Do You Buy or Do You Build?

Buy or Build for OTT personalisation services?

In the race for OTT providers to grab market share and maintain subscriber numbers, personalisation has long been recognised as one of the key drivers to deliver a loved service and move the dial on customer churn. However, personalisation models are often difficult to get right, balancing emotive feelings about TV content across millions of customers with changing business priorities. Can this complexity be delivered in-house or conversely do you want to risk outsourcing something so important?

Many organisations like Netflix, BBC and Discovery have made the decision to build these services in-house. They have built substantial teams to integrate personalisation into their customer touchpoints. Indeed, for Netflix you could argue that personalisation is core to their brand. The flip side is many other OTT providers that utilise outsourced providers to manage these services, delivering personalisation to complement their existing editorialised content promotion. Buy or build? What are the pros & cons to both approaches?


In-Housing Can Give You Better Control & Ownership

The decision to build in-house is often centred around keeping control. Owning the entire user experience for your customers and the ability to change it quickly is a motivation that any business will have a desire to remain in full control.

Protecting IP is also another big pull in the build direction, ensuring your competitors don’t steal your “secret sauce”. Whether this is your own models or code, many will see this as something they want to develop and own themselves.

Outsourcing can still offer the same degree of control but only with the right partner. A partner who uses an agency style approach, who will work with you on your customer experience and your core KPIs and not just deliver a SaaS solution. Control means the ability to make the changes you want at the pace you expected. You need to find a partner who offers this.

However, the secret sauce is important, and you need to know that the hours of continual optimisation across the many personalisation models that you are paying for will be something you own. Again, the right partner will allow you to understand the code and even the way various models are blended together. Transparency is key here and the right partners will show you what they are doing.

The likes of Netflix shares most of its models and approaches with the world. That is not actually their secret sauce. It is their way of using them and continual testing that gives them a competitive advantage.


Outsourcing Is Faster

Time to market is a key consideration. This is one area where in-house teams can often suffer from a lack of detailed TV recommendation knowledge. Machine learning is notoriously closer to an art form than a development process and so knowing how long solutions that really work will take to develop can be a risk. Many of the best solutions are blends of two of more models such as cold start options for recommendations which are closer to ‘most popular’. Working with an outsourced partner will offer all the benefits of the significant existing experience they have had of successful solutions.

You also need a unique kind of data scientist; they need to combine expert understand of coding recommender systems but also ‘get’ TV. They need to be able to look at a set of results (which are just TV show, kids, sports and movies) and see that they are working for that content and that customer. These people are hard to find.

For the reasons above, time to market is often better with personalisation partners as they bring years of proven experience with the right sort of people and culture and are focused on just this one area. Whether this is metadata enhancements, data management or personalisation algorithms, partners in this space offer experience, technical know-how and market innovation. The result is increased development speed in addition to receiving the very best in technical and theoretical development

Speed of optimisation is not just important on launch but actually is more important when the service is live; on-going, continual optimisation is needed to constantly improve all your KPIs like churn reduction or views per visit. You might have new rails added to your service, new content partners or new customer segments. An experience partner who has seen the problems before will inevitability be faster.


Outsourcing Partners Bring Data

To really get recommendations to the premier level, you need the right metadata that is relevant for recommendations and provides the correct degree of overlap to allow content matching. Whilst many OTT and broadcasters have a wealth of metadata, it is orientated for EPG displays not powering personalisation models.

Whilst no outsourced partners will share your proprietary data with anybody else, they do augment content data usually with their own tags. This was key to Netflix’s success in optimising its recommendations. Outsource partners will usually share these tags between all their partners where the content overlaps thus driving efficiencies.


Cost Differential Is A Close Call

In many cases, the build or buy decision will come down to cost and both approaches will require investment so less of a clear-cut differentiator here. Building the data infrastructure and technical capabilities to deliver personalisation services in-house is no mean feat and will require both significant investment in time, infrastructure costs and hiring the right people. An option is often to re-purpose existing technical capabilities, however transferring skills in this area is often difficult and you need the right mix of coding and TV knowledge.

Whilst the per month costs of contracting with a personalisation specialist can often be a the same as in-house, outsourcing often nets out as cheaper as it is quicker and less risky. The financial exposure from outsourcing is potential rising costs of additional services or costs to serve. “They charge me for every CR”. Again, it goes back to selection and management of any partner; find outsourcers with a fixed fee model to protect from service cost creep.


So, What Is The Right Approach?

As with all these things, there is never one right approach, it will depend on each businesses’ objectives, requirements, desires and resources. There is no doubting that in-housing has some major benefits in terms of control and IP ownership. You are the master of your destiny. But for some, the partner route, with the right partner, can prove just as affective and can often you help you compete faster. The right partner still puts you in control and makes you feel like you are working with your own in-house data science team. And remember, if you’re still unsure what the right approach is, most partners will offer some sort of trial period so test them out before you decide to take the in-housing journey. Test & learn is the answer you should be looking for.


The Filter’s Personalisation Products

The Filter offers personalisation of key customer touchpoints based on proven data science. We put the right content in front of a customer at the right time.

The Filter uses data science methodologies to combine content and user data to predict what content a viewer might want to watch at any given moment. This drives up content engagement for each customer, where content engagement is a core KPI that correlates to churn, active users, loyalty and ultimately customer lifetime value.

We pride ourselves on working closely with clients both during seamless technical integration and through continual performance development, making sure that our client’s customers are always getting the optimal recommendations. User interfaces change, content and promotions change and business priorities change, and all of this means we need to constantly refine and develop personalisation through data science. Always A/B testing our solutions to show the value each development brings and continually improving our algorithms.


What you get with The Filter

  • Advanced, secure and reliable technology solutions with easy integration and continual support;
  • A suite of over 30 different algorithms that enhance your personalisation journey;
  • Interactive environments where clients can visualise the output for any customer in the context of their viewing history;
  • Editorial control through tools such as content boosting, allowing clients to increase or decrease the prominence of any given title as business needs dictate;
  • Comprehensive analytics which delivers key insights so that clients can understand the power of the personalisation and wider content and customer journey insights;
  • Dedicated account teams that work with you to deliver the very best solutions offering support, guidance and thought leadership.