TechBytes with Enriko Aryanto, Co-Founder, QuanticMind

Enriko Aryanto
Co-Founder,  QuanticMind

Last month, QuanticMind secured $20 million in Series B that was followed by the data platform for intelligent marketing announcing an AI-powered solution for Google shopping. To understand how the new machine learning-based solution integrates with the existing predictive advertising platform, we spoke to QuanticMind Chief Technology Officer, Enriko Aryanto.

MTS: What are the key components of QuanticMind’s predictive advertising platform?
Enriko Aryanto: QuanticMind’s key components include: data science-based bidding algorithms, machine learning-powered automation that uses natural language processing (NLP) and semantic distance modeling and unlimited scale for both processing and storage.

My co-founders and I built QuanticMind because we saw major gaps in the world of digital marketing. Digital marketing is complicated, and it’s common for even the most talented and experienced marketers to overspend or underspend on their budgets.

Here’s what I mean: During our experience working at NexTag, we found there are ideal bid levels for every individual ad listing. Take paid search, for example. Enterprise-scale paid search managers have portfolios of millions of keywords they use in their search ads. And every keyword has an ideal bid level that will deliver the maximum number of conversions or maximum overall revenue.

Example: An apparel retailer runs search ads with the keyword “winter coats,” which has, let’s say, an ideal bid level of $3.50. This means, at a bid of exactly $3.50, the retailer will get the maximum clicks, conversions, and revenue. Spending more than the ideal bid, wastes money – it won’t get any additional conversions or revenue beyond that maximum level. Spending less than the ideal level potentially gets a lower ad position, less traffic, and ultimately fewer clicks, conversions, and revenue. This means lost opportunities to sell more winter coats.

So, as a digital marketer, you definitely want to find and bid at that ideal bid level. But my team found that hitting ideal bids for a keyword is difficult (or impossible) through manual management. Is this a deficiency on the part of marketers? No…marketers are smarter and more talented than ever. What’s the real reason finding a keyword’s ideal bid level is so hard? Data.

Data is the biggest challenge, and the biggest opportunity, in digital marketing today. Every individual advertising click contains thousands of important data points from key modifiers like location, device and time of day. Once you have a clear picture built on historical data – which customers click on your ads, where, at what time, using what device – you can locate and consistently bid at ideal levels, which leads to much stronger performance.

Here’s the problem: Getting those insights from your marketing data isn’t easy. Even a single click contains a huge amount of data. Quick example: There are 210 designated market areas (DMAs) in the US, 7 days in a week, 24 hours in a day, and 3 major devices for digital (desktop computer, mobile phone, tablet). Multiplying just these numbers together gets you 105,840 data points. Now, multiply this by every single keyword in your portfolio (again, potentially millions). You can see how this just becomes too much data for anyone to handle.

This is why we built QuanticMind on a core of data science to extrapolate the insights from all this marketing data. Our data science-based algorithms extrapolate the ideal bid levels for every individual keyword, and our machine learning-powered automation manages and adjusts bid levels over time while also mitigating data scarcity. As we built our technology to have no scale limits, our customers can import years of previous data and pull up crucial reporting in seconds.

MTS: How does QuanticMind help optimize long-tail bidding?
Enriko: We optimize bidding for the long-tail with machine learning-powered technology that automatically assigns bid-optimized ad groups using NLP and semantic distance modeling. By creating ideal bid models based on semantically similar keyword groups, we help our customers quickly ramp profitable long-tail bidding without having to rely on expensive trial-and-error.

Before we founded QuanticMind, my colleagues and I observed that bidding for long-tail keywords was an expensive process that involved “buying the data,” as paid search marketers call it. Long-tail keywords, for those who are not familiar, are extremely long keywords that are relevant to a specific use case, and often signal strong buying intent. However, most search queries tend to be much shorter and more generic, long-tail keywords usually have relatively little historical click data.

What does that mean for digital marketing? For “head terms” that have large amounts of historical data, bidding successfully is arguably a little easier. If you have large amounts of data on a keyword, and some way to extract the important insights from that data to inform your bidding strategy, you’re in a better position to find the ideal bid level for that term.

However, long-tail keywords, because of their specificity, just don’t have that kind of historical data. A long-tail query like “Nike shoes air max 2017 size 12 mens running on sale 94065” is something that relatively few people have ever searched for, which means it doesn’t have much historical data. So as an advertiser, without data, you’re left with guesswork to determine your bids. Then again, if a shopper does search for such a specific query, it’s likely they’re ready to buy. This is why getting long-tail to work is so important. It helps you capture those high-intent, high-converting clicks.

Unfortunately, because long-tail keywords are so data-scarce, many advertisers underutilize them or don’t use them at all. Because our technology’s NLP core is able to quickly determine ideal bids even for data-scarce long-tail keywords, we help our customers skip the expensive guesswork of “buying the data” and go straight to running profitable campaigns using long-tail keywords.

MTS: How do you integrate machine-level intelligence with analytics and campaign automation?
Enriko:  As mentioned, our technology uses machine learning with NLP and semantic distance modeling to accurately forecast ideal bids for data-scarce keywords, such as long-tail.

It also uses machine learning to automate bid management for our customers’ campaigns over time. While I’ve talked a great deal about the ideal bid level for individual keywords, there’s one other piece to that: Ideal bids don’t stay static. Previously, we talked about a hypothetical retailer selling winter coats. Obviously, when the weather gets colder, bidding on keywords for winter apparel gets more competitive. And when the clock ticks closer to noon each day, bidding on keywords for lunch restaurants gets more competitive.

Our machine learning technology is able to dynamically use the insights from all the relevant data we collect and adjust bids over time to their ideal levels. This is how we drive consistently stronger performance every time.

Also Read: Interview with Chaitanya Chandrasekar, Co-Founder and CEO at QuanticMind

MTS: How do social insights drive contemporary marketing performance metrics? What are the granular factors that impact the accuracy of such insights?
Enriko: Social is an increasingly important factor in digital marketing in general and performance marketing in specific. At a glance, social is a different world where engagement metrics are more-highly prioritized.

However, for performance marketers, it’s an important channel that generates its own set of data points, and as part of the changing customer buying journey, social is playing an increasingly large role. Shoppers don’t simply walk into a store and buy things anymore. They research them extensively and tune out any annoying advertising messaging that isn’t relevant to them in that exact moment – this is why global ad blocker use is on the rise, up 30% in the past year.

More and more shoppers are taking control of their own buying journey, leveraging channels like social to get word-of-mouth recommendations and browse reviews. This is why we consider it to be such an important channel.

However, the biggest factor that affects the accuracy of social insights is going to be the integrity of your data. As mentioned, digital marketers are drowning in data, but none of it is connected. To fully leverage data – including social data – to its full potential, you need to capture all relevant data from all relevant channels and connect it all together. This is how you can pull the insights you need.

MTS: What’s the next frontier for QuanticMind in bidding management and campaign management?
Enriko: Our team envisions a world of fully connected data. Digital marketers, as we mentioned, are already drowning in data, but they’re also struggling with another challenge – their data isn’t connected. They’re getting first-party data from search publishers (Google, Bing, Yahoo), but also session analytics, call analytics, inventory analytics, and third-party data. And none of these sources talks to any of the others. We already integrate all data from all relevant sources for search marketers. The next great frontier seems it’s not just about like integrating data specific to search, but also integrating data from related channels such as social, e-commerce, offline and others.

MTS: Would you tell us about your product roadmap for 2022? How would performance marketers use QuanticMind’s platform to leverage search engine marketing to drive ROI?
Enriko: We’re all extremely excited about the new products we have in development and ready for launch. You may have seen that we recently closed a $20M Series B round of funding, which we’re investing to enhance our flagship search product and launch new products. We have already been updating our social features and will be launching products for other important channels shortly.

Bigger picture, as mentioned, we envision a world of interconnected data where marketers never again miss out on opportunities or waste their budgets because their strategies are based on guesswork rather than the actual data-driven insights that help them bid at ideal levels and drive stronger performance.

MTS: Thanks for chatting with us, Enriko.
Stay tuned for more insights on marketing technologies. To participate in our Tech Bytes program, email us at news@martechseries-67ee47.ingress-bonde.easywp.com