't deliver on those expectations. Then it was pretty bad. So tell me about the forecasting process at Stripe and how you're using Anaplan to help facilitate. 

 

Hari Rajagopalan 0:06:28.3: 

Dana, you are spilling the beans on all the secrets. I mean that's something that most of the sales teams deal with, it's always a challenge, like you say that, I'm expecting this much of revenue from my customer and then you're held to it at the end of the year. That's what your performance is based on. So that's a challenge that we have even today, it's going to be there forever. It's an art of managing expectations from each individual sales rep. So collaboration becomes easier as it's not just relying on one single person's input, it's a whole team. So there are different sales reps, there are sales managers, and then the hierarchy of about who constantly keeping tracking with our customers. So it's always not going to be through one person, the collaboration itself together makes things easier. So that's where a real-time update, a real-time view for the leadership goes a long way. So when we gather forecasts the leadership should not be the last one to know what the forecast is. They have to be participative along the way, and having a platform that provides this view throughout is the best thing to happen. 

 

Dana Therrien 0:07:31.7: 

Well, Stripe is a subscription-based business, right? Are there terms and commitments that come along with Stripe like that, or is it more usage based, or is it a combination of these two things? 

 

Hari Rajagopalan 0:07:41.1: 

It's a combination of multiple things. Imagine it as installing a tap, there are multiple payments crossing companies, all of them can consolidate with your business. It's up to you to decide which tap to open when and what to consume, so that's how it works. So as long as it's easy to do your business, as long as we have most approval rates in terms of payments going through, I think our customers would love us. 

 

Dana Therrien 0:08:01.3: 

So if you're selling to a high-volume retailer it's much more valuable for Stripe than it would be for just a small mom and pop shop? Is that fair to say? 

 

Hari Rajagopalan 0:08:11.0: 

Our presence is more dominant in the e-commerce side. So more for the Internet economy. We do have point-of-sale devices too. Today if you go to restaurants in the airport or if you go into any other restaurants you can see some of our Stripe [unclear word 0:08:23.2] too. We have quite a few companies that do that, so we are in that space too. 

 

Dana Therrien 0:08:27.9: 

So what are you forecasting? Are you forecasting the signature of a new customer or are you forecasting the usage that occurs after that? That signature? 

 

Hari Rajagopalan 0:08:37.3: 

We do it, it's all in the pipeline, right? When you look at a CRM pipeline, first we know whether are we going to sign this customer? Once we sign them we need to know how much of revenue are we going to get through this customer and what are the projected growth. So it's a bit of everything. 

 

Dana Therrien 0:08:51.2: 

So the sales organisation is out to try to drive new signatures from new customers to get them to sign up for the platform, and then on the forecasting side is it a revenue forecast that goes after that? Just projected usage and things like that? 

 

Hari Rajagopalan 0:09:04.4: 

Yes, I can give you a full view. 

 

Dana Therrien 0:09:05.7: 

Yes, let's do that. 

 

Hari Rajagopalan 0:09:06.5: 

So from the sales side it's not just about signing a deal that has almost a mutable value. What goes a long way is how much revenue that we expect from a customer, or from a deal that they expect. So if we call that out in the beginning of the year, we look our front book, how much of the pipeline deals we have, how much of them are going to be realised this year and what is the revenue we expect. Then once they come back it becomes a full-fledged financial forecast and also a total revenue forecast for us. So our sales reps are accountable for both front book and back book. They will be able to produce forecasts of how much new business they're going to make and also how much revenue they're going to bring from the customer if they're already going with - it's a role of hunter versus farmer. You go hunt new deals and bring in new [?logos 0:09:55.2] into your deal. Also farm and grow the businesses that are already with Stripe. 

 

Hari Rajagopalan 0:10:01.3: 

When it comes to our revenue forecasting, on the finance side it's one step ahead, it is one step lead, it comes on, so later on that. So there we have an established process, it's not just a [unclear words 0:10:11.7] driven. We also need to provide our finance forecast at regional cuts, segmentation. The type of products we have and then Stripe accepts an insane amount of payment methods, so each of those payment methods can be tracked in our forecast, as well. So all of these go together in our revenue forecasting. So I can go into detail much more, but… 

 

Dana Therrien 0:10:32.6: 

Well, yes, but explain how you use Anaplan to automate this and improve the accuracy of it. 

 

Hari Rajagopalan 0:10:38.2: 

Great, so to give you a full picture, let me backtrack a bit to our revenue forecasting process begins with a machine learning model that's generating a very abstract forecast, which is directionally usually correct. We will have this forecast reviewed by our financial teams, there are multiple departments within finance who are responsible for fraud acts, we are responsible for regions, segments, GTM finance teams, GTM teams themselves. There are a lot of different teams who reviews this abstract forecast. Then they have to - in old world they had to work in their own silos in an isolated fashion, trying to update and make these forecasts accurate with their own set of data. They all are interested in different segment cuts. For example, a GTM team is more interested in the customers that the numbers are against, rather than what kind of payment methods they use. On the other side, the product team will be more focused on the payment methods and the segment the revenue's coming from, rather than which customer in particular it's going to benefit. 

 

Dana Therrien 0:11:40.6: 

Because does that drive the development and the product investments that you're making, based upon their usage and how those things are growing? 

 

Hari Rajagopalan 0:11:45.9: 

Yes, each of these developments represent different business functions within Stripe. So their person, their stakeholders, that's why the difference in the data that they're looking at. 

 

Dana Therrien 0:11:55.2: 

So we've got a sales organisation that wants to be able to predict the results to the executive team, for instance. Now I know that you and I joked a little bit, I said in days before the sales organisation would have spent thousands of hours rolling up a number, hand it to finance and then they'd throw it out. Then they would just take some sort of a statistical forecast. That's not happening any longer now and they're actually using the forecast that the sales team is putting together. So the sales organisation wants to see it, the product teams want to see it, because it affects the investment and the future of the product that they're developing. I would imagine do they also get visibility into the pipeline? That could potentially affect their product. 

 

Hari Rajagopalan 0:12:38.6: 

That's true. So the product team have their own narrative, so then the investment on product depends on how much we're going to get back on that investment, so that goes a long way for them. So they get feelers from the market through the GTM teams, they get to know how the market is feeling about a particular product. Obviously all of this is going to start with a small pilot, a few sets of users, gather feedback, just like how Anaplan does. Anaplan did Polaris with us and now we're going through with AI capabilities. So we have our own way of piloting some of our products, that gives us insights on how much of revenue this might generate in the future. If we are able to go along. That has, for the large part, shaped our product strategy.  

 

Dana Therrien 0:13:19.3: 

So then what was it like using Anaplan? Let's bring Polaris down into terms that the average Anaplan user can understand and why they would want to think about using it and the difference it could make inside their business. I think you're probably more qualified than anyone I know to be able to explain that. Especially since you've been around the ecosystem for so long. So what problem did it solve for you and how? 

 

Hari Rajagopalan 0:13:42.6: 

Got it, so I was beginning to explain the revenue forecasting process, so going back to that. So these different teams who are supposed to review our revenue forecasting, they are earlier doing it in spreadsheets. Spreadsheets have their own constraints, in terms of number of rows it can accommodate, number of tabs we can have in one single spreadsheet [?we don't want the 0:14:05.9] performance. Also the number of people who can collaborate at the same time. Unfortunately more dimensionality means more number of rows, which a spreadsheet is limited with. That's one of the problems that we wanted to overcome, we want to do all the follow off, collaborating, we didn't want another layer of consolidation to be consuming our finance users' time. We wanted consolidation to happen automatically as much as possible. Having the ability to forecast of the dimension that is closer to how our business operates was very important for us. That's why when we interacted with our finance teams we came to know that in order to do that you need more than 12 dimensions in our forecast. Spreadsheets can never handle that. 

 

Hari Rajagopalan 0:14:52.6: 

We tried first with Anaplan Classic, we tried a very simple model and only to find out that we were still not solving the full problem where every team was not still able to forecast in Anaplan. Because, again, we are constrained by the number of dimensions we can include in our [unclear word 0:15:09.2]. So all of that led us to search beyond Anaplan Classic. We knew that we needed a tool that would not only solve our problem for collaboration, but also solve our problem to accommodate every single type of user, in terms of dimensionality. So it's one tool that's granular enough for individual teams to predict and forecast and the same tool should be able to consolidate and provide a wholesome view of the corporate strategy to our leadership. That's when we stumbled across Polaris. We had a great team from Anaplan represent Polaris, we had a lot of help demonstrating its capabilities. Two things really stood out to us during those demonstrations. The first and foremost was its sparse model capability, which would solve our problem of the number of dimensions that we can have. 

 

Dana Therrien 0:16:02.9: 

So just talk about sparsity, just for a second. So that people hear that term, they know what you mean. 

 

Hari Rajagopalan 0:16:07.9: 

Sure, so for example, if I say that our revenue forecast is across all dimensions, not every single intersection of the 12 dimensions will have a number in it. In a classic Anaplan model it occupies space for every single intersection. The bigger advantage of Anaplan Polaris is only the cells that has a number value, non-zero value, occupies the space. 

 

Dana Therrien 0:16:34.4: 

So in Classic it still had a value and when you're doing these large computations at the scale of Stripe, it was probably slowing the calculation engine down as you were going through it? 

 

Hari Rajagopalan 0:16:44.1: 

One problem is it slowed down performance, another problem is the possibility of having all the dimensions, [?a date solve 0:16:48.9] is not possible. So that is only possible with Polaris, like I can have 100 dimensions, I'm just making up, 100 dimensions added into a Polaris model, but it'll occupy space only when there's data in those models.  

 

Dana Therrien 0:17:02.8: 

So you need Polaris, if you've got a complex, multi-dimensional model, and requirements that exceed the capabilities of, let's say, Anaplan Classic, definitely spreadsheets, because spreadsheets are out at that point. Also performance at that level, for that level of complexity, is of utmost importance to deliver the types of information for the individuals. Because you truly have a collaborative forecasting process where you've got all these different people involved. From the multiple sales reps all the way up to product leaders and people within finance. 

 

Hari Rajagopalan 0:17:34.0: 

Right, that's right. So the big thing that we found was when we moved our first model of Classic into Polaris, we were surprised to find that the same model, the same data set, occupied 50 per cent less space in Polaris. That word is a big revelation for us. That showed us that our data, [?house path 0:17:55.0] data was even at its consolidated form, that could fit in the Classics model. 

 

Dana Therrien 0:18:00.4: 

So there was a huge economic advantage to doing this, as well? 

 

Hari Rajagopalan 0:18:03.0: 

That's true. Huge economic advantage and the amount of possibilities it brought was significant, that changed the way our finance users saw our forecast. 

 

Dana Therrien 0:18:14.5: 

This is coming from someone who's more than just a casual Anaplan user, you've got ten years' worth of experience prior to doing this. 

 

Hari Rajagopalan 0:18:19.9: 

I have some of the Anaplan users here in the room today, as well, [unclear words 0:18:22.8] before. So we had a lot of ground on that part. 

 

Dana Therrien 0:18:27.8: 

Well, if you've got questions or concerns about Polaris then Hari's definitely an awesome person to speak to after this, or to reach out to, to help clarify some of the questions you might have. So talk to us about how you're using Anaplan across both finance and sales and how that has changed the relationship and collaboration between the groups. 

 

Hari Rajagopalan 0:18:48.0: 

Thanks. So all these happened in the finance world, right? So if you look at Anaplan's - sorry, Stripe's Anaplan honeycomb, you can see that we have multiple models in the GTM side, as well as sales and marketing. It all began with a simple territory planning model, four years ago. Very simple model where our sales operations team can form territories in a spreadsheet and upload that into Anaplan. It was when a spreadsheet defined everything, it defined which territory an account goes into and Anaplan just became a single source of truth. From there fast forward now, after four years we have - last year we deployed a model that can intelligently identify all the 1 million accounts that we have and put that into a territory based on the definition that our sales operation team has. It is even more intelligent to identify the accounts that are previously held by a territory owner and they'll retain that for them the next year too. 

 

Dana Therrien 0:19:44.7: 

1 million accounts, is that existing customers or is that a customer and prospects and everything that you're trying to go after? 

 

Hari Rajagopalan 0:19:50.7: 

It's the list of customers that have activated with Stripe. We have new customers come in from the pipeline deals, from Salesforce, every week or every month. 

 

Dana Therrien 0:20:00.3: 

So it's more than 1 million because you've got prospects that are coming in all the time and are you dynamically feeding those to your salespeople as they start to feed in, as opportunities? 

 

Hari Rajagopalan 0:20:10.9: 

Not all 1 million, Stripe's model is different, we don't sell, our salesforce does not sell to all of our customers. Today anyone in this room can just log on to Stripe.com and subscribe to Stripe for your business online and that's a self-serve, you don't need a salesperson coming to sell you. So those does not come into a salesperson territory, like a sales rep territory.  

 

Dana Therrien 0:20:35.4: 

So self-service would be a segment that you service and you measure at that point? 

 

Hari Rajagopalan 0:20:39.0: 

Channels at GTM, that is sold through a sales rep or sold through a sales [?mechanism 0:20:45.0]. 

 

Dana Therrien 0:20:46.8: 

Okay, are you doing any kind of account segmentation? 

 

Hari Rajagopalan 0:20:49.2: 

We do. So we do have a lot of segmentation happening, so Anaplan is our source of truth for annual planning segmentation. So every year a large exercise happens before all planning exercise in finance and… Well the first step for planning is making sure that the segments that we had last year is still intact and for the new customers that were tagged through our sales reps are still accurate. A lot of customers grow in size. We have a large customer base of startups, they tend to grow. We have seen some of our customers grow along with us. So they get promoted for being startups to enterprises along the way and the other way happens too, it's the nature of business. So all of these are tracked well every year and that becomes the starting point for us to start our revenue forecasting. So what we forecast as a segmentational forecast for previous year, we will not hold exactly through the next year with this dynamic shift. 

 

Dana Therrien 0:21:49.4: 

How frequently do you move accounts between segments? Is it annually or is it continuous? 

 

Hari Rajagopalan 0:21:54.2: 

We try to keep it static for a year. 

 

Dana Therrien 0:21:57.8: 

Yes, for quota-setting reasons, I would assume? 

 

Hari Rajagopalan 0:21:58.5: 

Exactly, for quota setting and for revenue forecasting reasons. That is what we see our revenue forecast also by segmentation. So it's important that it stays the same for the whole year. 

 

Dana Therrien 0:22:07.3: 

So when you re-segment a group of accounts, do you go back and restate the results for that new segment based upon the new accounts that are in there? Now you have baseline and you can see what the growth is from that point forwards? 

 

Hari Rajagopalan 0:22:16.4: 

Exactly. We'll be able to see what was the transformation of the customer or the set of customers from startup to enterprises or to any other. 

 

Dana Therrien 0:22:26.7: 

You know I'm passionate about sales performance management, also the relationship between sales and finance and I know you guys are using us for OPEX planning, as well, and headcount, as well. So what's the intersection between the forecast, the OPEX plan, the headcount plan? There's so many balls up in the air constantly and they're changing every single month based upon last month's results and what's forecasted for the rest of the year. How are you managing that problem? 

 

Hari Rajagopalan 0:22:55.9: 

So we have a vision to integrate a lot of these things together. Part of it is already done and we are still continuing to do that. One big vision we have for the future is headcount ROI planning, where we wanted to make sure that our headcount investments are assessed and planned well, before investment goes out. We will be able to track the sales capacity first, that comes in from the resources we have. We'll be able to track with the given resources how much of revenue we could earn in the next year and that is going to compare against our revenue forecast that we hear back from our sales teams and also what we field from the market and the market expansion, as well. If we are sure that this is going to - that'll help us understand whether we are well equipped to go and sell in a market. That'll also help us drive investments that we wanted to make in our sales teams. This, along with our OPEX data, where we can identify the cost per head, will tell us what would be our OPEX. So once we know what is the revenue that we're going to get, what is the OPEX and headcount OPEX that we would need for managing this much revenue with additional heads, we'll get a return on investment. So all these are going to be in an integrated fashion. 

 

Hari Rajagopalan 0:24:09.4: 

Territory planning will be incorporated into this, as well. Quota will be directly driven by the [?ONR 0:24:14.4] targets or the revenue targets that we have. All of these are - it is our vision, it has not materialised yet, but we'll be there soon. 

 

Dana Therrien 0:24:20.9: 

Well, it's a vision I've been calling integrated revenue performance management for quite some time, because all of this stuff is so interdependent. If you start with account segmentation, you understand what the opportunity is for that segment and then you start to do capacity planning based upon that. That's based upon a productivity model and then you decide about the investments and the ratios of investments and salespeople you want to make against that. That ties back into your OPEX plan and then you've got all these product managers who want to understand how the business is performing. Just for their slice of the business. It sounds to me like you guys are moving towards that vision. 

 

Hari Rajagopalan 0:24:53.8:  

Exactly. Sales teams are the primary beneficiaries of this whole planning model, so they'll be able to see an investment of additional headcount, or one or two headcounts, how much of revenue is that going to [?lead 0:25:05.0] to them? Even before making it. It also informs in the target setting for what they are holding. 

 

Dana Therrien 0:25:12.3: 

I think the pinnacle of revenue performance management is having the ability to cut, to track customer acquisition cost and customer lifetime value. Is that something that's on your roadmap or are you doing it now or have you thought about that? 

 

Hari Rajagopalan 0:25:24.6: 

We are not that far yet. We are still in the early stages of GTM planning. Compared to our finance side, our GTM side is still in the very early stages. We recently implemented a quota transaction model, in-year quarter [?transformations 0:25:36.6] like on transfers, how that can be easily affected and we can maintain a single source of truth. On Anaplan we very recently deployed an Anaplan model for that purpose. So once we have our quota planned for the next year, our sales operations teams will be able to transact in-year quarter transformations within that tool. 

 

Dana Therrien 0:25:57.4: 

As someone that's come from this space I bet you have a vision about where you want to go. How are you able to manage all of this for Stripe? Like you personally and whatever people you have supporting you. 

 

Hari Rajagopalan 0:26:08.6: 

So we have - we are a very small team, so comparing what we have achieved, we had like - if you look at this on [?honeycomb 0:26:15.7], we have like 18-plus models today. We started with a single honeycomb for income statement reporting four years ago. Then we expanded to all of these 18-plus models in the last three to four years, mostly within our in-house team of five model builders. 

 

Dana Therrien 0:26:31.2: 

Did you say you have team members out there in the audience? 

 

Hari Rajagopalan 0:26:33.9: 

We don't have our model builders here, we have our users though, one of our users here. 

 

Dana Therrien 0:26:36.8: 

Okay, well, hi everybody. 

 

Hari Rajagopalan 0:26:39.1: 

So we were able - we also sought help from our vendor partners every now and then, but for the most part is our highly-efficient model building team who cranked out most of these models for us. 

 

Dana Therrien 0:26:48.4: 

So you working personally with them, you tell them what it is you want to do, you help them shape the vision and then they go off and [over speaking 0:26:53.3]? 

 

Hari Rajagopalan 0:26:53.5: 

Yes, it's always that we have a team, we have leaders who define the vision and then react on that. We make sure that our users are well served, well equipped to forecast the way that they want or closer to how the business operates. That's what we make sure that our users are successful in their day-to-day jobs. 

 

Dana Therrien 0:27:12.0: 

Do you dare share any secrets about honeycombs that are not up there now that you've thought about? 

 

Hari Rajagopalan 0:27:16.9: 

So we do have ROI planning, which I think is not up here yet, and we also have some of our support operations that could come up there. 

 

Dana Therrien 0:27:27.9: 

So just tell me about before and after Anaplan, what it was like. What difference did it make? 

 

Hari Rajagopalan 0:27:33.9: 

So again, the biggest change was with Polaris. We were with Anaplan for a long time, so in the last few years the big shift we saw was after we migrated to Polaris. The biggest change was collaboration, like I mentioned earlier. 

 

Dana Therrien 0:27:45.5: 

So that was a cultural transformation? 

 

Hari Rajagopalan 0:27:47.8: 

It was a huge transformation. Like our leadership was very excited by the fact that they are able to travel along the whole cycle of forecasting. There was a time when all of our intermittent leadership reviews happened in an Anaplan screen and not through slides. That was a huge win. When we were able to break that and award the time it took for the finance teams to consolidate data, repurpose it in a form that is easily consumable for our leadership, all within the same tool, it proves to be a very great advantage. That helped us save a lot of time and all this time was reinvested in making the forecast more accurate. In the last few years our revenue forecast has - is close to more than 90 per cent accurate to reality. In a few cuts we are even more better than that, it depends on the teams and the kind of data we receive from different sets. For example, startups is one segment where we are still wanting [unclear word 0:28:45.6]. So there are different such cuts which we wanted to include, but for the most part wherever there's stability our forecast is much better. 

 

Dana Therrien 0:28:53.3: 

So the teams that were previously responsible for collecting all this information and putting it through spreadsheets or whatever tools it might have been, what do they do after you deployed Anaplan? If they're no longer responsible for all the tactical and administrative burden of capturing and normalising all of this data, what are they given the ability to do after that? 

 

Hari Rajagopalan 0:29:11.0: 

So once they have all of these things in one single data cube, it makes collaboration easier. 

 

Dana Therrien 0:29:16.3: 

Yes. 

 

Hari Rajagopalan 0:29:16.3: 

The amount of time they're spending talking to each other, talking to different finance teams, is drastically reduced. If one team inputs their knowledge of forecast adjustments into the model, that is immediately reflected to all other teams. So the entire team can always see what is a consolidated revenue forecast for Stripe in one place, earlier this was not possible, they were just using a model, a spreadsheet model that is confined to their own business. If they have to know what's happening on the other side, for the same revenue forecast, they need to make the additional effort to go and talk to the team, understand and interpret it in their own way. 

 

Dana Therrien 0:29:53.9: 

Well, it must have freed up a lot of time, given them some excess capacity, to think of things that are a lot more strategic than just the tactical elements of assembling information and making sure it's correct. 

 

Hari Rajagopalan 0:30:02.7: 

Exactly, and we also saw one advantage that we didn't anticipate for. We have a quarterly forecasting model, our users who do revenue forecasting some back every quarter, make sure they have the latest actual numbers and they compare it against their previous plan and try to generate what would be the forecast for the next quarter. With us having a fully-automated model, We were able to generate a revenue rate cut, with the sales volume and the revenue [?draws 0:30:31.5] we had, we are able to have a revenue rate card that caters to every single department. From that point it made our revenue forecasts much more frequent than quarters. We are able to forecast after each month. Once we have a month of actuals in the model we are able to find out what would be the outlook. Again, generated by our machine learning model, we just feed that sales volume, it's going to automatically generate as the forecast for revenue and with that forecasting at the end of the quarter is a matter of when than how. The forecast is always ready with you, you are able to re-forecast almost at the frequency of every month, without any additional effort. You just have to decide is the number accurate enough to call it a quarterly forecast. 

 

Dana Therrien 0:31:14.4: 

Well, that must give your executives some comfort. 

 

Hari Rajagopalan 0:31:16.0: 

It did. They saved a lot of time, it made their forecasts more qualitative and they were more comfortable in talking to their peers from other departments, from [unclear word 0:31:25.3] teams, from sales teams, to understand. It gave them more time to interact with them and the kinds of insights that they brought into the forecast proved much higher. 

 

Dana Therrien 0:31:34.3: 

Can you talk about CloudWorks and some of the other user flows that you have and how it integrates with Anaplan? 

 

Hari Rajagopalan 0:31:40.1: 

We use CloudWorks in almost every single model that we have, in all 18-plus models. There was a time, not very long ago, where we depended on external ETL tools to automate some of our processes, all inbound data flow, outbound data flow, any model-to-model transactions, all those need to be done using an ETL tool. With the option of CloudWorks, the first thing we solved was any internal processes, any internal set of data movement happened all through CloudWorks. We were able to schedule them, we were able to run them ad hoc from a dashboard. It gave our users more freedom to move the data as and when they want, however they want. The second problem we solved was pushing the data out of Anaplan. As much as the capability that Polaris brings in accommodating a lot of these dimensions, a forecast is only valuable until it becomes actualised. Once it was actualised it's not true anymore and we all try to make it not to be true anymore we want to beat our forecast. So we need to store this somewhere, so it was easier for us to do that using CloudWorks, where the finance team's… 

 

Dana Therrien 0:32:54.8: 

So it's a baseline for you? You take a snapshot of the forecast and then you compare it? 

 

Hari Rajagopalan 0:32:58.3: 

Exactly, we take a snapshot and then we push it into our downstream tables and today we don't need a systems team member to go and do that for us, the finance team members, they can do it themselves with the click of a single button. All enabled through our CloudWorks function [over speaking 0:33:12.6]. 

 

Dana Therrien 0:33:12.6: 

Do you go back after the forecast period and calculate what happened? This is what we said was going to happen, this is what happened, here's what the difference is. 

 

Hari Rajagopalan 0:33:21.5: 

We do and, again, powered by Polaris. Did it have a [?BBA plan 0:33:25.3] with it actuals model built? So every time we finish the forecast, we mobilised that into that particular model, we brought in actual information from our data warehouse models and we were able to compare the difference between plan and actuals. That gave us more insights on to how we are trending towards what we initially planned, and it's only useful to make our future plan more perfect. 

 

Dana Therrien 0:33:50.9: 

Great, so when you deployed Polaris, you worked pretty closely with our product teams, I think, and you were giving feedback in interactions. Can you explain what that was like and what that process… 

 

Hari Rajagopalan 0:34:02.2: 

So when we adopted Polaris we were one of the first lot to use Polaris. We didn't have a blueprint on how Polaris works, we didn't know how to model, we didn't know what to watch for. So we had to rely on our Anaplan sales team and Anaplan [unclear word 0:34:16.4] teams. They were very helpful. Anaplan sales team invested a lot in making sure that we were successful. There were many weekends when our model didn't work the way we wanted, they were available on call for us to support. Our platform managing teams, they helped us a lot too. They helped us prioritising some of the features that were more important to us, rolling them out to us before it was available for general audience. All those helped us shape our Polaris platform and create the [unclear words 0:34:41.8] in Stripe. 

 

Dana Therrien 0:34:42.9: 

Well, and I know you helped us a lot too to make the experience better for customers that came after you, yes. Did you use the Polaris migration as an opportunity to, let's say, rectify some overdue maintenance on models and maybe rewrite some models that you knew probably needed to be re-written at some point in time? Now that we're going to make the new models, we'll just do what we need to do. 

 

Hari Rajagopalan 0:35:03.9: 

That's right. So I shared our first revenue forecasting success, very recently one of my team member's implemented another model in Polaris and that was exactly - it was exactly the use case that you mentioned. We only had a model that grew all the years, painfully, it was still in Classic, we knew that we have to change it. When we knew that Polaris would work and handle sparsity well, we immediately grabbed on to that opportunity and now we are in a place where a lot of that happens in Polaris, as well. Like I mentioned earlier, it saves a lot of space for us, that is good economic value for us. Also along with our workflow tools, that Anaplan provide us an alternative option, we were able to automate a lot of those processes directly. Earlier we wanted some of our system team members to be running the forecasts for our finance teams, not the case anymore. We were - our finance teams are able to run the forecasts themselves. All we had to do was set the platform for them and make sure that they're all onboard. From that point to run and finish a forecast, it's all on there [unclear words 0:36:12.3]. 

 

Dana Therrien 0:36:13.5: 

Yes, all right. So one final question from me, now that you have this audience here, what piece of golden advice do you have for the members of the audience, for people that are contemplating a similar journey to what you've been through? Not necessarily around just Polaris, but just Anaplan in general. Because you've got so much experience at Stripe, also prior to that, than many other customers. 

 

Hari Rajagopalan 0:36:32.3: 

I'm not big enough to provide advice, but as an experience or from learnings from my side is before you move to Polaris, I think with Classic there are a lot of things that can be achieved. Moving to Polaris is significantly advantages, you'll be able to see a lot of value that comes along with it. Like how we saw the possibility of having a revenue rate card, which was not even on our cards, not even an agenda we had when we switched to Polaris. The opportunities are unlimited. Just keep an open mind when you move from Classic to Polaris and you'll find your own truth on what you are finding along the way. 

 

Dana Therrien 0:37:10.8:  

Well, thank you, Hari. Thanks for taking the time for addressing this audience, we appreciate it. Thank you all. 

 

Hari Rajagopalan 0:37:16.0: 

Thank you.