Lattice Engines: Predictive Model Building

The Lattice Data Cloud Explorer highlights the top fields by category and helps admins determine which fields should be exported to other platforms.
The Lattice Data Cloud Explorer highlights the top fields by category and helps admins determine which fields should be exported to other platforms.

Lattice Engines has taken the pole position in the emerging Predictive Analytics space.  In yesterday’s blog, I covered its pricing, value proposition, content, and integrations.  Part two covers model building.

When first launched, Lattice Engines and its peers had long deployments and black-boxed models that required data science expertise.  The firm now offers 24-hour deployments, simplified model building, and greater transparency around models and recommendations.  Furthermore, the system allows marketers to either build their own models or import industry standard PMML files constructed by their data science teams.

Predictive models are built by importing training files which are matched against the Lattice Data Cloud using D&B DUNSMatch logic and Lattice proprietary techniques.  Training models contain examples of both positive and negative outcomes (e.g. win / lose, renew / drop).  A model is typically available within thirty minutes of the training file upload.

Ideal Buyer Profile scores (Lattice’s term which is similar to Ideal Customer Profile scores) are available to sales and marketing and include both scores and recommendations.  Marketing can view the model via a graphical Data Cloud Explorer which highlights the key signals and variables in the model and makes the data available for export to other platforms.

To make the data more actionable for sales reps, Lattice provides Salesforce Talking Points which display recommendations and explanations that include Lattice data, transactional history, and buyer behavior.  A Lattice Buyer Insights CRM I-frame contains Lattice recommendations, talking points, company profiles, company fit, engaged contacts, engagement activity, intent analysis (surging topics), web activity, and purchase history tabs.

Lattice Recommendations with related scores. Sales reps can explore any of the recommendations by clicking on them.
Lattice Recommendations with related scores. Sales reps can explore any of the recommendations by clicking on them.

Future plans include a user interface for segmentation analysis and simplifying intent scoring to high/medium/low.


Part 1: Lattice Engines Overview.

Lattice Engines: Leader in B2B Predictive Analytics

Lattice Scores and Enriched Data are available within the Eloqua Canvas campaign builder.
Lattice Scores and Enriched Data are available within the Eloqua Canvas campaign builder.

In a 2016 survey of predictive analytics companies, Gartner sized the global market at between $100 and $150 million. Although Gartner remains bullish on the sector, the size must be disappointing to both the firms in the space and their investors. One of the early companies in the space, Lattice Engines, continues as a market leader with over 200 global deployments.

Lattice Engines supports both enterprise clients and high-growth companies with deployments beginning around $75,000. Pricing is based upon the number of managed leads or contacts in the instance along with the number of users. With revenue between $25 and $50 million (GZ Consulting estimate), the firm has a strong position in the nascent market.

Lattice Engines combines first and third-party data to build predictive models. External content includes firmographics, intent data, technographics, social data, and web crawled business signals. Content is licensed from leading vendors such as Dun & Bradstreet (WorldBase global company file), Bombora (intent captured from over 3,000 B2B media sites), and HG Data (technographics). The Lattice Data Cloud covers over 200 million global companies, 21,000 buying signals, 100 million tracked domains, and over one billion daily interactions. Internal content spans transactions, CRM, marketing behavioral data, usage data, and support services.

“Predictive analytics is one of the few types of marketing technology that has the ability to solve issues at every step of the funnel, because it aligns sales and marketing against the right targets, and provides them with the right data to create targeted campaigns.  By infusing fit and intent data into our models we enable teams to have a complete understanding of their ideal customer profile, which enhances the programs teams orchestrate against their targets.”

  • Director of Corporate Marketing Caitlin Ridge.

Firms can build multiple models to support various geographies, product lines, and scenarios (e.g. win/loss, upsell/cross-sell, renew/churn). Lattice scores and modeled data are integrated with many of the key SalesTech and MarTech platforms:

  • Ads/Web: DemandBase, Oracle Data Cloud, doubleclick (Google), AdRoll, Facebook
  • MAP: Marketo, Oracle Marketing Cloud (Eloqua), Pardot (Salesforce)
  • CRM: Salesforce, MS Dynamics, Oracle Sales Cloud, SAP

This platform coverage enables Omni-channel ABM campaigns across programmatic platforms, email, direct mail, and field marketing.  Scores, insights, and recommendations are provided to sales reps within CRM i-frames.

“Lattice remains the most visible “face” of the market,” said Gartner analyst Todd Berkowitz in September 2016. “With its focus on security, level of integrations and ETL tools, the company is a fit for enterprise clients (both in high-tech and other industries) and/or companies planning to deploy in multiple regions. Gartner clients report that the company’s go-to-market approach is unique in the way it addresses complex problems and help customers operationalize the insights from the models. Lattice is one of the few vendors that can recommend key plays at both the lead and account level across the entire funnel.”

According to Lattice, customers enjoy a broad set of improved metrics:

  • 2X Higher Conversion
  • 3X Greater Pipeline
  • 35% Higher Deal Sizes
  • 6% Increase in Quota Attainment
  • 85% Rise in Revenue per Customer
  • 20% Reduction in Customer Churn

The firm sells broadly across B2B sectors.  Customers include Amazon, Dell, PayPal, Staples, and SunTrust Bank.


Tomorrow’s blog will cover core Lattice Engines model building and recommendations.

SFDC Einstein: Once Again We’re Discussing AI

einstein-artificial-intelligence-in-business

In the late 1980’s, when my career was first beginning, I worked on a technology helpdesk for an insurance agency automation system (Aetna’s Gemini platform).  Many of the calls were routine with an easily road mapped set of resolution steps.  So the firm decided to invest in artificial intelligence (AI) and began interviewing its most seasoned experts to identify the problem resolution path.

After several months of development, an AI module would be unveiled that walked the user through problem resolution.  It was basically a set of if-then-else and case statements providing pre-coded branching logic.  Support reps started with a category and were walked through a set of questions to ask and resolution steps to convey over the phone.

The solution was expensive and lacked the ability to learn.  Thus, if new problems arose or the problem resolution changed due to new hardware or software being introduced, the rules no longer applied.

It was far from intelligent.  Heck, I’d coded a twenty-questions game in a first semester programming class that was more intelligent than the service.  At least my Q&A game had the ability to learn new questions to ask without requiring an expensive consultant.

Finally, it was only used by new hires as much of the routine steps were just that — routine.

Solutions like this quickly proved that Artificial Intelligence wasn’t intelligent and after a few years, the term AI fell from favor and returned to the realm of sci-fi killer robots.

Nearly three decades later, the term AI is once again being rolled out.  But now it does convey an impressive level of intelligence which makes our devices feel smart.  It’s why we call them smartphones.  They are able to leverage vast amounts of data and make decisions in the blink of an eye.  Whether it is asking Siri a question or having Google map the best route to a location subject to current traffic patterns and transportation mode, we expect our devices to be intelligent.

AI represents a massive change in technology. You might call it a “paradigm shift” or “disruption” or we could just stick with “massive change.” What we’re trying to say is, AI is kind of a big deal. And just like the arrival of the personal computer, cloud computing, and the mobile smartphone, AI is going to fundamentally change the way things work, forever.

AI is not killer robots. It’s killer technology.

So it was with a smile that I saw the term AI being used by Salesforce in positioning their new Einstein service.  Each year at Dreamforce, CEO Marc Benioff discusses a new underlying technology or cloud.  Most recently it has been Lighting (UI and workflows), Wave (analytics), and the Internet of Things Cloud.  At Dreamforce 2016, it is Einstein, their artificial intelligence platform to assist with sales, marketing, and service.

Salesforce presents AI simply as

Lots of data + cloud computing + good data models = smarter machines

So while much of this technology has been provided as consumer applications for over a decade, businesses have been lagging behind when the scope goes beyond a mobile app or e-commerce portal.

Shouldn’t the full transactional and service history be available to help understand past purchases, preferences, and potential cross-sell and upsell opportunities?

Wouldn’t we want it delivered no matter the touch point?

That is the type of intelligence that Einstein is looking to bring to Salesforce customers.  Einstein is “the world’s first comprehensive artificial intelligence platform for CRM. I’ve never been more excited about the innovation happening at Salesforce,”  said Benioff.

Einstein is available both programmatically (for developers) and “declaratively for non-coders,” said Benioff.  It is integrated directly into the SFDC platform and available across all of the clouds.  For example, an Einstein widget displays a set of insights identifying competitor news, recommended actions, and account intelligence.

Einstein Insights Widgets provide intelligence both programmatically for developers and data scientists and declaratively for end users.
Einstein Insights alerts widget.

 

Einstein can surface competitor mentions even if the end user hasn't trained it to do so.
Einstein Insights surfaces insights both programmatically for developers and data scientists and declaratively for end users.  It can even infer competitors from emails and deliver alerts within SFDC widgets.

 

 

 

 

 

 

Einstein builds models with no coding or initial training by users.  For example, the system is able to determine which trigger events are important to sales reps and surface news about competitors without asking “who are your competitors?”  The system also can make recommendations concerning high-scoring leads based upon both fit (firmographics, biographics) and behavior (e.g. recent viewing of a demo).

Einstein recommends actions to sales reps. In this case, it is suggesting an email requesting a meeting be setup with the VP of Sales at a high-scoring account.
Einstein recommends actions to sales reps. In this case, it is suggesting an email requesting a meeting with the VP of Sales at a high scoring lead who recently viewed a product demo on the website.

Not only does the system recommend activity, but it then offers recommended email copy including a proposed call time.

The platform is built on a series of recent acquisitions including RelateIQ (rebranded SaleforceIQ), MetaMind, Implisit, PreductionIO, and TempoAI.  The firm now has a team of 175 data scientists “stitching together this amazing platform,” said Benioff.

“The new platform will “democratize artificial intelligence” and “make every company and every employee smarter, faster and more productive,” continued Benioff.  “This is going to be a huge differentiator and growth driver going forward as it puts us well ahead of our CRM competition once again.”

The new platform infuses their sales, cloud, and marketing platforms with AI capabilities for “anyone” regardless of their role or industry.  According to Salesforce, Einstein lets employees “use clicks or code to build AI-powered apps that get smarter with every interaction.”

Einstein is positioned as having your own data scientist focused on applying AI to customer relationships.  Einstein has access to a broad set of intelligence including CRM data, email, calendar, social, ERP, and IoT to “deliver predictions and recommendations in context of what you’re trying to do. In some cases, it even automates tasks for you. So you can make smarter decisions with confidence and focus more attention on your customers at every touch point.”

Several predictive analytics companies used the launch to shout, “hey wait, we’ve already mastered AI for sales and marketing.”  LeadSpace CEO and former Salesforce CMO Doug Bewsher stated, “B2B marketers need a complete solution that works across multiple channels, in their existing marketing stack.”

“Bad data is the Achilles heel of AI,” continued Bewsher. “AI is only as good as the data available to it. Marketers who want to get the full benefit of AI need to address their data problems first, or they’ll see the same diminishing returns as with traditional marketing automation.”

Shashi Upadhyay, CEO at Lattice Engines was a bit more diplomatic in welcoming Einstein.  “After having led the market for several years, we are really excited to see the mainstream attention shifting towards AI-based solutions for marketing and sales.  The Einstein announcement from Salesforce is a great step forward, as it will serve to educate the market and signal that predictive solutions are here to stay.”

Image Credit: Salesforce.com

HBR: Digital Exhaust for Sales

Mark Kovac of Bain and Company wrote an interesting piece on the topic of digital exhaust in Harvard Business Review.  The short piece, titled “Using Digital Exhaust to Improve Sales,” provides three examples of how software vendors are combining big data and analytics to provide new tools to sales management.

Kovac defines digital exhaust as “the data generated from the regular activities of a sales force or their customers, to change the behavior of frontline sales representatives in ways that dramatically improve sales productivity and effectiveness” and provides examples from three firms including Lattice Engines.

The first firm, Volometrix, was acquired by Microsoft last year.  Volometrix performs resource analyses to determine where sales reps are spending their time and which behaviors are positively correlated with sales performance.  Unfortunately, the use case Kovac provides is about a firm that realigned company priorities providing more time for selling and therefore “too much sales capacity.”  While efficiency is desirable, if software solutions result in efficiency improvements without much improvement in efficacy (ability to sell), then they will be resisted.  This isn’t too say that Volometrix doesn’t provide efficacy gains (I only know them from the story), but case studies which focus on cost savings (efficiency) over revenue gains (effectiveness) may create situations where sales reps refuse to cooperate.

The second vendor discussed is GoToMeeting which is performing voice-to-text semantic analysis and discerning which phrases and approaches are more effective.  The data gathered is anonymous (though I still see issues with recording calls, particularly in certain jurisdictions) and provides insights in how to more effectively sell.  The software sounds a bit like SalesforceIQ but instead of focusing on email analysis, GoToMeeting is using conversations.

Lattice Engines stratifies leads by probability of closing with the highest probability leads immediately sent to sales and the remainder held for nurturing.
Lattice Engines stratifies leads by probability of closing with the highest probability leads immediately sent to sales and the remainder held for nurturing.

The final case study was predictive analytics company Lattice Engines which helps firms improve “call response rates, close rates and average order value.”  Predictive analytics for sales and marketing is a growing class of recommendation software with a broad set of competitors including Leadspace, Infer, 6Sense, and Mintigo.  Lattice Engines combines first-party and third-party data sets to score leads and recommend sales approaches.  Unfortunately, Kovac focuses on the matched third-party data instead of the digital exhaust captured by the firm (he does mention loyalty scores and product purchase history).

There is a growing set of SalesTech vendors that are applying digital analysis to previously inaccessible datasets.  While sales remains an art, these solutions are shifting the sales profession from a craft to a science.  If vendors are to be successful, they need to focus more on efficacy versus efficiency.  While sales could certainly become more efficient by reducing non-productive administrative activities (and reps are always happy to reduce dead weight time spent tracking problems, navigating contract signatures, and entering data), the focus should be on top-line growth and exceeding quota, not cost reduction.  Otherwise, sales reps will resist their adoption.