Leadspace continues to innovate by applying Deep Learning to help Marketing and Sales teams create accurate target account lists for ABM.
The problem of “Big Data” is one most B2B marketers will be familiar with by now (we’ve also addressed it extensively on this blog). But there’s another, lesser known phenomenon that’s just as much of a challenge: Small Data.
The challenge of Big Data is very simple: there’s simply too much out there to humanly process. Artificial Intelligence (AI) in the form of machine learning offers a powerful way to overcome the challenges of Big Data, as AI is able to process enormous quantities of information in a matter of seconds or less. In fact, the more (accurate) data you have, the better AI will work.
There are many types of machine learning methods, each with their own pros and cons (see a full list here). The common thread though is that by productizing machine learning of this kind, marketing tech companies have reduced the burden on the average marketer, who would previously have needed a PhD in data science in order to take advantage of AI.
But the problem with the way traditional predictive analytics vendors apply machine learning/AI for B2B marketing is that they still require a large volume of first-party training data to make them work. If you don’t have enough first-party data, they won’t be able to produce an accurate model — regardless of how powerful or complex their AI is.
This is often the reason many marketing and sales teams are failing to see results with AI. For example, if you’re looking for new leads with obscure job titles like “Hacker-in-Residence”; or for new accounts within very niche fields or industries; or if you’re trying to predict which new vertical will be the best to focus on next — in all of these cases it’s highly likely your company just doesn’t have access to enough training data.
This is what is meant by “Small Data.”
Solving the Small Data Problem: Look-alike Modeling
Leadspace’s Look-alike Modeling is the long-awaited solution to the Small Data problem, enabling marketers to discover net-new accounts that resemble their best existing customers, which they otherwise would never have encountered.
Unlike with traditional predictive solutions, Leadspace customers can provide as little as a single example to generate a highly accurate, comprehensive, qualified list of equally qualified accounts. So if you’ve stumbled upon a particularly great account, but aren’t sure where to look for similar ones… well… you don’t have to. We’ll find them all for you.
Customers like Tipalti have already used Look-alike Modeling to execute highly effective account-based marketing (ABM) campaigns, by automating the difficult, tedious process of target account list-building.
Deep Learning: At The Cutting-Edge of AI
Our Look-alike Modeling works by applying advanced Deep Learning to Leadspace’s comprehensive data coverage.
Deep Learning is an advanced method of AI that’s currently at the cutting-edge of machine learning techniques. Without going into too much technical detail, Deep Learning mimics the human thought process, and enables machines to learn “rules” from the data autonomously, without human guidance. (This helpful article provides a great primer on Deep Learning if you’d like to read more.)
Using Deep Learning to build our predictive model is a step-up in terms of AI productization. The kinds of machine learning techniques used in most predictive models are not quite as fully-automated as the term “AI” might imply. These more basic machine learning methods still need someone (typically a data scientist) to specify which values in the training data are most important, and that means marketers will still need to be involved in the training process.
While Deep Learning — like any other form of AI — still needs training data, the process isn’t as manual as other machine learning methods, because Deep Learning “teaches itself” by picking up patterns like the shared characteristics of your ideal customers. (And of course, as mentioned above, we can provide all the data for you, so you don’t need to worry about that either).
That greater level of automation makes Deep Learning a more convenient, low-touch solution for marketers — and also so much more powerful if applied correctly. The obvious advantage is that AI can process huge volumes of information in a fraction of the time it would take a human to do so, so the patterns it identifies may well have been totally overlooked by your marketing and sales teams.
The “resemblance” we’re talking about here isn’t something as obvious to the naked eye as industry, company size, revenue, etc. These basic attributes can be identified using regular machine learning, or even with good old-fashioned human analysis. But they’re also blunt objects, as it’s rare that your ideal customers will be exclusively located in a single industry, or in a single company size range. What’s more, it’s highly unlikely that your ideal customers are defined by a single shared data set or characteristic; for a truly accurate model, you’d need to consider countless combinations of potentially dozens of granular insights simultaneously.
Deep Learning can uncover vital hidden trends within your data by connecting the dots to learn the subtle, common traits which truly separate your best customers from everyone else. This could be anything from a particular combination of existing technologies, to a certain sized IT team, to a very specific constellation of job functions within a company, department or series of departments, and so on — all things a human marketer or data scientist could simply never figure out for themselves given the near-infinite combinations of data sets. Similarly, a more basic AI/machine learning model couldn’t connect the dots “for itself”, as it still needs human guidance and training to get going.
In short: Deep Learning combines humanesque independent thinking with the enormous capacity of AI, to get the best of both worlds and provide a tool with immense potential for B2B businesses of all kinds.
“Leadspace Look-alike Modeling has enabled us to automate the targeted account list building process,” said Rob Israch, CMO of supplier payments automation leader Tipalti.
“Leadspace is a particularly strong system for building out an account-based contacts database,” he added.
“We serve some niche markets that can be difficult to prospect in — for example adtech and ecommerce. You can’t just buy a lead list from your typical vendor. So we needed a B2B database platform that had intelligence built-in, that could use predictive and AI to find the right prospects in the right space, and provide tailored lists based on slices that aren’t immediately obvious.”
“Leadspace continues to invest in both the data and the AI methods that produce the best results for our customers,” says Amnon Mishor, Leadspace Founder and CTO. “Not all customers have the expertise in-house to know which algorithm will produce the best outcome for a given scenario, or have access to a complete enough data set to model with.
“Leadspace brings together the data, modeling expertise as well as understanding of how it can be applied to help our customers achieve their business objectives.”
Read the case study below to
To learn more about Leadspace lookalike account modeling, or about the benefits of Leadspace Audience Management Platform more generally, contact us for a free demo.
Picture credit: iStock