Building a Streaming Platform - Alternatives and Decision Making

What is SVOD?

SVOD stands for Subscriber Video on Delivery. Products like Netflix, Prime Videos and Disney+ are SVOD platform (more or less). It’s one of the possible variations of OTT (over the top) content products.

OTT just means that’s its a streaming media service offered to viewers over the internet. The revenue model is really what determines whether the platform is S(Subscriber eg Netflix) VOD, T(Transactional eg Sports Websites where you pay per game or season) VOD, or A(Advertising eg. YouTube) VOD. The underlying technology for video streaming remains the same.

Some time ago, a client reached out to us for help building a SVOD platform to provide lessons to people in the [redacted] industry. This article is a summary of the options that are available for entrepreneurs or businesses looking to build a video or audio streaming service as well as their respective tradeoffs.

Why is it so effing difficult?

It is difficult to get OTT video right primarily because of the unpredictable variations in the viewing experiences. Some of the things businesses need to consider are:

Location - where are your users located? Businesses need to invest in a distribution network that facilitates smooth access to content for those areas. This means you need servers, databases and a delivery network that allows your users to reach your content.
Device - what devices are your users using to view content? There are hundreds of different devices - computers, tablets and phones - that users could be using to try to access your content. Those devices don’t all support the same video formats. You need to create replica videos for all your content in different formats and write logic to serve the right content to the right device. More importantly, you need to figure out which content doesn’t work with what device.
Bandwidth - what kind of internet speed do your users have? What happens when they are trying to access content from a 3G network? Businesses need to create varying levels of video quality and ensure fallback happens smoothly.

What are my options?

There are generally three major routes a business can take when building a streaming service. Each route has trade-offs in upfront capital investment required, time to development, and long-term costs.

Infrastructure as a Service (IAAS)

This is the heavyweight option and the most customizable. It is also the most expensive to get up and running.

All the major cloud infrastructure, Google Cloud, Amazon Web Services, Microsoft Azure, etc, providers have functionality that allow engineers to build their own Netflix. The fixed costs to build infrastructure there is relatively low - a few hundred dollars - while the variable costs will change depending on how much usage your platform is getting. I don’t expect a new business to exceed $100 / month in usage costs.

The real cost drivers are the engineering hours you need to put in to solve the three problems we discussed above. These are the kinds of problems that can only be solved by specialist teams which means you’ll likely have find some engineers who are deeply familiar with Netflix’s (or competitor’s) infrastructure. Your costs could vary anywhere between $200,000 to more than $1,000,000.

So when should you pick this option? When you’ve proven your model and know you’re gonna hit that hockey stick growth. When you’re sure that a million dollar investment is going to be worth it.

Platform as a Service (PAAS)

The complexities of OTT content delivery are so difficult to solve that businesses have formed that offer this as a service. Companies like Anvedo and Wowza offer platforms that have been battle tested as a service. These platforms make it possible for the business to invest engineering effort only on the business logic of their platform with the complexities of OTT - bandwidth management, distribution network, and device variation - being offloaded to these providers.

For example, if you wanted to build the next Netflix, you’d invest your time figuring out how to implement that DELIGHTFUL ‘Skip Trailer’ feature and build a solid recommendation engine while leaving the content management to Wowza.

Costs vary here depending on what kind of functionality you want to build onto the product and the kinds of devices you want to support. A simple product with nothing besides authentication and subscription payments would cost less than one with a recommendation engine and review system. Building applications for IOS, Android, and TVs will cost more.

Byldd can be especially helpful here. Byldd uses its proprietary tools to automate away much of the redundant parts of building new software products. Things like payment processing, admin and user-level authentication system,s and review systems are modularized and can be added at the click of a button. It also creates the infrastructure to interact with your data by automagically building a complete REST API for your resources. This reduces the time and cost of building a streaming service by 50 to 70%.

Subscription to the platforms themselves has a fixed cost of about $50/month plus streaming cost which is typically charged by the number of hours of content streamed. For new businesses, I don’t expect the cost to exceed $150/month.

Software as a Service (SAAS)

There are other businesses who provide end to end generic solutions for OTT. Services like Uscreen will create a streaming web application for a few hundred dollars a year. These products are the least customizable and you have the least control over your user’s experiences.

These are white label streaming services that put your branding and logo on their existing streaming applications. While this is a great way to experiment, it is not a sustainable long term solution because they charge by the subscriber. The current basic plan on Uscreen limits you to 300 subscribers. You can find your costs expanding very rapidly on platforms like this.

The other drawback is that these providers own the software. So if you ever wanted to implement something custom, like a recommendation engine, you’d have to hire their engineers for custom work.

What option should you choose?

TL;DR - probably the PAAS option.

Long Answer: like everything else, it depends. In most cases though, especially initially, your best bet is going to be using a platform to host and distribute your content. There are plenty of reliable, relatively inexpensive platforms available for use. And if you ever experience the growing pains of Netflix's scale, you could spend time and money building your infrastructure.

Need help building a streaming platform? .