The Data Cleansing Myth: Why "Clean First, Map Later" Destroys Value in ERP Implementations

Introduction

"Clean your data before migration." This piece of advice appears in virtually every ERP implementation guide, methodology template, and consultant's playbook. It seems so logical and straightforward that few question it. After all, wouldn't you want to start your new system with pristine data?

Yet after guiding numerous ERP data migrations and rescuing plenty of troubled implementations over the past 20 years, I can state unequivocally: This conventional wisdom isn't just wrong—it's actively destructive to your business. Premature data cleansing is one of the leading causes of failed migrations, extended implementations, and eroded business value in ERP projects.

The Dangerous Allure of Early Data Cleansing

The standard approach seems sensible on the surface. Before migrating to your new ERP system:

1. Identify "dirty" data in your legacy systems

2. Establish cleansing rules and procedures

3. Execute cleansing activities

4. Verify the cleansed data

5. Then begin mapping to your target system

This sequential approach appears logical and methodical. Unfortunately, it fundamentally misunderstands both the nature of data quality and the purpose of data migration.

Why Early Cleansing Fails: Three Critical Flaws

1. The Target Model Problem: Cleansing Without Context

You cannot effectively determine what constitutes "clean" data without understanding your destination. Each ERP system has unique data models, field requirements, and business logic. What might appear "messy" in your current system could be perfectly appropriate—even necessary—in your target environment.

Consider a recent financial services client who spent three months "standardising" customer category codes before engaging us. They consolidated hundreds of granular customer types into a dozen broad categories—only to discover their new ERP system required precisely the detailed segmentation they had just eliminated. The result? They had to rebuild these distinctions manually after migration, essentially undoing their cleansing work while adding months to their implementation timeline.

2. The Hidden Value Problem: Destroying Business Intelligence

What looks like "dirty data" to technical specialists is often valuable business information stored in non-standard ways. Without proper business context, cleansing teams frequently eliminate critical intelligence embedded in seemingly irregular data patterns.

In one manufacturing rescue project, we discovered a client's previous implementation team had systematically removed what they considered "inconsistent formatting" in product description fields. What they didn't realise was that maintenance engineers had developed a specific notation system within these descriptions that communicated critical equipment specifications. By "cleansing" these descriptions, they had inadvertently deleted irreplaceable institutional knowledge, forcing maintenance teams to create entirely new documentation systems after go-live.

3. The False Expectations Problem: Misunderstanding "Clean"

Much of what gets classified as "dirty data" is actually perfectly valid information that simply doesn't conform to predetermined expectations. This is particularly true when data cleansing is led by technical teams without deep business knowledge.

A healthcare provider's data team once flagged thousands of patient records as "dirty" because address fields contained unexpected information. Investigation revealed these were actually homeless patients for whom social workers had recorded shelter locations or contact arrangements instead of traditional addresses. The "irregular" data was precisely what made these records valuable to care providers.

The Real-World Cost of Premature Cleansing

The consequences of premature cleansing aren't merely theoretical. They translate into substantial business impacts:

A global manufacturer spent over £250,000 and six months on aggressive data cleansing before engaging us for their ERP migration. Their well-intentioned team had systematically removed custom fields containing critical supplier qualification information because they appeared unstructured. After migration, procurement teams couldn't identify properly vetted suppliers, leading to compliance violations and emergency re-implementation of qualification processes.

Another client, a distribution company, "cleansed" what appeared to be duplicate customer records, not realising their business model required maintaining separate accounts for different departments within the same customer organisation. The result? Billing chaos after go-live, damaged customer relationships, and three months of emergency data reconstruction.

The Right Approach: Discovery and Mapping First, Targeted Cleansing Later

Successful data migration requires inverting the traditional sequence. Rather than cleansing first and mapping later, the correct approach follows this progression:

1. Comprehensive Data Discovery

Begin with detailed analysis of your current data landscape. Understand what data exists, where it resides, how it's structured, and—most importantly—how your business actually uses it. This isn't just technical profiling; it requires engaging with business users to understand the context and purpose behind your data.

During discovery, you're not judging data quality; you're documenting reality. Every field, every relationship, every usage pattern is simply recorded as-is.

2. Detailed Mapping to Target

With thorough understanding of your current landscape, create comprehensive mapping specifications to your target system. This involves deep knowledge of both your legacy data and your new ERP's data models.

Proper mapping identifies:

• Direct field-to-field correspondences

• Required transformations

• Business rules that must be preserved

• Data relationships that must be maintained

• Target system constraints and requirements

Only through this mapping process can you truly understand what constitutes "quality" in your data context.

3. Issue Identification Based on Mapping

With mappings in hand, you can now identify specific data quality issues that will impact migration success. These aren't abstract "cleanliness" judgments; they're concrete problems that will prevent successful transfer to your target system.

For instance, you might identify:

• Missing mandatory fields required by the target system

• Values that violate target system constraints

• Relationship inconsistencies that would break referential integrity

• Formatting conflicts that would cause load failures

  
  

4. Targeted Cleansing with Clear Purpose

Only after completing the first three steps should you implement cleansing initiatives—and these should be precisely targeted to address specific, validated issues identified during mapping.

This approach ensures every cleansing activity has a clear purpose directly connected to migration success. It prevents the wholesale destruction of business value that often accompanies premature cleansing.

Case Study: The Right Approach in Action

We recently applied this methodology for a global manufacturing client who had been advised by their implementation partner to undertake a comprehensive six-month data cleansing initiative before beginning their Dynamics implementation.

Instead, we conducted a six-week data discovery phase, followed by detailed mapping. This process revealed that only about 30% of their data actually required cleansing for successful migration. The remainder was either already appropriate for the target system or required transformation rather than cleansing.

By focusing cleansing efforts only on genuine issues, we reduced their data preparation timeline by 70% while achieving higher quality results. Most importantly, we preserved valuable business information that would have been lost in a more aggressive approach.

Conclusion: Invert Your Thinking About Data Quality

Data quality isn't about conforming to abstract standards of "cleanliness"—it's about fitness for purpose. In the context of ERP implementation, that purpose is supporting your business processes in your new target system.

The next time you're planning an ERP implementation, challenge the conventional wisdom. Resist the urge to launch data cleansing initiatives before you've thoroughly mapped your journey. Understand your destination before you start packing your bags.

By inverting the traditional sequence—discovery and mapping first, targeted cleansing later—you'll not only save time and resources but also preserve the critical business intelligence embedded in your data. Your implementation will be more efficient, your data more valuable, and your business outcomes more successful.

Stop wasting resources on premature data cleansing. Focus on understanding your data landscape first, then target your efforts precisely where they'll deliver value.