Words Drew Hyde
Collage Hannah Downton

Search is important. It's how we explore the internet, do our food shop and find our long lost relatives. Sometimes we even use it to find our next homes.

The funny thing about a great search experience is that you shouldn't notice it. Great search just works: it shows you exactly what you're looking for and does it instantly. It might even show you something useful that you weren't expecting.

Sadly, our previous search experience on The Modern House & Inigo fell short of this. In February this year, we released a totally rebuilt version of the TMH website, and Inigo followed shortly after. You might not have noticed this though as it looks almost identical to the previous one, just with shiny new tech behind the scenes that allows us to do new and exciting things.

Part of this re-platforming included a new Content Management System, a handy piece of software that helps us publish new listings and editorial features to the website. We chose Sanity for ours. Sanity has turned out to be a great choice for us. It's allowed us to build a system that matches how our team works and has simplified our workflows. The only downside though is that it isn't very good at search.

At launch, we implemented basic keyword search, which was the recommended approach at the time. For the most part it worked. If you searched "Stoke Newington" you'd see some matching homes, but you wouldn't see them if we hadn't mentioned the search term directly, or if a home was in a neighbouring area, if you'd made a typo, or if you didn't match the punctuation exactly. It was clear that there was room for improvement, so we set about designing what the perfect search experience would look like for us.

We knew our ideal search needed to be instantaneous, or as close as possible. It had to be flexible: handling typos, punctuation, and even understanding when someone searched for a neighbouring area. Most importantly, it needed to understand context: homes currently for sale should be prioritised before those under offer or sold, and homes for sale should take priority over our editorial content. We also needed to be able to easily tweak all of this ordering as our needs evolved.

We explored three main options. Sanity's own Embeddings Index was the obvious starting point as we were already using their tools. Though, while it was in beta, it lacked the features we needed for custom ranking and sorting. Google's Vertex AI suite had everything we could ask for, but it felt like overkill with its enterprise‑focused features such as managed vector search at large scale, granular IAM and audit, model training and versioning, and tight BigQuery integration.

Then we found Algolia, which hit the sweet spot: purpose-built for site search, recommended by Sanity directly, and equipped with the features we needed. One of Algolia's key strengths is “fuzzy matching” which essentially means it understands that "Stok Newingtn" is probably "Stoke Newington". Rather than requiring exact matches, it calculates how similar search terms are to our content, meaning typos and variations don't break the experience. This feature alone massively improved search results.

Our implementation is really rather simple. When a listing or editorial article is published in Sanity, a webhook (which is a way for systems to communicate and be alerted of changes) sends the record to a Vercel server-less function (a short‑lived API endpoint that runs on demand without managing servers). This function acts as a middleman that applies rankings based on the property's status (on the market, under offer, sold) and its type (listing, journal, almanac). These enriched records then get sent over to Algolia's index, where additional layers of intelligence take over, prioritising recently published content, applying our custom ranking rules, and making everything instantly searchable.

The benefit of this approach is that it happens instantly: when our team publishes a new property, it’s published and searchable seconds later. Most recently, we've extended the search to include Inigo properties in TMH search results, and vice versa. This cross-pollination means someone searching The Modern House might discover an Inigo home they wouldn’t have considered, and Inigo visitors can explore TMH’s more modern listings.

The most exciting thing about this approach though, is that it sets us up with a solid foundation for building new discovery experiences. Stay tuned.

This series will be accompanied by six original artworks created by Hannah Downton. Hannah is a Prime Buyer Specialist and collage and graphic artist; @hannah_downton_design