Applying Machine Learning to Business Outcomes at Travelopia

Applying Machine Learning to Business Outcomes at Travelopia

Travelopia shifted its focus from a technology-driven approach to business results, and adapted agile and lean to deliver machine learning solutions. This allowed them to deliver machine learning business models faster and better.

Sreekandh Balakrishnan, “Director – Innovation” for Travelopia, and Simon Case, Chief Data Officer for Equal Experts, spoke about machine learning at Travelopedia at Lean Agile Scotland 2022.

The first iteration of machine learning was very technology-driven, Balakrishnan said:

We took a technology stack approach and built a data lake before understanding business use cases. We took a big bang approach to delivery, a big team, with the promise of delivering multiple use cases once the data lake is in place. After 18 months we had 3 models in production but no fewer business units using it.

Suggested case to understand which business improvement you want first:

Start small, choose a value slice, and learn not only about the technology, but also how to define the problem, how to engage users, how to deliver results to users or downstream services, and what other organizational challenges (people , process first and technology last) are there.

Balakrishnan mentioned that they didn’t have the technical or business impact they wanted. They made a few changes to become leaner and focused on business results:

We had a second iteration using the new lean and agile approach and we were able to deliver 2 models in 3 months that are in use and creating business value. After this success, we adopted this as our methodology. Since then, we have grown to 10 models in production for 5 brands. All are used in the business, and some brands generate almost 21% more business revenue. In fact, I can now deliver a new model to the company in less than 10 weeks.

InfoQ interviewed Sreekandh Balakrishnan and Simon Case about machine learning at Travelopedia.

InfoQ: What did you learn from how you initially applied machine learning, and how has that impacted your approach?

Sreekandh Balakrishnan: We applied lean and agile principles – find out what’s valuable, deliver in small increments, and keep learning (and pivoting until you get it right).

We have shifted the focus from a technology to a business outcome. We took the time to understand what the company wanted and how they wanted it delivered. The team was too big, so we reduced it from 40 people to a team of 6. We found that a lean, cross-functional team was able to move faster and stay focused on what the business wanted.

We stopped building a data lake that would meet all ML needs and started focusing only on preparing the data we needed. It also had the side effect of reducing our cloud costs to 10% of what they were before.

We have also consciously moved away from GUI-driven tools. We had used them in the first iteration, but had trouble applying modern software development techniques (TDD, Pair Programming) when using them. Rather than speeding up our delivery, they were slowing it down.

Finally, I recognized that I needed buy-in from the company, which was more accustomed to big bang delivery. So I made sure I had an executive sponsor who understood and bought into this approach. It really helped our relationships with the company and made it easier to adopt ML models.

InfoQ: In your talk, you suggested not worrying about data platforms, but rather worrying about how your teams will self-organize. Can you explain why?

Case Simon: Machine learning is a team sport. You need the data scientist and data engineers to work together. They are different disciplines – data scientists are good at algorithms and math, but often lack the software skills needed to create reliable products. The temptation may be to separate them, but if you do that, they won’t learn the techniques needed to put their work into production.

Balakrishnan: Things started to change for us when we downsized the team and created one small cross-functional team. With data engineers, data scientists, and business analysts working as a team, they understood users better and could make quick decisions and trade-offs on the technology stack and ML model.

InfoQ: What is your advice to companies considering using machine learning?

Balakrishnan: Start small and don’t worry about buzzwords! Keep company stakeholders close and invite them to daily stand-up/planning meetings. Create the first win and go live. Remember, this is a learning curve for you, your team, and your business.

Case: Build your skills iteratively – start with one steel wire – the first vertical slice which provides useful commercial value. Use it to learn quickly, and when you’re happy with it, it acts as a model you can use for other ML models. If you’re lucky, it becomes a cobbled street – a working method that allows you to quickly introduce new models into the company.

#Applying #Machine #Learning #Business #Outcomes #Travelopia

Leave a Comment

Your email address will not be published. Required fields are marked *