Site Search Made 5x Faster: An Image Search Time Trial
By Lihi Pinto Fryman • June 04, 2018
One thing that comes up again and again when talking about image technology is how much easier it makes the shopping experience for a consumer. It’s easy to say that something is beneficial, but the only way you can really know is to experience it for yourself.
We’re putting our time where our mouth is today and testing out exactly how much easier it is to find exactly what you’re looking for, with image search and without it. After all, as ecommerce consumers, each of us has faced a futile online search where we just can’t seem to find what we want, so we end up wasting time fiddling around on website after website.
So we’ll try two different searches — one for clothes and one for furniture — and we’ll time ourselves while we do it. Place your bets! Ours is on image search.
Scenario One: The Bridesmaid Debacle
There comes a time in every person’s life where they have to follow a dress code or match a certain look, and one of the most stressful of those occasions is a wedding. Those who are in a wedding ceremony often have to adhere to “guidelines” so that everyone matches, yet they’re left to find their own outfit.
So you’ve just received an email asking you to purchase a dress. You open the attachment, and . . .
What color blue is that? Is that above the knee or below the knee? How long is it going to take you to find a dress like that? We’re here to figure it out. Game on.
First, let’s search Macy’s — they have a large selection of dresses, and it’s likely that their inventory has something that fits this description.
We start by heading over to the “dresses” category:
But that is a big category. Should we search in the “bridal” section — will they have a bright blue dress? Should we choose “formal” dresses instead? Ok, let’s just start with color and length:
Wow, that’s a lot of results.
And the categories aren’t exactly foolproof. Blue is a popular color.
After limiting the search to formal dresses, we found that Macy’s has one dress that matches the features that we need:
The total time it took was:
Some critical thinking, filtering again and again, and we got one dress. That’s pretty par for the course, but imagine wanting more options. This is going to be a rinse-and-repeat process over several sites, each with their own layout and categories.
Let’s try this again, but with image search. We head over to Intu and hit the visual-search button:
Upload our picture:
Oh, thank goodness! There are two options right away, and a third just one row down:
And it took only — 27 seconds, including looking through every single option that the search pulled up and finding the photo on a cluttered desktop. Not bad.
Intu is obviously faster than Macy’s, but it’s also easier. And if someone has a particular piece of clothing in mind — whether for a wedding or a costume, or just because they really love the looks Rihanna’s wearing these days — they will search to get it. If they know you have a visual search, your site is the perfect place for people to find what they want, without the hassle.
Syte’s aim is to make shopping that easy. You see it, you search it, and you get it. No trying to figure out what counts as blue or whether you want the “cocktail” or “formal” dress categories.
Scenario Two: The Bench, the Dresser, and the Wardrobe
Redecorating can be quite the challenge. After you’ve collected 500 pictures of beautiful master bedrooms, it’s time to pull the trigger and actually buy some of the furniture you’ve been eyeing for weeks. The only problem is that you need something within your budget — and you wouldn’t mind just ordering everything from one place to make for an easier experience, either.
While we can’t promise design expertise, we can tell you how long it could take to find furniture if you’ve got inspiration you’re looking to imitate.
To start, let’s head over to Home Depot to find out if they have a wooden bench like this one:
For the bench, we wanted to use site navigation to get to the right department; for the dresser, we wanted to use search terms — one of each to see if there was any difference in time.
There was. We started with the bench, and the website was hard to navigate. Finding the right menu and terms was difficult. For example, “shop by room” at the top ended up being inspirational photos of bedrooms, not bedroom items. Finally, we get to this menu and figure that “decor and furniture → furniture → bedroom furniture” is a good place to start.
Then, we go in for the bench. On the sidebar, there is, mercifully, a section just for benches:
We narrow down our search by putting in material and price:
But unfortunately, not every return is actually a wood bench:
After scrolling through two pages, we come up with two candidates for our look-for-less bench:
Between the website navigation, narrowing down search, and clicking on options for benches to get them to look like our inspiration, we’ve been on the Home Depot website for 5 minutes.
Then, to find a dresser, we use the search bar at the top. It autofills, and we go straight to the dresser section:
This is where things get sticky again. We play around (unsuccessfully) with the preset options for narrowing search. Should we put solid wood? That seems to return things that look nothing like the coloring we want. Height returns a lot of options. Briefly, this candidate appears:
It has a pretty close color match. But in all of the filtering and unfiltering, we lost the price control, and $1,029 is over budget. Finally, we put in a search with our lower price and the number of drawers, which does the trick. This guy is pretty close:
By this time, we’ve spent quite a while here, and it felt like a struggle to figure out how to get the items we wanted. Dresses seemed like a cakewalk in comparison.
Time to test with visual search. We head over to Taskers and are greeted with an image-upload option — a welcome sight after the puzzle of Home Depot’s website:
Then, we upload our photo, choose our bench when Taskers recognizes multiple furniture items in the photo, and slide our price down on the top bar:
The results are not bad, and after scrolling through results, we like the left two options best. Including sneakily looking at benches over our budget, it’s been 1 minute and 24 seconds. Time for the dresser.
Again, we upload the photo and set our price slider:
As with Home Depot, there weren’t a ton of dresser matches in our price range, but the top-left option has similar coloring and grain and is way below budget — enough money leftover to swap out the handles and match the look. Or, if that wasn’t our cup of tea, at least we only spent a fraction of the time finding that option. In under two minutes, we knew whether Taskers had what we were looking for.
Taskers is easier because it does the search for you, provides you with options, and saves you the hassle of trying to describe furniture. As a layperson, trying to figure out the perfect woods, stains, and height combination to match our inspiration pictures was frustrating. And this isn’t a Home Depot-specific problem. The difficulties can be encountered on many furniture and decor websites.
The experience of shopping on Taskers was more like an in-store experience, where you could ask a sales clerk if they had anything like what you’re looking for. It felt like we were being helped on our buying experience, not playing guessing games — and it took less than a fourth of the time!
More Than Just Convenience
Image search creates a richer ecommerce experience for customers, who get a more personal touch. It’s an incredibly convenient feature that gets shoppers what they want, fast.
But image search isn’t the only way that image technology can help shoppers. There’s a lot that image recognition can do, beyond replacing manual search. For example, you can also show similar items for shoppers as they come across something they like:
For casual browsers, this can pull up a plethora of options that are eye-catching and spark interest:
You can also make photos shoppable — so when a customer finds inspiration in your lookbook or promotional materials, they are able to see those items pop up, just like how we selected the bench and got shopping options on Taskers.
And this is all continuing to develop. Image tech powers a range of tools that help ecommerce sites better mimic the way people want to shop and find things. It’s less frustrating, more intuitive, and more convenient. Try it for yourself. Time how long it takes you to find something on Home Depot and Taskers. Or try Nordstrom and Intu. And when you’re convinced that image search is better, contact Syte, and we’ll get it up and running for you in 24 hours.
Today, we’re happy to share that Syte raised $21.5 million in Series B Funding. Viola Ventures lead the investment, alongside high-profile investors Storm Ventures, Commerce Ventures, Axess Ventures, and Lyra Ventures. The total funding to date is $30 million.