We're going to cover this final element of the Canvas cost structure. And here's how it might look for Enable Quiz. They have this assessment development of creating the quizzes. And this product development of creating the platform that the users use to run them, and they use to manage them. Both of these are what are called fixed costs. So they're relatively fixed, meaning that if they added another customer, they don't really incur a lot of additional costs here. Whereas a variable cost is one, like promotion, where especially if they're using paid media, especially if they're using pay per click media, that's highly variable. So we spend two more dollars, we get one more customer. We spend another $2, we get one more customer. It's variable. Whereas the product development is, we built it, now let's try to sell as many people through it as possible. That's one obvious basic thing to think about with your costs. These general and administrative are usually fixed costs. This is like accounting and HR, things like that that are part of running a company. I mean, they're a little bit variable with how big your company is, but they're relatively fixed. Support is another thing that I didn't put in here. And maybe consulting or any kind of onboarding, things they do by hand. What do you think they are? I would say variable because they correlate pretty directly with how many customers they have. More customers, more support, more onboarding help, fewer customers and fewer. So you want to think about that and how these things relate to revenue. That's what's really crucial. A good lens to come back to for this might be these three sort of fundamental engines of growth. In that they're a good way to think about how we need our revenues to pair with our costs. If it's paid, then we're really looking at the relationship between promotion and driving subscription revenue, which is really their main source of revenue. They've hypothesized that maybe they can get referral fees if they sell other stuff or advertisements on their platform. But that's not something they're focused on right now because it has nothing to do with their core product market fit. And it's not even clear it has to do with their business model. The paid engine is, really, we're looking at the relationship between these two things. Is our cost of acquiring a new customer less than the lifetime value of how much we earn on that customer? If the engine is sticky, what we're really looking at is this lifetime value and how it correlates with the way that we handle things. Like things that happen in the customer journey, and that we may have to spend extra money on support, or even kind of proactive consulting and stuff like that, which are variable. And is that actually working? If we call a customer that we see is inactive and show them the love, and ask them, hey, can we help you use the platform, what's happening, rather than just sending them an e-mail bugging them. Does that lead to more activity and a greater lifetime value, or not? Those are the kind of things we might want to test and focus our testing on. And viral, we're really looking at this customer journey, and our work on promotion is going to be a little more fixed. Because what we're trying to do is organize the customer journey probably online, so that the customers, we're doing experiments that increase their propensity to share and talk about the platform organically and drive revenue. So we're going to have maybe more of a kind of fixed cost since we're basically an infrastructure company and we have a relatively fixed customer journey of trying to amplify this, at least in Enable Quizzes' case. As you think about costs, another thing that I would consider is, and a place where costs get out of control, especially with mature companies is, if we're building something new and we don't know it's innovative. So by definition, we don't know if it's going to work. Are we thinking about costs in a way that minimizes the uncertainty around them? For example, if we're going to build a talking bicycle compass, the kind of, I'll call it scale friendly legacy view of how we would do this is, we write a long plan, we allocate budget to it, we get the costs down as low as we can. We package it up and worry about massively scalable distribution. Maybe we hold focus groups. And then, we hope that it's successful. But any new product, especially in today's hyper competitive fast-moving environment, has only kind of moderate odds of success, individually. Whereas the more innovation friendly approach is we go out and we observe personas and problem scenarios, and we just start from that. We learn about those things. We find out what's really on their A-list and whether we see an important problem that has a weak alternative. In which case, we use a product proxy to test whether the customer's really engaged. For example, with the talking bicycle compass, maybe we hand a phone that we've made a phone call to and turn on location tracking, to a cyclist. And we can track them and we give them verbal guidance when they need it, as they're taking a trip. And we see if they like that or not, they care about it. Wildly impractical, not scalable at all, in any fashion, but very innovation friendly. because it helps us not waste a lot of money on potentially doing all this stuff and getting something that doesn't sell. And then in the innovation friendly environment, we are iteratively scaling up our development of this thing based on the evidence that we see from the customer. And so this is a little bit of a kind of macroscopic approach issue that will also help you minimize costs. And I would be sure to think about that if your business model innovation project has to do with going out and doing something new and rethinking the business in that fashion. So those are some ideas about costs, where to focus, what to consider, questions to ask yourself.