A couple of weeks ago we had a hackathon to kick off 2018.
When it came to picking a theme, machine learning struck us as a timely choice – reports suggest AI could be responsible for a 40% increase in labour productivity by 2035. Global revenues from AI for enterprise applications is projected to grow from $1.62B in 2018 to $31.2B in 2025. And Facebook developed that scary program which invented its own language in a few days, causing general terror to all who read about it.
The starting point was simple – how could we use AI to improve Reevoo for both clients and consumers?
On the Monday morning we split into three teams, each with the challenge of building something that incorporated machine learning into our current platform. It quickly became clear that we were all tackling the same problem: how to help consumers make sense of all the noise. Or how to enable brands to do just that.
Finding the signal in the noise – how big is the problem?
You may have noticed that there’s quite a lot of content these days. From the cave paintings at Lascaux to Johannes Gutenberg’s printing press, we’re always coming up with new ways to create content. The internet has taken that instinct and democratised it.
Even in the last few years (since about 2014), the number of people with access to the internet worldwide has grown from 2.4bn to 3.8bn, taking the number to around half the global population. This sort of rapid growth has a knock-on effect. Many of the most popular services online source all content from their users – the big social networks or Wikipedia, for example. By 2025, all of this is projected to cause a 10x increase in the amount of new data created per day, reaching a gargantuan 463bn GB every 24 hours.
Again, the fact that the internet is increasingly a visual medium – as the popularity of Instagram or Snapchat will attest – is a major factor in driving this growth in data usage. Finally, just to add to the chaos, around 90% of that data is unstructured, meaning it has little or no metadata that would allow it to be quickly or easily categorised and interpreted.
We’ve already taken some steps to help consumers sort through all the noise. We let users filter reviews to see content from people like them, and we group our content by product or by brand to give consumers more clarity on what they’re reading a review of; more on that in this short video from our customer success team. Unfortunately, this still leaves the picture content we collect largely out of the picture.
Reevoo Experiences is based on the power of rich storytelling to help brands share more of their customers’ stories. We think any UGC strategy that leaves this part out is incomplete. That being said, we don’t want to be contributing to the noise – consumers want content from people like them, but not at the expense of being able to understand and trust what’s in front of them.
What are we gonna do about it?
This thought process is what drove the winning team from our hackathon to victory. They decided to focus their energy on image recognition, using AI to determine what a photo collected through Experiences is actually about. When you’re collecting thousands of images a day, the need for some kind of sorting mechanism becomes obvious pretty fast. The thinking turned to delivering something tangible that would enable our clients to share all of their exciting rich content. With that, Content hub was born.
The beta version allows brands using our Experiences solution to sort and find photos collected by theme, as determined by Google’s Cloud Vision solution. From now on, looking for that killer content to promote products from across their portfolio will be much easier.
Looking to promote romantic destinations perfect for a marriage proposal? Check out the photos from previous customers tagged ‘engagement’ or ‘ring’.
Want a cute cat picture to tell a story about how pet insurance helps protect your furry friends? Look through the pictures from customers who bought pet insurance, and narrow it down to just ‘cats’ (or ‘dogs’, if you’re more of a dog person).
In the long run, it’s easy to see how we could harness tech like this to sort, filter and summarise the content we collect. In a world that never stops creating new stories, new photos and new reasons to doubt what’s being put in front of us, an injection of order and trustworthiness is well overdue.
Content hub puts us on the road to making that a reality across the whole Reevoo platform, ultimately benefiting both brand and consumer.