Moloco’s DSP with Christie Zhang | Appvertiser Chat #1

Ok, Christie, let’s go! What makes Moloco’s demand-side platform (DSP) different from other platforms in the market?

Good one to start. Moloco differentiates by driving sustained and scaled performance for advertisers using operational machine learning (ML) models. We can look at this in three main buckets:

Automation – We maximize ROI for advertisers by letting ML do the heavy lifting, automating all aspects of the campaign, including targeting, bidding, and optimization — in real-time.

Scale – Moloco’s inventory is the open internet where users spend 66% of their time. We help advertisers gain incremental conversions outside the walled gardens by accessing 2.7 million apps in 190 countries reaching 6 billion devices.

Transparency and trust – For advertisers, we strive to maximize the value of their first-party data. Beyond giving data transparency down to the log level, we provide deep insights to inform business decisions and uphold the highest data and privacy standards, including anonymization, encryption, and dynamic anti-fraud protection.

We are very curious on how Moloco uses Machine Learning to optimize campaign success…

Sure. We use operational ML that leverages Deep Neural Networks (DNN) rather than linear models. This translates to the accuracy of predictions, virtually unlimited adaptability, and scale. Our models process huge amounts of data to make 9 million predictions per second for thousands of advertisers.

DNNs are intrinsically capable of supporting large models. The larger the model and the more data it can leverage, the more accurate the predictions to drive better performance. In a privacy-centric world where user data will remain limited, our DNN-based models can uncover and leverage a myriad of available signals to optimize prediction accuracy and update in real time for continual improvement.

Got it! Now, can you explain the process you use to select the most effective creative when there are numerous options?

Yes. Our creative optimization model automatically selects the best-performing creatives to use based on real-time results. For instance, our model would allocate more weight to higher performing creatives, while continuing to test all assets in the campaign.

Alright… and what creative strategy or approach has proven to be most effective for advertisers using Moloco?

Our approach is about broadening reach to as many high-quality users as possible. To ensure we have creatives to every inventory type, we recommend our advertisers use a broad range of creative assets including banner, video, native, and interactive end cards (IEC)/standalone playables. Furthermore, we automate the creative production process for our advertisers using tools like auto-generated images and dynamic product ads (DPA).

Crystal clear. Would you also share some insights into the scalability and flexibility, particularly when handling large-scale campaigns or serving ads to a global audience?

Speed is crucial in programmatic advertising because bid requests need to happen in real-time across billions of requests. Thus, DSPs need enough computing power to select a qualified bid request, predict user action, calculate the right price, and place a bid — instantly.

Moloco does all of this within 14 milliseconds for each bid request because of our scaled infrastructure that handles nearly 600 billion bid requests daily.

We’re exceptional at scaling performance with any budget, and offer features to do that for any advertiser goal. Our dynamic budgeting allocates spending to the best-performing days and times throughout the week, our multi-country campaign setting enables cost-effective geo expansion combined with real-time flexible budgeting, and our zero-cost training helps hit goals faster by training on offline data before campaigns even start.

As a result, we’re able to sustain performance levels on annual budgets as large as $100 million.

We’ve teamed up with knowledgable Christie Zhang, Product Marketing Manager to chat about Moloco‘s technology.

If you want more information on machine learning- powered performance advertising solutions don’t hesitate to visit Moloco’s website, where will find a wealth of information and resources related to this.