After receiving $3.5 million in Series A funding, Spotad announced it was taking its artificial-intelligence-driven mobile advertising platform into China, and in the process touting itself as the first Western demand-side platform (DSP) to enter the market. eMarketer’s David Green spoke with Spotad co-founder Yoav Oz about setting up in China and how Spotad’s service works.
eMarketer: What is Spotad’s background and what prompted the decision to enter China?
Yoav Oz: Spotad started from client requests. When we started, we were a creative platform for rich media, and we built a platform that can give [clients] in one minute the full QA [quality assurance] for creative, for every device operating in a system, in any size.
When we’d done that, the clients asked us to buy media, so we built our own ad serving [platform] and connected to a lot of publishers, using phone calls and emails to close deals. But that’s not scalable. So we built our own lifetime value [LTV] real-time bidder, connected to exchanges, and we had the biggest client in the world and the biggest exchange: Uber and Google.
We [eventually] focused on big data and when we listed the first predictive algorithms with the data that we collected in the first week, we saw amazing results, so we shifted more to being a big data and AI [artificial intelligence] company.
eMarketer: Why enter China?
Oz: Listening to clients, we saw how the world is getting more and more flat. Companies don’t want to go only to one market. With Uber starting to approach APAC, and more clients approaching us to scale into China, we saw a big opportunity that no one had done in the West before.
eMarketer: That was a relatively recent decision. How did you go about setting up in China, and what stage is Spotad at now in terms of working out partnerships with the necessary ad exchanges?
Oz: We want to be the first in China to integrate to all the exchanges that no one has connected to before us. We have Baidu, Alibaba, Weibo, Sina. We have the contracts, the APIs in Chinese, and we translated them to English so our developers can do the integration, and today we are talking to Youku, iQiyi and to Tencent.
To do so, we had to open a Wholly Foreign-Owned Enterprise (WFOE)—a special agreement with the government that’s called an Internet Content Provider (ICP) license to advertise ads inside of China. And because we are a tech company working with bidders, we also needed to get approval for infrastructure and big data services inside China to work with the exchanges.
eMarketer: How long did that process take?
Oz: We did everything at the same time, and for all the approvals for the infrastructure and everything, it took us around nine months.
eMarketer: Did you have any concerns about taking the Spotad model into China, or concerns more generally about working in the country?
Oz: When I went to China, everyone scared me by saying that if you do something right and you’re making a lot of money, they will try to copy you and do it themselves. In my experience, it’s been really different. Every time I talk to one of the big companies, Baidu or Weibo, or with clients over there—Momo, Ofo, Didi, Jingdong—no one tried to copy us. Also, we have an under-the-radar technology that’s in Israel—it’s really hard to duplicate.
eMarketer: What’s the most difficult thing for Western companies?
Oz: The most difficult thing happening to Western companies when they’re getting into a contract in China is that they’re trying to negotiate and explain to the Chinese how to do business. In China, you start negotiating after you sign a contract. After you show performance, then you can negotiate. This stops a lot of companies.
eMarketer: Are you taking your US clients with you into China, or are you also helping China-based clients as well?
Oz: We’re inside most big clients in China that are looking to go to US and Europe with their business. We see a lot of US and European companies that are trying to grow in APAC.
eMarketer: How does your revenue model work?
Oz: We only work with clients that are open to sharing data. We go for a small budget and scale performance, the downloads we want to get, and then build our predictive algorithms for that brand from the data that it sends to us. It can be user ID, or if clients don’t want to share the IDs, it can just be the attribution of the data. Both are OK, but only if we can get predictive algorithms from the data, across real-time changes by hour and date, to make the best performance for clients, who measure us on LTV.
So for example, Uber is paying us full CPI [cost per impression], and every CPI for every location around the world is a different price, but they measure us on cost per first trip [CPFT]. So say they’re looking for users that pass $5 CPI, but only if this ride spent more than $15. I haven’t see any other companies that are paying on CPI but measure by LTV. We are doing that because we can connect into DMPs [data management platforms] that share the clients’ data.
eMarketer: So you’re facilitating a demand-side platform [DSP] model sharing arrangement?
Oz: We don’t see it as a DSP, especially in China. DSPs have a bad reputation and most of them are connecting to supply-side platforms and AdWords. We see ourselves as a marketing operating system.
We’re not only doing mobile. We give clients opportunity to go as desktop and video, rich media, native in the same platform, retargeting and even social, in the future, in one platform, and we’re not only focusing on one market. We’re connected to almost 15 exchanges globally.
eMarketer: What are the differences between China and the US in terms of getting your hands on the necessary data?
Oz: The data is very different in China, and the regulations require you do different things, but if you have smart technology, you can use other DMPs to get the missing information. In China, it’s more difficult to get geos—the right geos, and even gender, but you can sometimes get point of interest and even keywords.
In the US and Europe, you cannot get point of interest or keywords. No one will give you that, but you really have good data on the geos and gender. The method depends on the machine, the predictive.
eMarketer: So you’ve said you are working with most of the major exchanges, and working on Tencent. Is that enough, or are you also trying to negotiate premium inventory directly, as is often necessary in China?
Oz: A few years ago the fight was around supply. After that, it was around demand. Now everyone can get the supply and demand. The real fight is not about demand or supply. It’s about utilizing the right data because a lot of the time that one premium publisher will go on one exchange and after a month disappear, but pop up on a different exchange.
“The real fight is not about demand or supply. It’s about utilizing the right data.”
We are connected to most of the exchanges, and we see sometimes the same publisher sitting in all those exchanges. In our system, we don’t care where we buy the inventory. The algorithm gives all the supply or all the exchanges as flat. You just choose the best-performing one based on a very smart AI.
eMarketer: So the algorithm eliminates discrimination based on the publisher?
Oz: Exactly. We’re not focusing on running it like a network and putting our stick in every publisher. It’s more important to utilize the right data from DMPs and exchanges, and most importantly from the clients. If they’re giving you data, you don’t really need any SDK [software development kit] to go after premium supply.
eMarketer: Obviously, ad fraud is a major problem in China. How does your platform account for that?
Oz: As a company that everyone measures on LTV—it says everything. If you’re not bringing good LTV, you are not going to work with us.
Besides that, we’re working with the biggest fraud detection in the market, and we are fully transparent. We give clients the ID they want, and we have also a solution that will give us ID for identification between mobile web and in-app. Finally, if a client is trying to check fraud detection, we give them all the tools that there are outside, and the tools that we have, and transparent reporting for both.