AI Can Do Heavy Lifting To Clean Up Ad Supply Chain: MediaMath's Archibald

AI Can Do Heavy Lifting To Clean Up Ad Supply Chain: MediaMath’s Archibald

It was technology that caused the digital ad supply chain to become cluttered and ineffective – so can technology put it right again?

In the battle to straighten-out the digital ad ecosystem, artificial intelligence is being leveraged to do the volume of work that humans cannot.

In this video interview with Beet.TV, Charlie Archibald, VP, Data Science, MediaMath, explains why AI is essential.

Scale beyond human

Some studies have shown, that for every dollar an advertiser spends, approximately 40 cents of that dollar, or 40% of it, is making its way to the publishers, because you have so many middlemen in there, taking their cut,” Archibald says.

“The challenge here is that, given the complexity of this ecosystem, the number of publishers that you have, the number of supply paths that you have in play, it just becomes a really tall order to ask the buyers and the traders to kind of manually day in, day out, kind of manage that.

“So it really requires an AI or machine learning-based approach to address that properly at scale, in an automated fashion.

“You’re just no longer at a scale where that sort of thing can be done manually solely at the hands of traders or buyers. To do that effectively, you really need to leverage, AI or machine learning based solutions.”

Supply path efficiency

Archibald says AI can be used across a wide range of issues, including supply path optimization, the practice of reducing the number of ad-tech partners in order to gain a shorter connection between buyer and seller.

“You’ll see efficiency gains for the DSP by reducing the number SSPs that they are working with for any given a set of inventory and reducing the load on the bidders,” he says.

“And you’ll see efficiencies for the advertisers too, so that they can figure out which supply path makes the most sense to drive the ROI for their business.”

Archibald says using AI can help alleviate strain on header bidding. That practice has been used to entertain bids from multiple bid sources simultaneously, rather than in sequential fashion, so as to achieve higher yield.

But Archibald says that has led to inventory being re-sold through a multitude of SSps, leading to a big spike in the volume of bid requests.

AI in the Brain

Archibald says MediaMath’s platform is using AI everywhere it can, but chiefly in three ways:

Automated performance: Automatically optimizing campaigns toward true business outcomes with little human intervention.
Transparent insights: Making clear how its AI algorithms come to decisions.
Open AI platform: Optionality for customers to run MediaMath’s AI on their own systems.

Transparent insights go to the heart of MediaMath – or, rather, its Brain. That is the name for its algorithm which determines which inventory to bid on and at what price/

“AI and machine learning can often feel a little bit like a black-box,” Archibald adds.

“MediaMath has become very focused on delivering granular transparency into the decisions that our algorithms make. We offer reporting insights into the decision that our Brain makes.

“That reporting gives our end users insights into what factors are most influential in driving the decisions about how much to bid for a given bid opportunity.”

You are watching “Media In Transition: How AI is Powering Change,” a Beet.TV leadership video series presented by IBM Watson Advertising. For more videos, please visit this page.

Published at Mon, 14 Dec 2020 02:15:00 +0000

Australian security agencies to build top secret private cloud

Australia’s security agencies have banded together and are planning to stand up a private community cloud at the top secret security domain capable of serving around 10,000 users.

The Office of National Intelligence (ONI) went looking for one or more cloud providers on Friday to deploy a cloud platform on behalf of the National Intelligence Community (NIC).

It comes more than three years after a review of the country’s security agencies called for the “establishment of an intelligence community computing environment”.

Until now, the NIC has had access to public cloud services up to the protected security domain, as well as traditional IT environments at the secret and top secret classifications.

The NIC consists of 10 agencies, including the Home Affairs, Australian Security Intelligence Organisation, Australian Signals Directorate and Australian Federal Police.

The top secret cloud – which is expected to provide IaaS, PaaS and SaaS services – will allow the NIC to rapidly develop tools and systems, including for data analysis.

The tools, combined with artificial intelligence and machine learning, will serve as a “force multiplier”, allowing the NCI’s “limited cleared workforce to focus on mission activities”.

“Data is the foundation of the intelligence lifecycle. The volume of data the NIC processes continues to grow in both size and complexity,” the ONI said in an expression of interest.

“The NIC is seeking solutions that will assist in accelerating its ability to scale the processing and analysis of large datasets.”

ONI is also looking to use the cloud to stand up collborative environments for shared and joint mission workloads to improve interoperability.

The cloud is expected to be hosted in “Australian controlled spaces, be geographically dispersed, have strong fault tolerance and be accessible from within Australia and abroad”.

It will also need to be “disconnected from the internet”, use cross domain solutions to transfer data where appropriate, and allow third-parties to operate SaaS services.

“Consuming SaaS services is the highest priority across the NIC and is predicted to provide the most benefit and efficiencies of all the service types,” EOI documents state.

“SaaS services should not be limited to only cloud provider offerings but also include the ability for third-party vendors to operate their services.”

Around 10,000 users from across the NCI are expected to eventually use the cloud, which appears separate to a private cloud platform that ASIO went looking for last year.

The platform – which is expected to operate across the protected, secret and top security domains, and support greenfields systems – is part of a broader IT transformation.

Published at Mon, 14 Dec 2020 01:30:00 +0000