Megan Gell
Aug 20, 2018

How it really works... AI-powered matchmaking

"With machine learning, the more you swipe, the more it learns—and the better the recommendations become."

How it really works... AI-powered matchmaking

Grip began in London just two years ago and now powers event matchmaking for the likes of UBM, Reed, Cannes Lions, and the Dublin Tech Summit. It has a presence in the UK, Europe and the US; and has been used in Hong Kong and Singapore.

Grip founder and CEO Tim Groot discusses how AI is taking the serendipity out of making the right connections, and by doing so, delivering better returns on time and investment for stakeholders.

How is AI-powered matchmaking different?

AI-powered matchmaking differs in that it uses machine learning to make better initial recommendations that get smarter over time. When left to chance, organisers see a 22% success rate with recommended matches – those being the “yes” swipes – but Grip’s solution delivers a 55%
success rate.

Other systems are based on static data and macro trends rather than individual priorities. The event may have a “rule” that entrepreneurs want to meet investors, for example, but perhaps the entrepreneur just secured a major investor and wants to meet with journalists to talk about it.

No matter how many investor recommendations they decline, it will keep suggesting them. With machine learning, the more you swipe, the more it learns – and the better the recommendations become. When you’re attending an event with 10,000+ people, that’s huge.

How does the system make recommendations?

For the initial recommendations, Grip uses registration data from the organiser such as product interests, subject interests or a problem you’re dealing with. We combine that with social data that participants may use for populating their profile, as well as keywords from a summary or headline.

By combining those two data sets we further understand more about the user and what their interests might be.

Then, though you may have not had any previous interaction with the system, we may have had other users with similar data points so we use a Netflix-style of recommendations. There is also a sophisticated algorithm that weights the various data points.

The system itself is looking at the most relevant data, which is not necessarily the first guess you would have, even as an industry specialist looking at the data.

Does it comply with privacy laws?

No personal information is ever received by Grip, and any data collected or generated belongs to each individual event.

It’s like it borrows your bike to learn how to ride. Of course it gives you your bike back, it just keeps what it has learnt.

Source:
CEI

Related Articles

Just Published

1 hour ago

Berjaya Sompo's heroes pivot to a pandemic mission

The insurer's fivesome of 'Silent guardians' take on new protective duties in a spot by FCB Malaysia.

1 hour ago

Why the buying exercise is flawed in a media pitch

Especially during the pandemic, trying to lock in a media-buying position in a state of so much uncertainty is bound to create distrust and disappointment.

5 hours ago

BLM: Brands don't need to be saviours, but can't be ...

Netflix and Yorkshire Tea earned praise for their reactions to the Black Lives Matter movement, but Starbucks came off as inauthentic, according to a report from TBWA's 65dB social-insight unit, which looked at how BLM brand activism landed around the world.

5 hours ago

It's do-or-die time for account management

With the discipline at a watershed moment, how will its future be defined? Wunderman Thompson's global chief client officer shares five ways to set up client leads as strategic growth partners of the future.