Through the help of a model called a contextual multi-armed bandit (CMAB), an algorithm from the field of reinforcement learning, Amplify Analytix are bringing their clients higher return on investment and are saving resources on paid search using machine learning.
How was the idea behind your startup born? Present your team and let our readers know how you all came together.
Established in December 2017 by Laura Murphy, Anna Thomas, and Daniel Strahinov, Amplify Analytix started with offices in The Netherlands and Bulgaria. We started as a small team of 7, serving two clients. By the end of 2018, we became a team of 12 professionals serving 6 long-term clients in both B2B and B2C segments. We covered a variety of industries from Electronics and Retail to Manufacturing and Utilities, with projects in marketing optimization, churn reduction, pricing, and others.
In 2019, we opened the 3rd office in India. We rapidly started to appear on the competitor lists of various analytics companies. During the same year, we signed our first PSP (Preferred Service Provider) contract with a multinational conglomerate corporation. We became ISO 27001:2013 certified and a member of the CISQ Federation. The certification allowed us to onboard a Premium Retail client with revenue over EUR 2 billion.
We attribute our success to talented and results-driven professionals in our team. It is always the people behind the numbers that make great things happen.
Can you please present your solution? What makes your solution stand out from the rest?
Our product helps paid search marketing teams be more efficient and bring greater results. We saw our clients waste budgets on paid search, so we built our product to help them increase return and save resources on paid search using machine learning. We developed a product that optimizes advertising spend across campaigns/keywords and their bids to get the best results based on clicks, CTR, or Revenue.
The type of model we are using is called a contextual multi-armed bandit (CMAB), an algorithm from the field of reinforcement learning. What we came up with is a non-traditional way of using the CMAB algorithm to test how your historical bidding strategies work, how much you could have earned and how to tweak it for higher future returns based on your specific goals and target customers.
As the model learns from past behavior and by exploring new campaign strategies, it learns which actions lead to the best increase in KPIs defined. Over time, the model explores different strategies to get the best return via defined KPIs and then exploits what it’s learned to maintain the level of these KPIs at their new height.
Which Data Provider do you collaborate with and how has your experience been? What made you choose them and their challenge?
We solved the challenge for JOT Internet Media. Their challenge of optimizing marketing ROI fits well in our expertise with current clients – we usually work with marketing and sales teams.
We saw a gap in many of our clients’ strategies where automation would be the most efficient option. This is how we created our first product, which was the result of a perfect match between our skillsets and JOT’s proposal as a data provider.
You have come a long way since the start of the EDI incubation programme. Can you tell us more about the evolution/traction of your solution, company and team during the programme?
Before EDI Amplify Analytix was a purely service company, providing bespoke solutions to our clients. Developing the product made us take a completely new route that we have not tried before.
We had to define our product-specific target audience and marketing approach. We dedicated a team to product development and hired a product manager. Everyone in the team had a chance to share their thoughts about the product. We even created a game to come up with a product name!
Some of our current clients got very interested in the product, many were eager to try it out. It showed us that we are moving in the right direction and that we can do both – create an impactful product and provide a high-quality service.
What challenges did you face and what lessons did you learn?
Developing a product meant for working in a completely different framework than what we were used to from our custom project-based business. We had to come up with a scalable idea that would still solve the exact challenge of each customer from our target audience. This made us realize there are a couple of challenges ahead.
First, there is no way to fully the context and specific challenge together with the customer upfront and “once and for all”. There will be adjustments. Second, you need to balance very well between standardized and customized solutions. It’s a fine line in between.
Key takeaways? It is OK to not know all the details – start by making assumptions and hypotheses. Talk to your customers and ask for feedback often to validate those hypotheses. Break down all ideas into smaller parts and test them in an agile way.
What impact has your solution achieved/ you are planning for it to achieve?
Our estimation, based on the JOT’s historical data is that the product could result in 40% and upwards higher ROI on average based on CTR. Our goal was to increase the ROI by 10-15%, and as we are testing the product in the real-world environment now, we are confident we will see this continued trend in live deployment and A/B testing experiments.
The current results we see are impressive for both us and our Data Provider, JOT Internet Media. Our collaboration has gone more than well, and we agreed to be developing the product together further down the line together. We defined our future goal to be optimizing for revenue returns rather than clicks because, after all, ROI is what matters for our clients, and therefore to us, the most.
According to your experience, what is the secret behind a successful data driven startup?
Trying things out. And not being afraid to acknowledge when something is not working. Sometimes, new insights come in during the data analysis stage that completely change the roadmap of a project. It is important to understand that things won’t always work out from the first try. And it is even more important to not give up.
Another tip is to listen, to your customers, to your mentors, to your team members. They are the people who will succeed if you succeed as an organization, so their feedback is highly valuable.
And of course, it is all about the team. We always pride ourselves on the wonderful, talented team that we built.
What’s next for you? Are you looking for partnerships, a new round of investment, new piloting or something else?
We are now looking to scale the product and start implementing it in our clients’ systems. We want to make product development a solid part of our business, now that we have gathered so much experience and expertise from our EDI mentors.
We are eager to test the market for our next product. Keep an eye on us to find out what it will be!
How would you describe your experience with EDI?
EDI made our dream come true. We are proud and happy to have gone so far in this challenge, which not only gave us practical experience and advice for product development but also introduced us to a great European startup ecosystem.
We have gone a long way from an idea to a functional product. The webinars and mentoring sessions have been an invaluable experience for us on this journey.