In today’s metasearch complexity, one single property can have over 700,000 bids to cover all the different search scenarios. Machine Learning algorithms process large data sets to calculate the best bids faster than a human can by using millions of different data points.
With the roll-out of a brand-new bidding platform, Click, DerbySoft’s data science team launched upgraded versions of the machine learning algorithms in an ongoing effort to improve optimizations and results for clients.
DerbySoft found that the new algorithms could attribute to significant increases in ROAS (+63%) and incremental conversion rates while keeping CPCs low. In the end, DerbySoft was able to determine that this new approach was a success.
Why is Machine Learning critical for metasearch campaigns efficiency?
Every minute that your team spends on the manual data analysis of thousands of bytes and bid calculations equates to that much time they are not focusing on your marketing strategy. This is basically minutes of lost opportunities and revenue.
Since the beginning of the metasearch industry, bidding has been a guessing game for many advertisers – testing, succeeding and failing. This process requires more and more resources to perform data analysis for effective bidding. This data analysis has become more complex and nearly impossible for human resources to perform due to the constant evolution of the metasearch channels’ technology, their increased numbers of users as well as their necessity as a distribution channel for the hotels’ direct bookings.
The main obstacles for hoteliers could be:
- – Not enough data or technology to make the right decisions
- – Not enough human resources or time
So what is Machine Learning?
Machine Learning is a system designed to perform specific tasks without using explicit rules but relying on patterns, models, and predictions extracted from large amounts of data that we have access to. Human interactions are only needed to provide the end goal to this system.
DerbySoft is one of the first providers in the hospitality technology industry that has been using effective Machine Learning algorithms to solve complex metasearch problems exploiting the vast amount of data that channels make available to our clients.
Machine Learning brings dynamism and precision to the bidding process by employing real-time data analysis of numerous sources and data points. This process allows DerbySoft to decide the best bid for the desired result.
How does it work for metasearch?
The DerbySoft algorithms create clusters of large data sets and can predict cost and revenue alongside other variables. These predictions are adjusted over time since the algorithms learn from the experience and interactions with the channels. The data points used for these predictions can include (but are not limited to):
- – Demand levels, clicks and impressions for each property, country or device
- – Information on what each user has previously booked and what has not been booked
- – Your price parity score
The more information we have, the more precise the decisions become. DerbySoft tests multiple algorithms against each other simultaneously so the company can choose the most successful version.
Automation is Key to Driving Revenue
In the ever-evolving world of metasearch, you need strong technology to be able to manage this key driving revenue channel effectively. DerbySoft offers cross-channel optimization tools and results-oriented algorithms that do the most difficult parts of the job for you.
Our goal is to continuously deploy optimal marketing bidding algorithms that maximize customers’ targets and exploit the vast amount of data and expertise that DerbySoft has accumulated.
See the latest case study on how the brand-new algorithm from DerbySoft improved the ROAS and revenue for one of our clients.