Interview: David Ma at Front

I caught up with Front's Sales Operations and Strategy Manager David Ma all the way from San Fransisco's Bay Area. I was particularly excited to hear all about his recent, and astronomical, task of creating and implementing a new forecasting system at Front...

Rory Brown (RB): David, thank you so much for joining me today. To start with, it would be great if you could tell me a bit about your route into the world of sales ops?

David Ma (DM): Like it is for many, my route into sales ops was a little bit circuitous. The first job that I had when I left management consulting was for a health tech startup that had a two-sided marketplace, doing ad sales to pharmaceuticals on one side, and to doctors on the other. In that role, I found myself serving go-to-market teams from an analytical capacity. So we hadn’t called the role sales ops, but I did a lot of things that looked like it, like book planning, territory planning, rep productivity, incentive design, these types of things, and I was also a Salesforce admin just by necessity.

I took a pretty long winding path from there across a bunch of different things, looking at product, looking at starting my own company. But a very consistent theme across all of these experiences was I felt like I didn't have the right knowledge or the right education on what go-to-market motions look like. So the opportunity to join Front, to use some of my experience in the past, and then to grow that muscle at a place like Front that's scaling at the rate that we are, I thought was a fantastic opportunity. I'm really excited to be at Front and to be talking about sales ops.

RB: Nice. Succinct and humble as well. I appreciate that. Today, we've chosen the wonderful topic of forecasting. In this particular scenario, we're talking about a scenario where we have fast-moving deals and fast-moving business. All of a sudden, we've got these bigger, larger upmarket deals, and we're trying to forecast across two very different situations. Maybe we could start with, where did the project come about and what kind of brought on, "Okay, we need to look at this differently."?

DM: When I think back to when we started, forecasting wasn't a muscle that we necessarily had developed at Front. This was about a year and a half ago. As we were getting toward the end of the quarter what inevitably happened is we put a big board of deals on the table or on a whiteboard and asked, "What's our path to 100%?" That was kind of the way that we had done it. But then we brought in mature revenue leaders and people who had, obviously, been at companies where there were much more rigorous forecasting processes. It became really clear that we needed to start doing this.

The first thing that happened is the revenue leaders started to meet once a week to talk about where they think they're going to land at the end of the quarter. But oftentimes, you don't have a lot of rigour to that process. It's a lot of like finger in the air “we think we'll do this.”

It became really clear, especially in a couple of quarters that we had in 2019 that were a little bit more challenging, that we didn't really have a process for understanding where we were going to be 90 days from now. We were tasked, as a revenue and go-to-market function, to deliver a forecast across what we call new business and the post-sales segments.

If I flash forward to where we are today, we have a weekly meeting, all of the revenue leaders are there, the CEO attends every week and the CFO attends. We talk about our potential landing at the end of the quarter. If it's not going to be 100%, what help do we need from the rest of the revenue group on getting there and what areas were blocked.

Let's say July is the start of our quarter and September is the end, if we walk in projecting 80%, we're able to intervene in July as opposed to scramble in September when we have no clue and it's already too late. It allows us to pull a whole different series of levers that have a little more of a long-term outlook of three months versus three weeks.

RB: Awesome, I love that point. Often, we hear a lot of people talking about, maybe even in a boastful way, this is how accurate my forecast is, this is how far out my forecast is accurate, but of course, as you just described, a forecast should be an early warning for when things aren't looking so good and then course correction can happen.

DM: Yes, it's actually impossible for us to forecast using only Salesforce opportunities that far out because, as we talked about, for our fast-paced moving business, more than 50% of our revenue that we book in a quarter doesn't necessarily even exist as pipeline until the beginning of the quarter.

When we walk into July, we don't know how many inbound trials and demo requests we're going to have. We haven't qualified the outbound pipeline. We know that, especially with the faster cycle, a lot of that stuff is actually going to come from things like a hot press release in the middle of the second month. For us being able to forecast that is not a particularly easy thing to do.

RB: Excellent, if we take a step back. Obviously, you had this scenario of you meeting once a week. There was no real process of framework. I'm guessing the output wasn't the desired output, either in terms of accuracy and helping you steer the ship. When you were tasked with just, "Okay, go and create a great forecasting process and roll it out," where did you even start? What did you start to look at first?

DM: Anytime that we're trying to do a forecast, we want to look at signals that are not too crazy. We want to look for parts of our business that we think are stable. I come from a pretty analytical background so the first thing that I thought about was, for at least the high volume part of our business, how stable was that and how much does that change week-over-week?

For us, on the sales side that's companies that are less than 20 employees, they come in, they buy in-product. So we have a portion of revenue there. The question was, how consistent is that? I basically did an analysis to look at, on a weekly basis, how much is that?

The tricky part about doing an analysis like that is you have to figure out what the cutoff is. Is it based off of company size? Is it based off of ASP of purchase? We basically decided that it is right around a certain dollar threshold. So we tested a couple of different thresholds, and it turns out, the revenue is really stable and really predictable in the quarter. It may vary by X% week-over-week, but over a course of eight weeks, it looks pretty much the same. It became really clear to us that we could use that as a signal to say, "If we enter the quarter and expect us to do very similar to what we did in the prior quarter, we know that this stream of consistent small-dollar quick turning revenue is going to be X." Then we track it week-over-week to make sure that we're not overestimating

RB: So you have a fairly consistent baseline, it’s fairly dependable so you can forecast that number because it's just a volume of data running through the system, right?

DM: Right, exactly.

RB: After that cut-off, this next section of deals, what did you do differently there?

DM: The other thing that's always been hard, as with any maturing sales team, is varying levels of pipeline hygiene. For us, coming up with the right forecast methodology was the right way there. When we talk about that, I think it's really important to think about the differences between stage and forecast category. Because stage represents sales motion. Forecast category helps us understand what we’re going to land.

For us, we had started with the assumption that stage was going to be the right thing, for example. It turned out that was totally wrong. I'm sure you've heard this a thousand times, just because it's in stage four negotiation, doesn't mean it's got a high likelihood of landing. It could be negotiating a really bad deal.

For us, it was instituting a very rigorous forecast stage process that had clear definitions, and all the reps knew it. Then we bubble up the forecast where the reps were basically responsible for looking over their deals and understanding where their projected landing of those particular deals in pipe were going to be.

This cascades its way all the way up from rep to manager, to head of sales, where, by the time that we do our forecast meeting every week, the head of sales has a weighted forecast. But then at each stage, they have the right to override certain deals as a managerial piece on top of it.

To summarise our forecast, we divide it into what we call a quant forecast and then a deal-based forecast. The important thing is to make sure that those two don't overlap because then you double count. So you have to draw a really clear line in the sand where everything over that certain dollar amount goes into our deal-based forecast. Even if a rep happens to forecast a really, really tiny deal, we just straight up will ignore it because it's under our dollar threshold.

RB: I see. So the forecasts are there to support each other or complement or challenge each other, not to be blended in any way.

DM: No, they are blended, as in, we add the two together.

RB: Of course, yes.

DM: All of that is our landing number, but we have to make sure that we don't double count on the Venn diagram there. You have to draw a really, really clear line down the middle of what counts for one versus what counts for the other.

RB: Okay, that makes a lot of sense. You've created these two models, that's great. Then we've got a situation where we've got a bunch of human beings in sales leadership and the sales team that have got to do new things or change the way that they're working or change the way they were forecasting before. Where does adoption start to come up in your mind when you're working on a project like this? Is it early, is it later, is it middle?

DM: I think, ultimately, it's the head of sales' job to deliver a forecast to our CEO every week, and so it starts from him. In the early, early phases, we weren't a massive sales team so he knew every deal. Every deal that mattered that could swing our forecast, he would say "I've been on those calls. I know the difference. I can choose to commit it or not." What this doesn't do is it doesn't teach the reps and the managers below to own their forecast.

The second stage was basically to say, in the next round the manager has to commit a number and roll that up to our head of sales. Then from there, the managers would say, "Well, you AE, every week, you tell me that this deal is not in commit. I sit on the calls with you, and pretty sure this one's going to come in. It's not written in blood yet. It's your job now to commit." So we trickle it down from the top to the bottom.

RB: Excellent, that's a really important point. What you're describing is, you're creating visibility through the spine. For example, what we often see is this hedging scenario where a rep might commit six deals and a manager comes along without having a discussion with the rep, and reports up and says, "Actually, I think this rep is going to close these four.” Would you say that you're actually keeping it an open view for everyone to see?

DM: Yes, we review the deals in a spreadsheet every week and we look at the deals. We're not big enough that we have to look at a massive spreadsheet. The visibility of deals that swing our quarter is on the magnitude of dozens. We were able to look through those and understand there's not enough volume for a manager to send back up to his manager.

RB: Yet.

DM: Yet, that's right.

RB: Awesome, when you started to roll it out to the people themselves, were they involved in the design process? Were they just, "Okay, we've built it now? Here you go. Let's onboard you."? How did that process work?

DM: I think for us, we let the managers handle that all the way down. We certainly did enable them, we made it very clear what the categories mean, but we left the reps to forecast as they would and then the managers basically do the correction.

One thing that we did was we ended up sending automated emails to all the reps once a week, "Your forecast is due tomorrow. Here's what I'm currently showing you're at, please correct me if I'm wrong." It adds a little bit of an extra motivation for them to realise that this is top of mind every week.

RB: Brilliant. I'm interested in the results. You've obviously gone through this process. How long has the new system been done?

DM: We've had an iteration of it now for probably three quarters or so.

RB: When you're thinking about rolling out a project like this, where does what success looks like start to come into your mind? Do you agree that upfront, or is it, well, obviously, success is forecasting, so are there other things that you consider as well?

DM: Yes, for us the success is, ultimately, predictable revenue. We should prove to ourselves that at the beginning of the quarter, we have a decent idea of where we're going to land. No one is necessarily held accountable or paid on whether our forecast is correct, but we look back at the beginning of every quarter to the prior quarter to make sure that our process didn't break and we didn't miss by a huge amount.

Last quarter, we had a big deal that came in on the last day that, literally, the sales process happened within two weeks of end of the quarter. We were never going to get that right, and we have to acknowledge that parts of that are just going to happen.

Our head of sales had us landing right around 99% to 100%. That deal put us at 108% or 107%. Huge deal, but we had no idea. We call it the bluebird. For us, it's just saying, "Do we have a path early on so that we can course-correct?" The mandate that we were given, at the beginning, was always closest to the pin. We don't want to see the forecasts just tick up every week because the methodology is wrong. We want to be accurate from the first get-go to the end of the quarter.

RB: Fantastic, this is a part I really want to dig into and talk to people about a lot as a business as well. You get an understanding early, because you've got this system in place as to what might happen. Let's say for example, right now we're 85%. Once you've got that set up, you talked about levers you can pull, and I imagine it's different per scenario. Could you give maybe some examples of when you're talking about levers, what are you talking about? What levers could be pulled? What conversations might be going on in the room?

DM: The pipeline levers, when you talk to a salesperson, could be hundreds of things, but it all really boils down to a couple. You can either increase the close rate, you can increase the pipeline, or you can increase the landing value. Everything has to map to one of those things.

What we'll then do is say, "Which of these things seems maybe the most off?" For example, it seems like pipeline may be the issue. We may say, "This week, we know that inbound is down, and we know that we need to increase inbound because it usually delivers X. Therefore, our focus for the next two weeks is going to be juicing our pipeline." Then marketing, who have been involved in all these calls, will come back with a plan to say, "We've registered for these virtual events that we're going to do. We're running a website test that does this, and we expect pipeline to grow by Y." That's one thing that we can do.

Another thing that we can do is, there are certain deals that perhaps weren't forecasted, and so you want to increase your close rate. What that means is, can we pull in deals that are taking a little bit longer? That may mean having our CEO email them. It may mean sending them a gift basket. It may mean giving them a better price, whatever it may be to accelerate. We basically map our levers to the different actions that we know can impact that number and then designate an owner just to follow up in the next week's forecast meeting on progress to that stream of work.

RB: I love that. The C-suite is connected to the deals and the forecast and where that's going to help the company finish. They can be pulled in and involved if necessary to help that scenario rather than just watching it unfold.

DM: Mathilde, our CEO, she loves this kind of stuff. She likes talking to customers. I think if you do a poll of CEOs who like talking to customers, it's all over the place. We're a type of company where we want our executives to be talking to our customers and our prospects. If it means that we have to borrow our head of product's time to get on a call, we do it. If it means that Mathilde, our CEO, needs to hop on because we want to show a certain prospect that they matter, she'll hop on. We reserve a lot of her time for that because oftentimes, your best marketer and your best salesperson is your CEO.

RB: I totally agree. I think a lot of people reading this will look at that and think, "We could do that so much more." I’ve heard of a couple of companies that do it really well, actually. It's a big lever because your CEO could influence two or three deals a month and if they're big, that could mean a lot.

DM: She has to sometimes beg us. She'll be like, "Put me on more calls, guys. What's going on?"

RB: That's what you want.

DM: "We need to escalate more customers." I think that's a really good muscle and a type of culture to have.

RB: One big question here. You rewind three quarters, they say, "David needs to build us a new forecasting process. This isn't good enough." What would you do differently if you did it again?

DM: I'll actually talk about our post-sales forecast, which is a lot more complex than our new business forecast. Post-sales, not only has to include the ups but also the downs, the churns, and the renewals. Obviously, it gets a lot more complicated. Our post-sales business also has this complexity, what we call organic growth, similar new business where customers can go into product and hit the plus button, and we would never know until they hit the plus button until after we’ve billed the customer. It's been complicated for us to figure that out.

The surface area of potential areas that we can go wrong is a lot bigger. What I would have done differently is think through more of that process from the beginning. I bolted on the things that I thought existed. I know renewals is a thing. I know expansions, I know organic, and I forecasted it and thought that was it.

In the middle of COVID, my forecast was so wrong that we had to get our investors involved. They were like, "We're going to set you up with some people that really know how to do this." What I would have done differently is, I think I would have been a lot more thoughtful at the beginning about what the different levers are and map that out. I would have designed a much more robust process that understands the different levers in our business better.

When you're talking about missing the forecast by a magnitude of more than 10%, 20%, you're now talking about changing your revenue trajectory which has an impact on the business in terms of hiring, in terms of all sorts of things. People often think of forecast as just like, "Oh, it's a number. It helps us predict," but the ripples are huge, and getting it wrong and underestimating its importance can have a big impact on the business.

My learning is, if it's important, then you have to spend time on it. If you want to get it right, you have to spend a lot more time at the beginning. That's what I totally did wrong. If I were doing it again for a new business, I would spend a bunch of my first 90 days or 30 days really thinking through the levers and designing a process that makes it clear how to forecast that.

RB: In that circumstance, the levers you're talking about were presumably, the data input. There were actual levers that you'd missed out, for example, that plus button example. Is that something you just hadn't considered?

DM: That one, we considered. We basically were underestimating our forecasted churn because customers were actually starting to remove seats in product. What we didn't realise was that there was amount of seats that were going to be lost in the future, that were not in the Salesforce opportunities that required a projection.

This was a lever that was not considered before COVID because it was a very small volume. Suddenly, the world turned and people were just flipping seats out at a relatively predictable rate 12 weeks out into the future. I just totally missed that as a lever because it wasn't a problem before.

RB: Interesting, so it was a very subtle lever, a very subtle trend, that maybe if you'd been more forensic, you'd have spotted, but obviously, it wasn't that obvious. It became obvious when the market changed. One could be forgiven for that.

DM: Yes, I know. I took it pretty hard, and it was fine. It's hard to predict that the world was going to turn that way, but the way that I would have thought of it is, if I were doing my job the right way from the beginning, I would've thought of it.

RB: Yes, that's a very good way to look at it.