John Nardone

The media buying fee for the placement goes from whoever the agency or the DSP is conducting the media buy to the publisher.

So it's a pure tech fee.

We are not really using AI today.

So we're going to have our first generation of machine learning coming out at the end of this year.

But to this point, Flash Talking has been very focused on heuristic-based targeting and allowing customers to have a lot of control and to execute their target

targeting strategies based on their heuristics.

So I know that if it's raining out on the gap and I know it's raining out where the customer is, I know I want to show products from the outdoor catalog and I want to show sunglasses when it's raining out.

I don't need machine learning to tell me that that's the right way to think about my targeting.

There are a lot of circumstances that where machine learning is very, very helpful, particularly when you're doing pure performance based marketers.

not as interested about branding or long-term customer value development.

So there's a place for both the heuristics-based targeting as well as machine learning targeting.

There's a long customer list.

I would say that there's about 40 clients that account for 80% of the revenue.

And those are

on the big global advertisers, both in the US as well as in Europe.

We're in Europe where it's British Telecom and British Airways and O2, which is, and Vodafone, which are the two largest telcos in the UK.

So those are the kinds of clients that we focus on, not just for their size, but for their sophistication.

FlashTalk is a platform that's really built to meet the needs of the most sophisticated marketers.

to create competitive advantage by deploying their data in a very, very smart way.

Smaller clients don't either have the data assets or the sophistication generally to take advantage of what we do.