SalesPredict Offers Highly Automated, Highly Flexible Predictive Modeling

A couple of weeks ago, I wrote that “predictive everywhere” is one of major trends in data-driven marketing.  I meant both that predictive models guide decisions at every stage in many marketing programs, and that models are used throughout the organization by marketing, sales, and service.

I might have added a third meaning: that systems to do predictive modeling are everywhere as well. SalesPredict is a perfect example: a small vendor with a powerful system that just launched earlier this year. Back in, say, 2008, a product like this would be big news. Today, I simply add them to my list and try to understand what makes them different.

In this case, the main technical differentiator is extreme automation: SalesPredict imports customer data, builds models, scores current records, and deploys the results with virtually no human intervention.  This is possible primarily because the painstaking work of preparing data for analysis – which is where model builders spend most of their time – is avoided by connecting to a few standard sources, currently Salesforce.com and Marketo with HubSpot soon to follow. Because it knows what to expect, the system can easily load customer data and sales results from those systems.  It then enhances the data with business and demographic information from public Web pages, social profiles, and third party sources including Zoominfo, InsideView, and Orb Intelligence.  Finally, it produces models that rank customers based on how closely they resemble members of any user-specified list, such as customers with deals that closed or who failed to renew.  Results appear as lists in a CRM interface or as scores on a marketing databaset. The whole process takes just a few hours from making the Salesforce.com connection to seeing scored records, with most of the time spent downloading CRM data and scanning the Web for other information. Once SalesPredict is installed, models are continuously updated based on new CRM information and on feedback provided by users as they review the scored records. This enables the system to automatically adjust as buyer behaviors and conditions change.

User interface is a second differentiator. CRM users see a ranked list of customer records with a system-assigned persona derived using advanced natural language processing, suggested actions such as which products to offer, and the key data values that influenced the ranking.  Users can drill further into each record to see more customer and company information including previous interactions, products owned, and won or lost deals. The company information is assembled from internal and external sources using SalesPredict’s own matching methods, so results are not at the mercy of data quality within the CRM. As previously noted, users can adjust a ranking if they feel the model is wrong; this is fed back to the system to adjust future predictions. Another screen shows which data values are most powerful in predicting success.  This helps users understand the model and suggests criteria for targeting increased marketing investment. Although there’s no great technical wizardry required to provide these interfaces (except perhaps the name and account matching), they do make results more easily understood than many other predictive modeling products.

The final differentiator is flexibility.  The system can model against any user-defined list, meaning that SalesPredict can score new leads, identify churn risk, or find the most likely buyers for new products. Recommendations also draw on a common technology, whether the system is suggesting which products a customer is most likely to buy, which content they are most likely to download, or which offers they are most likely to accept. That said, SalesPredict’s primarily integration with Salesforce.com, user interface, and company name itself suggest the vendor’s main focus is on helping sales users spend their time on the most productive lead.  This is somewhat different from predictive modeling vendors who have focused primarily on helping marketers with lead scoring.

Is SalesPredict right for you? Well, the automation and flexibility are highly attractive, but the dependence on CRM data may limit its value if you want to incorporate other sources. Pricing was originally based on the number of leads but is currently being revised, with no new details available.  However, it’s likely that the company will remain small-business-friendly in its approach. SalesPredict currently has about 15 clients, mostly in the technology industry but also with some in financial services and healthcare.

KXEN Packages Automated Predictive Models within Salesforce Apps

As I mentioned in my Marketing Automation Beer Goggles post, KXEN introduced a free lead scoring app  for Salesforce.com users at Dreamforce. KXEN has since given me a closer look at the lead scoring product, underlying technology, and future plans.

First for details on Predictive Lead Scoring itself. As originally reported, it’s free and requires no configuration to set up. The trade-off for this simplicity is users have no control over which variables are included or what the models predict. The variables will include all standard and custom fields on the lead object, which isn’t too bad except that there might be useful data on related objects such as activity details.  At best, marketers could summarize such data and add the summary to the lead object, but that requires human intervention. Predictions are limited to conversions from lead to contact.  This isn’t always what you want, but it does stay within the lead object’s contents.

Some of these limits might be relaxed in future versions of the app.  However, KXEN is wary of making deployment more difficult or letting users make poor decisions such as removing variables they should keep. The system does provide reports showing the contributions made to the scoring formula by different variables and by values within those variables. Recognizing that this is already more than many users will care to review, KXEN plans to add simpler reports over time.

Although the lead scoring app attracted more interest than KXEN expected, it was really developed to illustrate the power of KXEN’s new “Cloud Prediction” model-building engine.  This uses the same automated modeling methods as KXEN’s established on-premise product. What’s new is a REST API that lets external applications send inputs over the Internet, wait while the engine builds a new model, and then receive the completed model formula. Scoring happens within the external application itself – Salesforce.com in this case – allowing the system to update scores as Salesforce data changes without running on KXEN’s own servers. Similarly, the app relies only on data stored within Salesforce.com’s own database, so KXEN doesn’t have to keep a copy.

The limits of the lead scoring app are a design choice: the cloud prediction API allows as much end-user control as KXEN’s on-premise system. But KXEN isn’t planning to expose the full API any time soon.  Instead, it's pursuing the app-based model as a way to expand use of its technology beyond its current base of relatively skilled users.

The next Salesforce.com app KXEN will release – pretty much any day – will tackle prediction of the “next best activity” for a given customer. This is substantially more complicated than the lead scoring app, since it creates a separate predictive model for each activity and then chooses the activity with the highest probability of response in each situation.  This one won’t be free: list price is $50 per user per month.

The next best activity app also requires more user effort to set up, since users must define the activities to model and specify eligibility rules for each activity. The system recommends activities randomly at first, to build some experience with different situations.  After the initial models are built it will still make occasional random selections to keep the models current.  Unlike lead scoring, the next best activity app reads from several Salesforce.com data objects in addition to the lead object.  Eventually, users will get control over which data elements to include.

The next best action app also relies on data stored within Salesforce.com. It won’t store a full history of offers made and rejected, because this would take more data than Salesforce.com can economically hold. That means the system won’t know when agents decide not to make the recommended offer. This can be a problem because the models based on a user-selected subset of cases.  It's a common issue with recommendation systems.

Whether these and similar issues cause serious problems for KXEN's cloud modeling apps remains to be seen.  Some amount of skilled human intervention may be essential to apply modeling effectively.  But it's worth exploring what's really needed: the ever-growing volumes of data and decisions make low-cost, automated predictions increasingly important for marketing success.

 

Rapid Insight Provides Low-Cost Options for Desktop Data Transformation and Predictive Modeling

Summary: Rapid Insight offers low-cost desktop tools for data transformation and automated regression modeling. They're a good choice for companies that need something simple yet powerful.

Predictive modeling is widely used by consumer marketers to select names for mailing lists and to decide which products to offer existing customers. These models are typically built by statisticians with tools like SAS and SPSS. In other cases, marketers can build them for themselves with automated tools like KXEN that are tightly integrated with the marketing automation system.

But most marketers still don’t have a marketing automation system or even an integrated marketing database. Nor do they have the skills to use a product like SAS. This group needs stand-alone tools to do two key things: assemble data from multiple sources, and build and execute the models themselves. (Okay, three things.)

Rapid Insight offers exactly those two (or three) capabilities in a reasonably priced package.

Veera is the data assembly tool. It lets users connect to most standard data sources and then define a processing flow to filter, merge, aggregate, transform and otherwise manhandle data into a form that makes it useful. Veera also provides some basic analytics including descriptive statistics (mean, median, value frequencies, etc.), cross tabs and graphing. The flow is set up as a sequence of icons with a drag-and-drop interface, which means users don’t have to learn a scripting language. Rather than go into more details, I'll just point you to the vendor's on-demand demo.


The predictive modeling tool, prosaically named Analytics, builds logistic and least squares regression models. (Logistic models predict yes/no outcomes such as whether someone will respond to a promotion; least squares models predict continuous numeric outcomes such as lifetime value.) The Analytics interface is more sequential than Veera: users get a set of tabs that lead them through the steps of loading data, selecting variables, building the model itself, assessing the results, and scoring an audience. At each step along the way, users can make their own decisions or allow the system to choose for them.

I’ve seen quite a few automated modeling systems over the years, and was impressed at how well Analytics provides users with information to understand what's happening and take control when desired. This should let the system satisfy knowledgeable statisticians looking for a productivity enhancer, as well as novices who want to rely on the system's choices. Analytics also has a good online demo.


Veera and Analytics both run in client/server or desktop configurations. They load data into system memory (RAM), which means very large projects could be problematic. The vendor says a half-million rows with a couple hundred variables is a reasonable universe to model.

The two products are sold separately. This makes sense: many companies could use a generic data assembly tool like Veera for purposes other than modeling. For example, marketers might use it to construct a multi-source marketing database for promotions or analytics.

Pricing is $3,000 for the first Veera user and $5,000 for Analytics, with discounts for additional licenses. This is quite reasonable compared with other automated modeling systems, although other products often provide more than just regression models. Annual maintenance for each product is $1,750 per license. Rapid Insights has been selling its products since 2005 and has more than 150 clients with over 200 licenses.