Predictive Marketing Blog

The 2018 Predictive Marketing Landscape: From Manual to End-to-End Automation

Predictive analytics isn’t just for Fortune 500 companies these days. Brand marketers, at companies both big and small, are getting cozy with their data and are using it to their advantage.

Yet, with so many variations in predictive analytics solutions, it’s more difficult than ever to discern which program is the best fit for your company. It’s a task that can be maddening at worst and downright confusing at best.

So to help you gain perspective on approaching predictive data analytics for you, your brand, or your clients we’re taking a closer look at predictive analytics solutions to see what makes them tick.

I. Large Enterprise Analytics Providers

Enterprise analytics solutions, while usually focused on business intelligence (reporting, OLAP, visualization, etc.), often include strong predictive capabilities, typically provide custom models, and build profiles detailing the unique characteristics which make someone a good prospect.

A custom analytics solution is incredibly heavy software which requires implementation and is specifically built for data scientists, not marketers. Often, these solutions provide very little in terms of capabilities to clean and enhance your data. Custom analytics solutions require lots of manpower, and even after these solutions are set up, it can take weeks or months to receive a custom report since only parts, if any, of the solutions are automated.

There are many of these custom tools for data analysts and scientists, meanwhile marketers are often held up waiting on teams to build filters and models which don’t always guarantee results.

II. Traditional Data Vendors

Today, some data vendors who offer customer lists claim to offer predictive capabilities, but these are often “pseudo-predictive” solutions, meaning the model is a generic, pre-built model. In these cases, it is the company that must fit the model.

Many of the lists provided by these solutions are pre-built, which is a large part of why they are not client-specific and why it is the company that must fit the model and not the other way around. These lists generated from pre-built models are built on only a handful of use cases where you pick a couple of univariate filters what you anecdotally believe will meet most instances.

We know that in the real world, marketing to consumers is not a one-model-fits-all solution and picking a handful of variables is not going to cut it. To make matters worse, sometimes traditional data vendors will black box their solutions, and they’ll do it in one of two ways:

  • The data vendor has a pre-built model that you never get to see, meaning you have no idea which variables were or were not involved in making the list they provided you with, or
  • The data vendor shows you the pre-built model but they don’t show you any of the names that you will be marketing to.

The best case scenario with traditional data vendors is that they offer a white box solution and show you both the model variables and list names, but even these solutions are semi-predictive at best.

More often than not with vendors, building lists is still a manual process. Their solution is not only pre-built, but it’s a process that could not have been performed without data scientists and wasn’t automated.

III. Automation of Niche Predictive Tools

Some niche vendors have automation, but not end-to-end automation, only performing part of the predictive process and working within specific industries or use cases. Niche predictive vendors offer automated modeling but don’t automate the wrangling – which covers roughly 20% of the predictive marketing process, the rest of the time is spent finding, cleaning and preparing data. So you might get a model from a niche vendor, but then have to take that model to another vendor for names. These solutions don’t necessarily have the wrangling or visual reporting, ease of scoring lists, ability to export data, appending variables, etc, and some require that you get your data wrangled by one tool, modeled by another, and some won’t give you a list.

Of the niche predictive tools, some still have not actually automated. This process, like the above solutions, requires a team of data scientists and means that receiving a niche predictive model is a very time consuming process.

Alternatively, predictive companies just don’t have access to enough third-party data sources, and only have industry-specific data. These companies can often provide models and lists of prospects for specific use cases only.

IV. Data Co-ops

Data co-ops are formed when marketers band together and create pools of their first-party transactional data. The upside to joining a co-op is that you gain access to more consumers and their past shopping behaviors, but you have to give the co-op your own first-party consumer data first and ensure that you were provided the right consent from your customers to resell or share their data to a third party.

Co-ops are typically used for list acquisition and data processing, but do little in the way of predictive modeling and lead scoring. For the most part, co-ops use the same black box approach that data vendors do, giving you a list of people based on some predetermined filters and expect you to blindly trust that those are your best prospects.

Hands down, the biggest downside to using co-op data is that everyone is paying in their own data. Not only do co-ops fall short in the same ways that traditional data vendors and niche vendors do, but you give all of your customers and prospects to your competitors. Sure, you receive a list of your competitor’s customers and prospects, but your competitors have a list of your current customers and prospects so now you’re all fighting over the same pool of people. To measure the impact of a co-op list on your business, you have to consider how much ROI you lost by giving away your customer and prospect lists.

V. True, End-To-End Self-Service Predictive Platform for Mid-Market Companies

Nowadays, speed and efficiency are the name of the game. This is where Reach Analytics really shines.

Our advanced, self-service predictive analytics solution uses best-in-class AI to target your best audience and generate predictive profiles which help you identify and reach consumers who are mostly likely to become a customer and respond to your campaigns, no data science experience needed. The advantage to this, of course, is quick turnaround on scoring and targeting so that you can immediately apply data about existing and new customers to your next marketing campaign.

Our self-service predictive platform is made specifically for marketers and allows brands and their agencies to get all the benefits of ultra fast, scalable predictive marketing – best-in-class wrangling, algorithmic modeling, visual reporting, transparent lists – without the overhead of a large data team or the added cost and hassle of installing huge customer predictive analytics platforms.

In fact, at Reach, we’ve gone up against other parts of the marketing landscape. We have held bake-offs with an enterprise client, who weighted the test in their brand team’s favor, but our modeling software has routinely outperformed them – even with a robust enterprise solution at their fingertips -, improving their response rates by 53% on average. In another instance, we went head-to-head with a data vendor on a net-new customer acquisition campaign, and not only was our model built in one-tenth of the time, but our model doubled the data vendor’s performance.

It’s the speed and efficiency of our software combined with our custom solutions that make us different. And you don’t want to miss it.

Request a free demo or sign up to try it for yourself.