Home > Articles > Data > SQL Server

SQL Server Reference Guide

Hosted by

Toggle Open Guide Table of ContentsGuide Contents

Close Table of ContentsGuide Contents

Close Table of Contents

Customer Relationship Management (CRM)

Last updated Mar 28, 2003.

Have you ever noticed that kids have a lot of brand loyalty? When I was a kid, I loved one cereal: Quisp. It's basically a sugar product held together by some sort of organic glue, which made it, of course, the best cereal in the history of the universe. I wouldn't let my mom buy anything else.

Adults are different. They are far less likely to stick with a particular brand of anything. They shop around more, and they're willing to change brands based on price, quality, performance, and a lot of other factors.

Plus, they buy almost never buy Quisp. As a matter of fact, now that I'm older, I've hidden its very existence from my daughter.

For companies, however, it's clear that it's far cheaper to keep an existing customer than to get a new one. So how do they battle the desire of their clients to move on to another product?

We've been looking at various uses of SQL Server as an application back-end. In this article, I'll explain Customer Relationship Management, or CRM. The goal of CRM is to help firms gain new customers and keep current ones. Today, we'll examine how CRM works, and how SQL Server fits into the equation.

Customer Relationship Management Defined

To survive, a firm needs to sell products. They sell them to customers, who in turn need service on the product and hopefully will buy more stuff. CRM software keeps records of all those activities, and routes that information between its various functions.

Put another way, CRM is the software that enables you to identify who your prospective customers are, whether they bought anything from you, what they bought and when, where they are and where they bought it, why they bought it, what problems they had with it, what you did about it, and how happy they are about all of this.

Looking at that description, you can see that CRM encompasses several departments in the company, such as Sales, Marketing and Customer service. In the past, these departments held either a duplicate or a separate view of a customer or process. CRM provides for a single view of the customer. Not only that, CRM allows for manufacturing, finance and even logistics to coordinate customer-based information.

This kind of information makes heavy use of two key technologies: databases and business intelligence. CRM has the distinction of having a highly transactional component as well as a heavy dependence on reporting and analysis. Those two functional requirements makes implementing a responsive CRM system a bit of a challenge.

To better understand how CRM works with SQL Server, let's follow the chain of a typical set of transactions. It's important to keep in mind that CRM vendors implement these features in various ways, so the systems you use might have a different approach to the process I'll describe here, or they might have even more features than I'll describe.

From the data-flow perspective, a CRM transaction begins with a potential customer, sometimes referred to as a prospect. The department that is most affected by this information is marketing – the people charged with developing sales for the firm.

At the very least, the information stored about a prospect includes typical contact data, and most systems store lots of demographic information as well. In addition, CRM systems include information about marketing efforts used to find those prospects. On the reporting side of that marketing effort is usually a software business intelligence component to determine the best use of marketing dollars. (Such as: Did including a free coupon for Quisp increase propsects' likelihood to buy our dental services?)

Once prospects buy a product, their information normally converts to client records. This doesn't mean that they are no longer marketed to – quite the contrary. Terms like up-selling, cross-marketing, and organic growth all have to do with getting current clients to buy more, better, or other products. The reporting side of these transactions normally involves sales metrics by product, salesperson, and so forth. The sales department and the marketing department share this information.

After the client receives the items, there might be service implications. A helpdesk function within CRM software tracks the issues the user has and what is done to resolve them. This information has a reporting implication as well, alerting managers to defective products or to inefficient helpdesk processes. As an offshoot of this information, a knowledge base might be built to allow both clients and employees access to issue resolutions. The sales department uses this information to measure defects on products sold, marketing uses the reports to understand the firm's strengths and weaknesses, and the helpdesk pulls information from the reports to build the knowledge base.

Behind these three functions lie the products that the firm sells. The product information, including costs, warranties, margins and the like, are used in yet more reporting, normally to develop a strategy for upper management.

CRM systems have many more features, but those are the core functions it provides and gives us enough material to begin our SQL Server discussion.

SQL Server and Customer Relationship Management

SQL Server is used in several commercial CRM systems, and of course you can create your own. Let's take a look at some issues you'll need to consider if you purchase a solution, and then detail some of the high-level decisions you need to consider if you want to build your own.

Deployment Decisions

The first thing to consider in a CRM deployment project is the number of servers. While many high-end applications recommend separating the application and database servers, CRM goes a bit further. In many cases, it makes sense to have at least two database servers in a CRM landscape. The reason for this is the dichotomy of the transactional and the reporting components of the application. When the business intelligence reports are run, severe delays can occur if the transactions are high. These separate servers are sometimes referred to as the "main" and "reporting" servers.

The second overall concern is the system's security. You normally have to allow clients (and even vendors) to see parts of the system, so a firewall for exposing the public portions of the application is a must. In addition, you'll have to maintain that security separately from your network integrated accounts.

Information distribution is next. This area includes mail integration and perhaps even telephony connection. Most certainly, you'll have a Web input/output component.

Memory: CRM systems vary in the amount of memory they need, but as a general rule, the reporting server will need more than the main server.

CPU: Most of the CRM systems I've worked with need at least dual processors on all servers. Blade farms are great for this arrangement.

Hard Drive: A SAN is best for any application, but with the dual SQL Servers make sure the rig you use can handle a "split" configuration so that you can separate the reporting server onto its own channel.

Network: The primary concern here is security. You'll want to work very closely with your network admins to make sure that the clients can see their information without compromising the network. Most CRM vendors provide a setup guide to help with this.

Design Decisions

As with most off-the shelf systems, you normally don't have to design the database for these commercial products. If you do, there are some basic design concepts to keep in mind.

As an overall design consideration, make sure you build the database objects so that they can be federated, or separated onto various systems. Even if the entire program sits on one server, the design should allow for this separation. This provides scalability and the ability to put reporting on its own system.

The first step is designing your primary application databases. These areas include products, prospects/clients, helpdesk, knowledge base, and sales and marketing. I've covered basic design in other articles, but there are a few overall concerns in a CRM system.

You'll need to focus on the Business Intelligence portion of the application, right from the outset. Make sure you plan for a separated reporting component. SQL Server's Analysis Services are perfect for this part of the system.

Because the information will be used by a mobile sales force, you'll need to set up replication. In most cases, snapshot or merge replication is best suited for the disconnected clients.

In addition to disseminating the information through replication, you'll need to plan for mail access, potentially both in and out of the system. SQL Server provides native functionality for this feature. Your requirements might also include a telephony interface, and SQL Server does not provide that function natively.

These are just a few of the design issues that you need to consider. Customer Relationship Management is a huge undertaking, so plan for a significant, lengthy development project.

Online Resources

ITToolbox has a good section on CRM where you can find a discussion on a lot of the issues I've mentioned here.

InformIT Tutorials and Sample Chapters

You can check out a sample chapter from the book "Managing Your CRM Project" by Jill Dyche on CRM here at InformIT. Also check out the books in Safari on CRM. There's quite a few.