- About Us
- Data Services
- Consulting Services
- Customer Cases
- Blogs & news
BNA is a Dutch Association of Architects that unites almost 1100 architectural firms. Driven by the power of architecture for the living environment, BNA promotes modern and creative entrepreneurship.
The association contributes to strengthening the positions of architects and the architectural sector through advocacy, exchanging knowledge, creating good conditions for entrepreneurship and facilitating networks. BNA is looking for new market opportunities for architectural firms and thrives to actively contribute to a healthy business climate.
BNA's mission is to strengthen architectural firms that offer added value to clients and society in promising markets. They do this by connecting architects all over the country, sharing information about the industry, the latest trends, legal advise and research on their website. Their tool, "find an architect", helps to connect visitors with professionals within their region, in fact, it is the most viewed page on BNA's website.
However, having all these consumers coming to "find an architect" page didn't quite result in a success as analytics showed the exit rates significantly higher than those from other pages. The intent to find an architect was there, but the existing solution didn't serve the purpose. What's more, it took a lot of time to keep it up-to-date, and agencies themselves had to reply to queries submitted within the portal. It was clear that a new solution was needed.
At Crystalloids we started developing Archy with a team of data scientists using Google Cloud technology. The first step was to identify those pages that were showing projects only. That means if someone searched for "school", the engine returned only pages that contained school projects. With the help of machine learning, we trained a model using a medium sized training dataset (+/- 2000 train examples) and after three months retrained it again to allow further improvement.
Project-filtered search behaviour of users told us better what are project pages and what not. Using this behaviour feedback resulted in a bigger training dataset, which contributed to advancing the model and thus more accurate project page filter. We then applied this model to all associated architect URLs and finally delivered a model API.
Archy is not a traditional platform where companies have to create a profile, upload and maintain the content. Instead, the engine pulls all the data from the web real-time and returns the best match for the person seeking an architect.
There is a great potential in the further development of Archy as a search engine into a platform on which BNA can offer relevant content and services for the target groups. Moreover, it can be embedded in other places where potential clients come. Thanks to machine learning, the search engine will also become increasingly smarter based on usage.
Archy works as a unique search platform for BNA. Both, professionals and individuals can benefit from this solution. Individuals can easily find an architectural project based on the entered criteria and search terms. Architect agencies do not have to access another portal to keep their content up-to-date as all the information about their projects is automatically being pulled from their website. What's more, the more Archy is being used, the smarter it gets.
Share this story