Data science for the public good

Can we optimise the site search capability of the ONS website?

The new ONS website was officially launched on 25 February 2016. Once the site was released to a wider audience, we identifed the site search feature urgently needed improvement. More than a million users visit the ONS website every month, with more than 60% traffic coming from Google. This implies people are bypassing our internal search engine by using Google; a behaviour that indicates users needs were not met by the internal one.

Our work

Off-the-shelf solutions in Search Engine Optimisation (SEO) are mainly designed for commercial websites with different goals and constraints than government websites. We explored alternative options with a strong emphasis on understanding our users’ search behaviours and intents to help ONS web developers retrieve and rank search results in a more meaningful way.

What’s the data science?

We analysed Google Analytics data of the ONS website, in particular the search terminology, using Natural Language Processing and unsupervised machine learning techniques (e.g. clustering, Latent Dirichlet Allocation, Word2Vec) to uncover insights on users’ search behaviours and intent.

What’s the impact?

This project will help to:

• ensure the user will find relevant information to their search regardless of the terminology used
• inform web developers on how to rank search results in more meaningful way

Our partners

This work is part of a wider research and development effort led by the ONS Digital Publishing team. Beyond this collaboration, we hope that our findings will benefit other government organisations that share the same needs to develop user-centric websites.

If you’re interested in learning more about this project or would like to get involved then you can get in touch via email or Twitter