Data for subscriptions podcast - Episode #2
The future of the streaming video industry is usage
Guest
Stephen Hateley
Head of Product Marketing at DigitalRoute
Stephen Hateley
Head of Product Marketing at DigitalRoute
The growth of the video-on-demand industry has stagnated in the past year or so. Video subscription services like Netflix are struggling to gain new subscribers and even maintain their existing customer base. How can these companies keep growing and sustaining their revenue stream?
That’s what we are trying to answer in this episode of The Data for Subscriptions podcast. Stephen Hateley, Head of Product Marketing at DigitalRoute, joins our host Behdad to discuss how companies like Amazon Prime, iTunes, and AppleTV can manage subscriber churn and reinvigorate their growth in the current SVOD landscape.
We are all familiar with the most common or the most popular approach, the subscription video on demand (SVOD), or what Netflix is doing. When you talk about video-on-demand, everyone sort of jumps to this. The subscription video on demand or SVOD is largely geared towards customers who want to avoid broadcast TV and the ads and are willing to pay a flat rate for it.
In this episode, host Behdad Banian speaks with Stephen Hateley about new modalities and trends we can expect in SVOD.
Well for starters, the industry got too comfortable, and now we have way too many players. Subscribers have just too many options to get their entertainment and can easily switch from one to another.
On top of this, many are starting to realize how much they are spending on subscriptions every month. They’re facing a subscription fatigue; paying for too many options but being unable to use all of them.
The realization is prompting many to cut down on their subscription. And of course, we have the password-sharing problem that VOD companies have faced from the very beginning. Many are starting to realize that a significant chunk of their viewers is not paying for it.
Netflix is, quite controversially, adding an ad-based tier, which only time will if it will be successful because it really can go either way.
On one hand, you’ve got people who think they don’t want to watch ads at all, who see that as a selling point of video-on-demand.
On the other hand, you have viewers who won’t mind watching a couple of ads if it means getting a cheap subscription. And most surveys suggest that the split is really fifty-fifty between these groups.
Of course, it’s not just Netflix, other companies are looking into more modalities as well.
Well, let’s say I’m the CEO of a video-on-demand company that prides itself on subscription-only content and our customer and revenue growth has stalled. An ad-based tier is not an option, because we’d lose more customers than we’d gain.
So, we need to optimize the revenue per view and create an offering, maybe in a pay-as-you-view or pay-as-you-binge format. But at the same time, we’ll have to offer something more for our existing subscriber base.
We’ll also have to build better relationships with content providers as well. For this we need a granular understanding of our customers, create a light model or a cheap subscription to entice people to join the platform, like a freemium or freemium+ model, build an enhanced experience for premium subscribers, and motivate content providers to help grow our library with smoother payments and settlements.
Unknown Speaker 0:11
Hello, and welcome to the data for subscriptions podcast where we focus on how to succeed with subscriptions. And as a service businesses. I’m your host, Ben Bornean. And today I have the pleasure of welcoming Jonas, strategic Product Manager at Digital route. Welcome. Nice to be here. Thank you for that. Great. So today we’re gonna focus on front runners of this industry subscriptions, usage based pricing, as a service, all the stuff that can we kind of talk about in one form or the other. But before we look at the exact company, let’s kind of level set a little bit on these different terms. So let’s go right at it. So usage based business or usage based pricing, what is that? It can be a very long answer. But the short one is basically that you price and charge for something a product or a service based on your consumption of the service instead of flat monthly fee, or just outright selling the product. Let’s talk about some of the obvious benefits were new move to this type of business model. Yeah. So why would you do that? I think one of the maybe underlooked reasons, which I personally like a lot is that it’s fair in the sense that I, as a vendor want the customer to stay with me for a long time. So I want to provide quality service or product that has a long lifetime, because I keep earning over that long lifetime. That’s why one thing that makes it really attractive, there are other more tangible economic benefits, like you reduce the barriers of entry by making it really cheap to get started with your product. When you’re not using it so much in the beginning, then you can do the classic land and expand strategy where as adoption grows, your revenues also grow. And in this way, an upsell, which can be work, either through nudging and advertising or through sales outreach becomes on goes on autopilot, whenever they use more, you get more revenue. So that’s another big benefit. It also aligns sales incentives, right, instead of a salesperson chasing the next sale, and then moving on to a new one and sort of forgetting about the old customer. Now, to keep generating that revenue, you need to stay with your existing customers. And that’s another way to aligns incentives. And then finally, for the shareholder parts, it has been quite clearly measured that the companies who get this right tend to grow both revenues and profit faster than those who don’t adopt usage based pricing. So it’s as it has a lot going for it. Have you seen during the last two or three years with the pandemic? Have you seen a difference between companies that are providing usage based pricing versus those who are not if they’re, they have performed differently, I think that we see, it’s kind of this subscription exploded during the pandemic, right, we everything was a subscription, everything was online. And then we have subscription fatigue. Now, streaming services is like the poster child for this where you can’t have 15 different streaming services. So a usage based model might be more attractive. So think that that’s one thing that has happened, we got more subscriptions during the pandemic. And now there’s so many that usage based pricing becomes a more attractive way to keep those customers instead of churning them.
Unknown Speaker 3:44
Let’s talk about some of the challenges and cons. Yeah, so like, why would you not do this? Well, one of them is peace of mind, for the customer, maybe it’s safer to know that you’re always paying a fixed cost every month, like a flat subscription fee, maybe maybe the customer base becomes nervous about not knowing exactly what they’re going to pay. That could be one reason your CFO would be a bit freaked out. Because the finance and revenue flow will become less predictable than if you have flat subscriptions. For subscriptions is easy, you just like tier times number of subscribers and you know what’s going to happen next month with usage based pricing, it can vary a lot more. And then it can be a bit more technically complex to get right. There’s more data processing more systems that need to work properly. Also, sometimes companies struggle to find the right metric, like what what feels fair for the customer. So that those would be a few drawbacks or issues. So this predictability question that you mentioned, is that really a concern to have or is it more of a timing question that you know, when you draw a timeline, then the predictability is not an issue while in the shorter timeframe or window? It can be a very good point. I would argue that when you start out it
Unknown Speaker 5:00
It is more of an issue because you have a small customer base, and the variations are more visible. Whereas if you particularly if you’re selling business to a customer, or you have a large customer base, the law of large numbers eventually starts cancelling the variability out. And then it becomes more predictable again. So it’s particularly noticeable for a small customer base, or an early step in this journey. So what best practices have you seen? Well, first, spend enough time to research your metric and actually figure out what to price on it can sound trivial, but it’s not always like that actually interview your users, what is the thing they feel they get the value from, you know, and then I think you need to talk to your CFO to your shareholders about this revenue, stream behaviour change that we will maybe see a dip for a while, we may get less money from our small customers, long tail customers, we may get more revenue from the big customers, but that’s going to shake things up. And also that if we do the journey, right, you will eventually see a kicker where revenue starts picking up faster than before. But that might take a while to get there. And like steady through all the course through that period. And often forgotten part is that you need to talk to your salespeople. So they don’t do what we said before, they just take one customer go chase the next go chase, the next that does not work well at all, with user base pricing, you need to make them feel responsible for the long term success of your customers. And that can be difficult because sales compensations might change, there will be a lot of stress in the sales organisation. So that’s not to be underestimated. And then you need to be on top of your it. So let’s look at some of the companies that you’ve looked closely at. So we’ll start with the first one. Yeah, sure. So I took a look at one company that I we use here at Digital route, actually, it’s called Data dog. It is a SaaS company that sells observability services basically for it. So they help you to see what’s going on with your SOS has insane growth story over the last few years. And I think it’s inspiring. It’s also a really nice product to use. So what did they do? They started out a few years ago with usage based pricing, actually from day one. So they had a very simple metric, I will charge you based on the number of hosts that you monitor. So if I have one server, I pay for monitoring one server, if I have 100 servers, I pay for 100 service, easy to understand. So they did work on their metric. And they grew, or this model, they did like $20 million 2015, all the way up to 1 billion USD last year. So the beta has been an insane journey. And of course, they have a really nice product. Product is kick ass, the users love it. But the pricing was also a factor. So she will look at the pricing. Yeah, have a look. We have it up here on the screen even. So oh, there we go. So here we can see that they started out with one service only. But today they have 17, I think Yeah.
Unknown Speaker 8:27
So what happened here, they found adjacent observability needs to their first service. And they started selling those two. And all of these don’t have the same usage metrics. So they have also taken the time for each of these services to research like, what is value for this service? What is what am I giving my customer for this service? And how can I price it? So if it’s normal log retention or whatever, it’s still per host. But if I’m monitoring serverless application, there is no server, right? So I can’t do it by hosts, I need to do something else. And I can price it by how many functions or invocations Am I having? Or is it by security threats? Then okay, how many logs Am I crunching for my customer to find the threats and so on. So they have done that work in order to figure out how to price these 17 services. And what they do with this is like a smorgasbord as we just worked, where they have taken like an approach similar to Amazon Web Services. They have a usage based pure usage based pricing for each service. And then the user can mix and match using only the services they need at a given time. As their needs grow. Typically, they would add more services to their menu. So this is a very, it’s not just usage based pricing for one product, but it’s also usage based, adjacent product expansion. That’s really a good way to grow, I think for a SASS company.
Unknown Speaker 10:00
and this has been interesting. Do you recall where which which service they started with initially? Now, you caught me red handed. I think it’s the infrastructure one, but I wouldn’t be 100% Sure, do you know the pacing of introducing new services? Well do the math, right 17 services in six years. So three new per year, roughly speaking pretty fast. Yeah. But it’s also cool, right, because each service is not mega super complex, it’s doing one thing. And then they of course, have their in dashboard, where you can see all this stuff in one place. So they kind of plug in each new service in an existing framework in a really good way. And because what I find fascinating with companies like data dog, or any of the other examples that we’re gonna get to talk about over time, is in traditional businesses, you spend a whole lot of time understanding your customers, their needs, you doing surveys, you’re doing, you know, different kind of user groups and whatnot. And you often speak about that you need to be fast and agile, to introduce new new type of product or services. And you want to maintain a steady relationship with your customers, with these companies that are providing true usage based pricing. That is the basically the fundamentals for
Unknown Speaker 11:10
many, the type of service provider, you’re constantly getting feedback, minute by minute, day by day, based on consumption from customers. Oh, yes, it is, like I sometimes call it like it’s mass customization, or pricing, you know, when you when you buy the Nike shoe, and you can choose the colour and the logo or whatever, or when you buy a car and you design it, choosing your components. That’s customization, right? It’s the same factory, but it’s churning out lots of different things. usage based pricing is mass customization of pricing, every customer gets actually a different amount in on their bill every month, right? It’s customised. There’s no tears, there’s no like static mould I need to fit into, I will get a customised price just right for me. And I think that’s one of the strengths of this model that you can tailor it to so many people and not lose people who get squeezed between tears, or don’t feel like you’re either paying too much or getting too little. And this is really interesting. Do you have any insights in terms of the value based kind of pricing that you have to look at when you provide a new type of services? How you go about finding the right rates? How to come? How should companies think about that? Yeah, so that’s there’s a bit of science but a little bit of magic to it as well, I think so obviously, you need to talk to your users a lot and and test it, and not just test it by talking to them, but also testing it on them, right. So there’s a two step thing there first, talk to them to research, figure out what you think will work and then you need to test it. There is why this is why the SAS industry in particular, has adopted usage based pricing so much because it’s very easy to run bluegreen AB testing in this industry, you can change this pricing for a few customers, and see if that increases adoption or not, you can tweak it. So so that’s a very powerful tool that you must use. If you’re doing usage based pricing, otherwise, you’re gonna miss out on revenue. What else did you wanted to share? What did it I wish it would go to the next case you had in mind? Well, we can we can talk about data log for a long time, but maybe just a couple of interesting benefits that they got out of this. Because it’s pure usage based some some companies have tears and usage. Yeah, this is pure usage. And I that’s interesting, because what it does, fundamentally is like we’re reducing the number of user decisions that need to be taken here. I don’t need to decide what to buy, except I just want to buy data. That’s it, then I mean, and that’s powerful. That’s really powerful. Actually, it’s just like the credit card, you don’t see your money going away anymore. So you took away a decision point. This is sort of like that, psychologically speaking.
Unknown Speaker 14:01
And upsells don’t really require human outreach. You can still do it for for large accounts, but you don’t have to. It’s just more usage. And you can prevent predict churn based on the usage as well. You can do cross sell recommendations based on us, which looks exactly what they’re doing here. Right. They’re positioning new services based on the other ones.
Unknown Speaker 14:24
And you don’t need to do contract negotiations to consume more services all about this reducing friction. Like that’s something you need to keep Centre in your mind when you’re designing this that you need to reduce friction points. And another consequence, I think, just to be expected this happened to data dog, it will probably happen to you as well, is that you will find that your long tail of users will give you less revenue than before, whereas your top customers will start giving you more revenue than before and both of those are imported.
Unknown Speaker 15:00
And you can’t just cater to the big ones, you need to also care for the long tail because those will some of those will grow to be the next big users. So it also it skews your revenue mix quite quite a bit. Yeah, we could go on talking about data, how does their customer mix look like they get more and more revenue from their big customers nowadays, actually. And growth rate of small customers is much higher than for the high revenue customers, obviously. But I think now it’s like 80% of revenue comes from the bigger accounts, which is a lot less than five years ago than there was about half half. So they have increased their revenue from large accounts. And also they have increased the number of accounts using more than one service. So that’s also something that often comes up when we talk about usage based pricing like, do we have product led growth or not, right product lead growth being like I build the product and market the product and use yourself sign up for the product. It’s not really a sales lead, go to market strategy. And that’s, I think, a lot where data dog started out. But as their accounts grew in size, they have added sales, even enterprise sales into the mix. So they now cater to three different segments, like the small self sign up users product lead, the outreach team for the mid tier, and then the enterprise sales, heavy, interactive approach for the large accounts, which is a pretty natural journey. And if everybody who’s successful with this will probably go through the same. Alright, so before we leave data, dog, I just want to leave you with a really nice quote from their CFO. So his description of what they were doing regarding to pricing was this. For a long time, we relied on a bunch of really smart people putting together a lot of very complicated spreadsheets. And I mean, end quote. And eventually that broke down due to complexity of products, quotes, usage contracts, so they automated their processing, which is a very natural growth evolution. So yeah, it’s worth thinking about, we’ll come back to this one, because we have, obviously the finishing touch, which is the execution. So we’re gonna keep this in mind. So let’s go to the second use case study that you wanted to do. So we’re going to flip to machine. Yeah, let’s roll over just to not only talk about the everybody’s sauce darlings, like snowflakes, and Amazon. And that’s already been done 1000 times. So I figured let’s do something differently. And look at Michelin, the tire manufacturer. It’s an interesting case, because here we have a physical product moving to usage based pricing based on IoT, because of course you how do you measure the usage, you need to know how much that tire is actually revolving, right? So Michelin has done a lot of work adding sensors and connectivity into their tires. And once they had this in place, then they could introduce usage based pricing, right? Do you know how many revolutions a tire has, you can then charge the tires by mile as a service, instead of selling tires? It’s pretty cool. When and I mean, it’s what the use of one size, less hassle, they don’t need to care about the state of the tire, the tire will report when it needs changing, you get a new tire you stay with Michelin, you become a loyal customer. So I think it’s the right way to go. Also, if we talk about alignment of incentives and the fairness of the usage based pricing, totally coincidentally, right when Michelin started introducing usage based pricing for their tires. They also launched a new series of tires with a longer durability time than before, significantly longer. It’s good for everybody. So this aligns incentives in three ways. Actually, customers get tires that last for longer. That’s great. less maintenance, interruptions, less hassle. Michelin earns more revenue for each from each tire because it lives longer. And hopefully also we produce fewer tires. So the planet wins because it’s less resource heavy on our earth. So it’s a triple win win win situation. And that’s something that is particularly evident when we talk about physical products going to usage based pricing. It has it’s harder than for a sauce. But when it works, it’s also better from a sustainability perspective. Yeah, that is interesting how incentives change right now. Yes. So what are the benefits? I mean, you talked about a couple of things. Obviously, the main benefits of moving to this business model. What kind of additional benefits have you seen what they’re moving? Well, I think I’m not sure because I haven’t talked to Michelle indirectly. But I believe that just as many other IoT projects, it’s actually didn’t start out as you
Unknown Speaker 20:00
usage based pricing, it started out as real time monitoring and predictive maintenance service. So if we take mean you think of Michelin as making tires for your normal car, right? But Michelin does a lot more than that, for example, they make tires for these really insanely big mining trucks that hold tonnes and tonnes of rock around. So they have a metric, which is called tonne kilometres per hour was like that is the coolest business metric I’ve come across yet. And in that environment, if a tire breaks, you can imagine the amount of revenue you lose while everything’s halted, because there’s a huge truck blocking the entrance to the mind. Right. So monitoring those tears in real time could prevent revenue loss, quite significant revenue loss. And I think what happened to Michelin just as many other companies is that once they had that data coming in from their tires, somebody got the idea, alright, so could we not also based pricing on this, give the customers this peace of mind of not owning the asset, but actually having somebody who could come in and fix it for them and have the whole lifecycle in the vendor instead of at the customer. And then once you have that, you can you have everything you need to do the usage based pricing you your first step is always say like, you need to connect your asset, you need to collect the data. And then you can go from there in different directions, predictive maintenance, alerts, monitoring, and eventually pricing. But one thing that also is different from a sauce when doing this is that you have a drawback, which is like, if I’m a factory, and I’m making tires, my whole life is centred around, supply chain in, tires out.
Unknown Speaker 21:49
Now, if I turn into a usage based priced company,
Unknown Speaker 21:53
suddenly my life turns into sending tires out monitoring how the tires are doing, going over to change the tires when they break, bringing in old tires, figuring out if I can recycle the tires, and refurbish them and use them again, or put them back as in my supply chain. Right. So that whole part of the operation also changes, this is a bigger deal for a physical product company than it is for a pure SOS software company. And that’s definitely not to be underestimated. So this is a really fascinating example, because the final point that you mentioned in terms of how we fundamentally shifts a company from being one type, to becoming essentially a data centric company in many ways and moving, what it really means to move to a service based. Alright, you and I said let’s talk execution. What do I need to get started? And one of the items we discussed before was, Do I need purpose built software? Or can I just use any generic data integration data management tool to get this done? Yeah, that’s a great question. Many ask how to get started. And I think if we start with a software question, you don’t necessarily need purpose build software, we see many companies who start out with Excel that don’t do that. But we see companies doing that there’s a lot of homegrown solutions, generic data integration software, using data lakes, and so on. And this can work for a start, but they usually tend to hit a wall once their usage scales, and then they reach out to us. And I don’t know if you heard that quote about like data, scientists spend 60 to 80% of their time cleaning data in those companies. That’s what happens. I actually wrote a LinkedIn post about this called, you won’t get it until you’ve tried and failed. So that is totally directed to these companies and the pains and the symptoms that they tend to experience with this approach. Other companies, they do their research from day one, they understand that this usage data actually represents revenue, and they really need to handle it well, for various reasons. They use our software from the beginning. And they scale from there confidently. And what I mean by scale is, let’s take an extreme example, reliance, geo a telecom operator in India, they used our software to scale from zero users to 400 million users in four years. All of it is being charged in real time.
Unknown Speaker 24:18
Those are hundreds of 1000s of use search events per second being processed at reliance, and they did their homework, they knew they were going to explode, and they put the infrastructure in place from day one. But bringing that back to maybe scale of most companies because realise is big. Any company that wants to make revenue from usage has a version of the same quote to cash processes, same revenue generation process the same way to get money in their accounts from usage data.
Unknown Speaker 24:53
So it’s a global need. It’s Regardless of industry, the same place to very modest way
Unknown Speaker 25:00
have you to say you can use Excel or alternative options to get started, but with the examples that you gave to take the data scientist time that you spend just to clean and normalise the data sounds like an awful business case to me, if you then take the example of Reliance you Yeah, it’s an extreme situation, but just gives a very concrete proof of when we say scale and complexity, what we’re dealing with, to me, it sounds like, you know, depends on if you want to get it right. Or if you want to really try a harder path, if you ask me. But suppose then we take some of the instances with customers who are already on their journey, they’re using an ETL data integration tool, and they say, Why do I need to move? In my case, towards a purpose built? Software? What do you say to them? That’s a very common question as well, right? You already paid for something you have it in house? Why would you use another tool, I usually advise them to look at two things. Two things. One is the total cost of what they’re trying to accomplish. Like, if we put aside for a moment, operational problems and stuff, like just look at what is the cost of the software, but what is the cost of the integration work that you need to do? Probably with a system integrator. And just look at that typically, when we are brought in the integration time is lower, and therefore the spend there is lower. Whereas if you use a tool that’s not really purpose built for users data, integration, time tends to be a lot higher, because you need to custom build much more things in your solution. So that’s one look at the total cost. We actually see ROI. So six to 18 months, a little bit, depending on the circumstances, which is very short.
Unknown Speaker 26:49
There, the second thing to look at is revenue leakage. So like, what can you automate? And what can you prove? When about the usage data that you are processing? And here, many companies believe that now, we don’t have a big problem here, we come in and automate the processing with our software, and then we find out that there’s anywhere between three to 12%, revenue leakages, you can imagine, in extreme cases, there’s actually pays for the software within six months or even a week, you have even seen faster payback times, actually. Because you see where the revenue leakages you’ve automated away, and you can prove, after the fact that yeah, there is no revenue leakage share anymore. That is valuable. So, wrapping up a little bit, your your question is that, yes, we can do this with other tools, but you may not necessarily get the results that you’re hoping for. That sounds quite clear. Alright, so leave us with your step by step Action List of how to get this done. Where do I start? How do I perfect my execution? Yeah. So I think first of all, before you get into software or data, there’s a couple of things that you need to do to get right, figure out your whys, whens and preconditions to your specific market won’t go into detail. But this is really important, you need to get buy in from the top, your shareholders, your management, and you don’t need to go all in from day one, you can start small, all good advice, solid advice we have seen this work before, once you get to that point, then you need to start measuring the usage. And it’s got a couple of parts, right, you need to connect to your usage, this can be as simple as just reading a log in sauce, or it can be actually an IoT sensor in a Michelin wheel or whatever it right that’s complicated for physical products, but it needs to be done. Otherwise, you cannot do it. And then when you have it, you need to clean it. So either you dump it in a data lake and let your data scientist loose on it, and they have lots of coffee, or you cleaned it up on the way in. So it’s actually ready for business use from the start. And your data scientist will love you trust me. So that’s something you need to do spend some time on, then you need to find a pricing metric. A lot of work there too. I won’t go into detail, but figure out what works for you what’s valuable, and then simulate, use your metric. What would have happened if I had used this for usage based billing based on my current consumption patterns. This is a great way to build confidence internally, you can like shadow run your billing, show it to your CFO, look this works. What are the use revenue predictions from this? This is a way to get buy in internally. And then the last page is just make sure you also test that your automation, are you leaking revenue, can you prove end to end that usage comes in revenue comes out. It’s all accounted for many don’t take that seriously enough and then you end up with revenue leakage again. So so that’s really important actually. And finally, make sure to make and get your salespeople on board. And then when you have all that ready then
Unknown Speaker 30:00
You should launch it. usage based pricing.