Repairing the CRM link

Customer relationship management is one example of the over-hyped, under-delivering software booms of the late 1990s. Danny...

Customer relationship management is one example of the over-hyped, under-delivering software booms of the late 1990s. Danny Bradbury investigates why so many integration projects to support CRM systems went wrong, and unearths some sure-fire ways of knitting your systems together.

Customer relationship management (CRM) has become a dirty word in the IT sector. At the height of the dotcom boom, almost every supplier had a CRM product, and all of them were promising huge benefits to organisations which were willing to use these products to gain a holistic view of their customer base. Three or four years later, CRM has joined a growing pile of other once-popular three-word concepts that failed to deliver, such as application service provision and wireless application protocol.

There are many things that went wrong with CRM, but one of the most common problems was a lack of integration at both a cultural level and a technological one. Many consultancies selling CRM as a business proposition in the late 1990s took an all-or-nothing approach. Gaining a company-wide view of the customer called for company-wide integration of your systems, went the philosophy. The problem was that integration was much harder than people thought.

Even in 2002, when you would think that basic management theory would have filtered through to even the most provincial of organisations, one of the biggest problems with integrating systems to support CRM is that companies still do not understand what they need to integrate, says Chris Wakerley, managing consultant at CRM consultancy Cobalt Technologies.

Taking the wholesale integration approach and tying everything together at once is complex and expensive, and fails to deliver the most value for businesses at an early stage, he explains. The early return on investment is particularly important in today's economic climate, making it more critical than ever for CRM implementers to be selective in the data they integrate.

Nick Rhodes, principal consultant at Extraprise, another CRM consultancy, recalls an energy customer with whom he was dealing. The IT department knew the system configuration that it wanted, but the company did not have a business roadmap against which to validate the proposed system. "IT over the past 20 or 30 years has been forced to take a disciplined approach to project management, but I don't think that the front-end on the business side has that discipline," Rhodes says.

In building a long-term business plan for IT, business managers must ask themselves which elements of data are most important to them when building a long-term relationship with the customer, says Wakerley. Different things matter to different customers at different times, and all of these data elements must be available to call centre staff, who must also possess good customer-facing skills.

In trying to bring the IT department in line with the needs of business managers, companies need to base their business blueprint - the document that sets out the short- and long-term goals for creating a customer-centric view - on business processes, argues Wakerley.

Understanding how a customer charts a path through the organisation, which activities they are likely to take part in and which teams they are likely to deal with in various logistical areas will enable you to better define the data that you need to integrate at an early stage.

At the same time, says Rhodes, it is important not to go too far the other way; working to understand what the business needs will help you to get your integration process right, but on the other hand if you spend too long examining your data structure to try and pluck low hanging fruit, you may find yourself stuck in analysis paralysis. Structuring your analysis around business processes will help you to become more focused.

Mathew Stewart, head of CRM strategy at Cap Gemini Ernst & Young, says that deciding which data elements to integrate is only a small part of the battle. Different people within your organisation will own different sets of data residing on disparate systems.

This raises both technical and political questions that need answering before a project can be completed. For example, should two different systems be rolled into a single one? If not, then one system will have to be the master with the most up-to-date records, and the other the slave that reads from the master's file. Are the owners of both systems prepared to accept the decision, and if they are, do they understand the technical challenge involved in carrying it through?

"If you're putting in a champion system for CRM, you can make a strong case for holding the customer file in that system," says Stewart. "But then you may get three-quarters of the way down, and find that the legacy people can't handle passing off the ownership of that data," often because they haven't understood the implications on their own systems.

Things become even more complicated when politics and technology mix. Stuart explains that often, technically capable people will be unable to agree about the interfaces that they should implement for political reasons. For example, a call centre manager may be unhappy at a request from another business unit to conduct validation checks on data produced by his department if this validation increases the length of the average call.

It is for these reasons that it is important to secure the buy-in of senior management in any integration project. This is especially true for an integration project supporting a CRM system because the latter is so business focused. "Sometimes you need to get to the chief executive to understand his priorities: is it cheap, low-cost services; premium, quality service; better company information or better cross-selling? If you try to take these decisions in a vacuum, that's not going to get you the best system," warns Stewart.

Unfortunately, IT departments wanting to integrate systems to support a CRM project are stuck in a catch-22 situation, because getting senior management buy-in relies heavily on quick results. This is another reason to take baby steps with your CRM project, adopting modest integration goals to start with and building up to a grander project later by garnering management support through early successes.

At the same time, it is still important to maintain your long-term vision, says Jonathan Radcliffe, an analyst at Gartner Research. Failure to maintain a wider view can affect your long-term integration strategy, creating a loosely coupled CRM project with isolated silos of information.

If you're planning to focus heavily on analytical CRM where data will be mined extensively, you should pull all of the necessary data into a single warehouse that can be used to feed the CRM system, says Ian Dunsire, managing director of Atos KPMG Consulting. You must have such a view of your data to understand the business from a customer perspective, he explains.

There are some processes that you will have to go through to build such a datawarehouse. Renovating a house properly requires a thorough overhaul of its basic infrastructure, checking everything from wiring to plumbing. Without this, anything that you build on top of it will be vulnerable to failure. Similarly, data modelling is a vital part of any integration process.

It is important that you use physical data models that map the various elements of your data to physical resources and entity relationship diagrams that help you to visualise your data structure logically and match it to business processes. Unfortunately, it is a time-consuming and expensive undertaking, which won't deliver any immediate return on investment, making it invisible to the board.

On the other hand, modelling your data gives you a good opportunity to identify those data elements that are dispensable, and which pose a drain on the organisation. Eliminating them - effectively spring-cleaning your data structure - will ease the integration task ahead, especially if your fledgling CRM project gets off the ground and develops into something grand. It also means you can assess your data quality, and do something about it if it isn't up to par.

Dunsire says he has known banks that have tried to merge databases and ended up sending mailouts to customers containing the wrong data. Nothing is likely to destroy your customer relationships faster than that, so this is a critical issue. "Data cleansing needs to be done on a regular basis, but the big problem is to go right back to the source of that data," he warns. "You need proper rules in place, so that you understand that when you catch a piece of data, you go to a master customer file and then change that file."

Astute readers may notice a conflict between the idea of a datawarehouse - generally perceived as a huge bucket of data in which the value comes from the sheer scope of the data covered - and the "gently does it" approach advocated earlier, in which integration happens on a piecemeal basis. Dunsire argues that the two are not mutually exclusive.

When he worked at the TSB bank, he explains that the company decided to build a static datawarehouse with basic demographic data first, to prove its worth. This gave the company useful information about the way that the customer base split down. When the bank felt ready to start storing dynamic data in the warehouse, documenting every contact that the customer had with the bank, it became 10 times more powerful. But it learned to walk before it broke into a run.

Jennifer Kirby, analyst at Gartner Research, argues that for integration to truly succeed, it should happen at a cultural level as well as a technological one. She explains that some companies she is working with are now experimenting with fluid teams, where people with different competencies come together to service the customer in different areas.

This move away from a purely product-centric view of a company, to one in which teams try to see things from a customer perspective stretching across different product areas, supports a customer-centric view which should always have been a central tenet of any CRM project.

In the post-dotcom fall-out, many product categories and industry sectors will die out altogether, or least be severely pruned back. Others, which had a basically good philosophy but which suffered from a lack of good implementation, may well survive. CRM is likely to fall into the latter category, but some of the companies that have tried to implement it and some of the suppliers that have been selling it need to adopt a more mature approach to supporting the integration process at a cultural and a technical level.

Ten tips for integrating business processes
  • Talk to your business managers and make sure that you have a clear understanding of your integration goals as part of the business blueprint

  • Use data modelling techniques to understand where your data is located, and who is using it

  • Take baby steps - don't try and solve all your company's customer relationship management problems at once. Pick specific business processes that can be integrated relatively easily for a high return

  • Don't make the mistake of isolating these small steps too much. You want them to form a cohesive whole further down the line

  • Make sure that your CRM system is supported by business processes that stretch throughout the organisation. CRM is no good if it is not supported by logistics, for example

  • Be aware of internal politics between different business units that may hinder your integration process

  • Let the business requirements guide your choice of integration tool, rather than the other way around

  • Try to find a balance between meeting the needs of the business and falling into the analysis paralysis trap, where you spend all your time evaluating integration requirements and never get around to implementation

  • Support the technical integration underlying your business processes with a cultural one, tailoring your business so that it can support a customer-centric view, rather than simply a product-centric one

  • Make sure that you get senior management support before you begin your integration project.

Read more on Business applications