With Adgorithms, AI Meets Real Location Data
Who is the smartest campaign optimizer of them all? Online marketing company Adgorithms would say it’s "Albert."
Using the powers of artificial intelligence – in particular, its platform, Albert – Adgorithms provides a programmatic media-buying platform that leverages real-time bidding for display, mobile, and video advertising campaigns. CEO and founder of the Tel Aviv-based company, Or Shani, talks with GeoMarketing about how advertisers looking to effectively geo-target can prosper from machine learning technology.
GeoMarketing: What is Adgorithms? When was it founded and how has it evolved?
Or Shani: We founded the company in early 2010. What we have done so far is to take the programmatic phase to the next level, where it is really autonomous. [Our technology] is a little bit similar to what Google is trying to do with the autonomous car, where you really want to say, “I want to drive to Utah” or wherever, and the car will just take you there.
To run a campaign with our technology, you just need to say, “Okay, this is my company, this is my goal, these are my KPIs,” and our system, which we call Albert, will take you there. In traditional programmatic – if you can call anything about programmatic traditional– most companies will give you the best tools possible to do everything you can, but it really depends who is in the driver’s seat. I’ve found, that in most cases, people will have a really hard time using those systems themselves, especially if they have limited budgets.
When you start a campaign with us, it is completely on autopilot.
Who is your primary clientele and where do they tend to be based?
Up until now, we have been working with direct advertisers and some agencies. We’re mainly focused in Europe right now, but are opening a New York office in the next six months. We are going to bring in more advertisers and agencies from the USA. We have clients in the US, but we feel that it is not our biggest territory now. This is why we are opening the New York office.
Can you walk us through how an advertiser or agency sets up a campaign with Adgorithms using the AI platform, Albert?
Sure. There are two ways of working with us: One would be the traditional managed service, where you will be in contact with one of our account managers; and the other option is to use our self-service platform, which should be out in the next two to three months. But the process is actually the same, and in both cases.
What you would need to do for the campaign you’re running – let’s say, it’s an app download campaign for mobile – [first] you define a budget, which can be anything between $5K and $100K [on managed service]; on the self-service, your budget can be between $1K and however much you want. You [establish] your daily budget, your goal, and whatever else is needed to run the campaign.
How do geo-targeting and other geo-components play a part in all this?
First of all geo-targeting is super-important. It doesn’t matter if it is done by state, city, DMA, and so on. You will always see the differences between [various locations], and you cannot say, “Okay, New York is always best at this and that, and L.A. is better at this and that.” It changes.
I think that geo-targeting is a perfect example of why artificial intelligence or machine learning can really leverage a campaign’s performance. In a nutshell, what do we need programmatic for? What do we need machine learning for? What do we need artificial intelligence for? The way I see it is: we need it to take all heavy lifting of the optimizing of the campaign, and do it in the background by machine.
As technology has become more precise [such that] we can now target people not only on the country level, but also on the state level, the city level, by the zip code, by the block, it becomes an almost impossible mission for a human being to optimize. Geo-targeting and machine learning, or artificial intelligence, is one of the best matches out there for this. It’s just too much work for a human being, the way I see it.
Does Albert simply blast out a series of geo-targeted ads and then adjust in real-time based on the immediate results?
No. Albert is not starting the campaign as in, “Let’s try everything and just eliminate what is not working,” it is actually much, much smarter than that. He takes the campaign information, and he tries to guess what he thinks is going to work the best for the advertiser. So, the first day, you may run an even smaller campaign than you expected. Once Albert understands that something is working, he will expand and evolve from that.
Let’s say that, for whatever reason, Albert thinks that a certain type of campaign will probably work better in smaller cities, or bigger metros. We would try those first, while maybe 10 percent of the campaigns would try other [areas]. Albert will then check if his guess or prediction was right. If it was right, he will continue to push that way. But if he sees that something doesn’t work as he thought it would, he will try other things.
Most companies in this industry, at least from my point of view, run campaigns that start very, very wide, and then eliminate [various profiles and touchpoints as they go on]. This causes a lot of waste in terms of advertising dollars. With Albert, from the get-go, you say, “Albert, just buy what you think might work for me, because you know that this is my game, and you can see that this is the type of offer, this is the type of campaign.” You can save a lot of money.