Should You Outsource Customer Analytics? Everything You Need to Know Before Making the Decision.

March 20, 2014

Getting customer analytics right is easier said than done. This article will examine what is required for customer analytics and whether you should perform it in-house or outsource it.

What steps are required for customer analytics?

Analyzing customer data involves a lot more than “analysis” alone. Let’s break it down in 7 steps:


1. Collection

First you have to collect customer data from across all customer touch points, which means connecting to many different database and software systems. Think email systems, your website, your order processing system, your billing system, your customer relationship management system, your call center or help desk software, your loyalty program, and many more.

2. Data Cleansing

After you have collected all customer data it needs to be cleansed, de-duplicated, linked, and validated. You want to validate that email addresses and mailing addresses are correct, that names are spelled correctly, that no profanity has entered your database through a web form, and much more.

3. Data preparation

Before more advanced analytics can be performed, you will need to aggregate and summarize the data, and perform a series of “pre-calculations”. For a typical customer data set, about 400 calculations or more need to be made before the real analytics can begin.

4. Analysis

Only now can true customer analysis begin. Read our post on predictive analytics methods for marketers for an overview of the types of models that can be applied to customer data and why marketers should care.

5. Presentation

After you are done with the number crunching, the results need to be interpreted. Visualizing the data can greatly help humans with this task.

6. Actions

Eventually either humans or algorithms need to recommend a course of action based on the analytics. For example, every customer with high predicted lifetime value and low likelihood to buy you could get an offer.

7. Connectors

Finally, you need to set in motion actions based on your decision. In some cases you could trigger an action automatically: in the example above the offer could be sent automatically by connecting to an email service provider. Contrary to popular belief, the most time consuming tasks are related to data collection, cleansing, and other data preparation steps – not to the analysis itself. New data needs to be collected, cleansed, validated, enhanced, and prepared daily whereas the analytics models themselves can stay in place for much longer.

What people do you need for customer analytics?

For different steps in the customer analytics lifecycle described above, you need different types of people:

Collection IT staff/DB administrator/data replication engineer
Data cleansing Database administrator/BI engineer
Data preparation Business analyst or data scientist
Predictive analytics Data scientist
Presentation Business analyst or data scientist
Actions Data analysis or data-driven marketer
Execution Marketing manager


In addition to a data scientist you will also need IT resources, a developer or designer, as well as business or marketing executives who can help interpret the data and act on it. The Data Warehousing Institute recently performed a survey which concluded that knowledge of the business was a much more important skill for data analysts than knowing math. Extrapolating, it would make sense to outsource the number crunching while keeping the creative and interpretation of the data in-house. A model for the future could look something like this:

Collection Outsource to the cloud
Data cleansing Outsource to the cloud
Data preparation Outsource to the cloud
Predictive analytics Outsource to the cloud
Presentation Outsource to the cloud
Actions Data analyst or data-driven marketer


By grouping customers with similar business problems (for example retail marketers) providers of a cloud-based customer analytics platform can effectively “share data scientists” across their customer base, thereby significantly reducing the number of data scientists required in the US economy.

What systems do you need for customer analytics?

In order to do customer analytics fully in-house you would also need to build an Enterprise Data warehouse. A typical setup may involve hardware and software from the likes of an IBM, Teradata, or SAS that can cost you millions of dollars. ModernCustomerAnalytics Advantages of outsourcing customer analytics? There are strong arguments to outsource the data collection, data cleansing, data preparation, predictive analytics, and presentation – while focusing in-house on the interpretation on the data and the actions that were decided upon. Doing so could bring the following advantages:

1. Solve for the shortage of talent

The shortage of data scientists is well documented. Even if you find one, chances are they will receive regular offers from other companies and they may leave you soon. This website lists over 50 analytics programs that have been established at US universities recently to try and fill the gap.

2. Don’t go to jail

Performing analysis on customer data means handling personally identifiable information (PII), which can be risky. If you mishandle or abuse the data in your care, you might violate a variety of privacy laws and, in a worst case scenario, end up in jail. If you handle all PII yourself, you will need to find out how to properly protect it. This might mean hiring a security manager to help with the task.

3. Costs

Hiring your own in-house data science team and procuring the software and hardware they need to do their job can be a very expensive proposition indeed. “Sharing” IT infrastructure and “data scientists” is much more affordable.

4. Results

By leveraging learning across clients with similar business problems, often in similar industries – an outsourced provider (whether it be a consultant or a platform vendor) can deliver significantly better results.

How are other companies doing customer analytics?

For more points of view on whether to perform data-driven marketing and customer analytics in-house or to outsource it, join our upcoming webinar:

WEBINAR Wednesday March 26th 9:00-9:30am PDT

How to build data driven marketing teams.

Learn about data driven marketing teams from these experts:


Kaiser Fung

Kaiser Fung

Author, Numbersense; VP Analytics at Vimeo and Sirius Satellite Radio




Mark UhrmacherMark Uhrmacher

Former co-founder and CTO of Ideeli (also VP Marketing at Expert Foods)




Diego Saenz

Diego Saenz

President and CMO of (previously CIO at PepsiCo)




Register now for this unique virtual event on building data-driven marketing teams.