Following user feedback the Financial Markets Authority has used Elastic to enhance the search experience for visitors to its website.
The improved experience means 80% of queries are now resolved on the website which has cut the volume of routine enquiries to FMA staff while delivering the foundation to support public LLM engagement and queries.
As the conduct regulator for financial markets, the FMA promotes fair, efficient and transparent markets to support confident participation by businesses, investors and consumers. Its website provides information to a large, regulated community, from banks and financial advisers to compliance managers and fintechs, as well as members of the public looking for plain-language direction.
The website holds thousands of pages and documents on guidance, licensing information, media releases, publications, podcasts, and PDFs, which are updated frequently and must be kept for long periods under retention requirements. Over time, the growing volume made accurate discovery difficult, even for experienced users.
Customer feedback was consistent: people could not easily find what they needed. The FMA’s previous keyword search often returned irrelevant results, forcing users to sift through long lists to locate the right page or PDF. FMA staff also spent time adding metadata and tags in the content to try to improve findability, while frontline teams fielded avoidable email queries for information that already existed online.
To address the issue, the FMA engaged Datacom early to look beyond a simple search engine swap and instead redesign the overall search experience – front end, back end and user journey – so search was “front and centre”.
Datacom then ran a structured evaluation across multiple platforms, weighing relevance, flexibility, governance and long-term cost.
The FMA selected Elastic for its ability to be tailored to the regulator’s needs and as a commercial model that suited a high-volume, content-rich site. The decision also marked a local milestone: the FMA became Elastic’s first Intelligent Search government customer in New Zealand.
The solution design integrated the FMA’s Silverstripe content management system (CMS) with a separate Elastic search platform that continuously indexes website pages and documents. On top of that, an AI summarisation layer powered by Azure OpenAI – generates plain-language responses that are grounded in the FMA’s own indexed content rather than searching the wider internet which helps to address governance concerns.
Implementation was delivered in phases. First came an Elastic integration to lift the quality of traditional keyword search. Next, the team introduced hybrid semantic search and an AI-generated answer panel that cites the most relevant pages and PDFs. Elastic provided a trial environment that let the FMA team test real search queries, identify patterns and edge cases, and iterate through multiple rounds of testing, pre-production and UAT, before going live.
“The trial was eye-opening – we could see the same questions coming up in lots of different ways, and where people were getting stuck,” says the FMA’s Strategic Product Lead Lindy Shuttleworth, who led the initiative. “Testing with real query data helped us spot the edge cases early and tune the experience and content, so customers didn’t have to learn our internal language to get the right answer.”
Since go-live, the FMA has been tracking how effectively people can self-serve online. It set a target of resolving more than 80% of searches on the website and is achieving that level – reducing the need for customers to email or call for routine clarification and freeing call center staff to focus on higher-value work.
A key requirement was that the AI search stayed contained to the FMA’s CMS and website content, helping address governance concerns and support internal approvals.
“For a regulator, trust matters – so we needed AI that stayed inside our governed content and could point people to our source material,” Shuttleworth says.
The project’s early results have been shared beyond the FMA with Shuttleworth presenting the use case to a government AI community of practice and digitalisation forum, showing other departments a practical pattern for improving public-facing information services with a controlled, content-grounded approach.
“AI is changing what ‘good search’ looks like,” says Datacom Head of Digital Elias Billeh. “People don’t want to trawl through pages and PDFs – they want a clear answer, fast. AI-generated search lets organisations meet that expectation by turning governed content into plain-language responses that cite the source, so users can trust what they’re seeing. The result is a lower-friction journey: fewer dead ends, less rework, and more people getting what they need on the first try. This is a first step in helping organisations prepare for an AI-enabled future across digital channels.”
Now, the FMA is exploring how the same foundations could support a more conversational, chatbot-style experience, while also planning a “My FMA” portal for authenticated users.
For Datacom, the engagement has become a repeatable blueprint – combining search optimisation, user-centred design and governed generative AI – to help other organisations make complex information easier to navigate.
“Datacom worked like an extension of our team” says Shuttleworth, “and that collaboration is a big part of why we could move from a frustrating search box to something that genuinely helps people find answers.”
We create technology experiences that optimise what you do today and prepare your business for tomorrow, so you can achieve the future you have in mind for your people, customers and organisation.
