Notice: Undefined index: order_next_posts in /nas/content/live/gadgetmag/wp-content/plugins/smart-scroll-posts/smart-scroll-posts.php on line 194

Notice: Undefined index: post_link_target in /nas/content/live/gadgetmag/wp-content/plugins/smart-scroll-posts/smart-scroll-posts.php on line 195

Notice: Undefined index: posts_featured_size in /nas/content/live/gadgetmag/wp-content/plugins/smart-scroll-posts/smart-scroll-posts.php on line 196

The art of visual searching

Web Designer talks to the brains behind the specialist search engine Niice, built to inspire and engage


Name: Chris Armstrong | Title: Designer | URL:

For those people who have never set eyes on Niice, can you define exactly what it is?
It’s our job as designers to come up with ideas on demand, and that can be really difficult. Niice is a tool to help you find that spark. It allows you to search across millions of images from the web’s top design sources to give you a face full of inspiration, presented in a way that encourages serendipity and pattern-matching.

What was the original thinking behind Niice. What were you hoping to achieve when you came up with the concept?
Originally, Niice was created to scratch my own itch. I find moodboards are a really useful way to ‘sketch out’ an aesthetic before I start designing, and I wanted to be able to easily create a digital moodboard that I could reference on my iPad while working.

So the moodboard-style presentation and responsive layout were there right from the very beginning, but it was only later that I realised plugging into multiple sources and searching across them would allow me to create these ‘instant moodboards’.

Once the concept was set in your mind how did the planning evolve?
It all happened quite sporadically. I built the initial prototype in a weekend, using the recently-deceased Zootool as a way to store images. I registered the domain back in April 2011, (inspired by my colleague David Hughes Belfast accent), and then the idea kind of sat dormant for a few years.
In February 2013 I was sitting in Toulouse Airport, waiting for a flight to London, when I came across a website that had a simple search bar across the top. It gave me an idea for how I could provide search functionality on Niice but keep the focus on the images. I designed a mockup on the flight, sent it to a few friends for feedback that evening, and started work on a prototype. Pete Hawkins joined me that week to help plug the site into different sources, and we launched the following Monday!

What did the build process involve? Which technologies did you decide to use and how long did the build take?
The first prototype took a weekend, the second took a week and we’ve been iterating on it ever since. Initially, all our ‘search’ did was go to a source site’s search page, run the user’s query, and pull the results back into Niice. It was kind of funny seeing some debates on Reddit about what kind of algorithm we were using, when the site was actually very dumb to begin with.
We’ve been gradually making it smarter over the past year, and have been building up our own index of design sources. The site’s tech stack is pretty standard: built on Ruby, hosted on Heroku and we use Elasticsearch to run the queries. A possibly interesting fact for you here: when we switched to Heroku from another hosting provider, our average visit duration jumped from  3 minutes to 15 minutes. It’s currently 20 minutes. Performance matters.
Because it was just Pete and I working on it in our spare time, we soon found that the only way to keep momentum up was to go for the quick wins, tasks that could be completed in an evening. Any time we tried to add a larger feature, we just slowed to a standstill (and the project nearly died a few times because of it). Since January, we’ve been dedicating more serious time and energy to the site, which has allowed us to take on some bigger tasks. However, it’s still surprising the impact that ‘quick wins’ can have: the ‘Surprise me’ button took me half an hour to build one evening, but is one of the most-loved features.

Does content have to  adhere to specific criteria in order to appear in the search engine? Is there a way that designers can submit their designs?
The key criteria is that it has to be content that has the potential to spark an idea. A Sixties ad man called James Webb Young wrote a great book called A Technique For Producing Ideas; in it he states that an idea is created by combining two existing – but previously unconnected – ideas. The element of surprise, of coming across something unexpected, is key. There are lots of fantastic design galleries and communities out there, and by their nature they all tend to have a particular aesthetic… but if we’re always exposed to the same thing over and over again, our ideas won’t be as good. I’d like Niice to be a site that exposes designers to ideas, aesthetics and communities that they wouldn’t normally come across, so I’m keen to find more sources that have content that is well designed, but a little off the beaten track.
I want to keep Niice a ‘curator of curators’, keeping search results to a hand-picked set of sources, so the best way for someone to get their work on Niice would be to add it to one of our sources, like Behance or Designspiration, and tag it well.

What future developments can we expect?
The next big things on the roadmap are suggested searches (so if you search for ‘squirrel’ we may suggest you search for ‘acorn’) and a ‘More like this’ button on images (also known as The Rabbithole Feature). We also have a big new feature coming early next month, but we’re not quite ready to talk about it yet.