Why Messy Data Isn’t Always a CRM Data Cleansing Problem

The Enterprise-Sized CRM Data Cleansing Challenge

Most enterprise organisations have data flowing in from hundreds of sources, sometimes thousands. Only around 20% of it is structured. The remaining 80% is unstructured, meaning it doesn’t fit neatly into a database’s rigid format.

Cleaning all of that “messy” unstructured information to fit a rigid structure is both prohibitively expensive and almost certainly unfeasible. And besides, it’s not always the right move.

Unstructured and semi-structured data are a goldmine. One that’s remained largely untapped – until recently.

Deloitte reports that just 18% of organisations use unstructured data. Yet those that do are 24% more likely to have exceeded their business goals.

AIs embedded in engagement platforms, CRMs, and CDPs are already helping to surface insights from unstructured data.

Access isn’t the challenge. It’s knowing what to keep and what to purge. That’s where we start to challenge conventional wisdom around CRM data cleansing.

Why Common Sense in CRM Data Cleansing Is Only Half Right

You’ve heard it a hundred times: clean, structured data is best, and you should regularly clear out the excess. It’s not completely wrong. But it is only half the picture.

Although an abundance of duplicates, missing fields, and errors is a problem, there’s a difference between bad data that bloats your database and messy data milling around in a data lake.

Finding that line is the key to distilling value from unstructured data.

Cleaning vs Distilling CRM Data

Cleaning CRM data means keeping the database accurate by removing errors, fixing duplicates, filling missing fields, and standardising structured data formats.

Distilling CRM data extracts usable insights from unstructured or semi-structured data, often to inform strategy, segmentation, and decisioning.

To distil messy data into usable insights, you need to know what you’re looking for.

Unstructured data is worth holding onto if it earns its keep. In other words, all the data retained under your CRM data cleansing policy should have a purpose.

The issue you might run into is that unstructured data doesn’t look useful in its original form. It’s only when you distil it that it starts to shine like gold.

We’re not advocating for a free-for-all by any means. You still need processes, tools, systems, and safeguards for unstructured data. But you don’t need to go as far as deleting or reformatting every iota of information that doesn’t neatly fit a table ~ Neil Hughes

How to Clean Up CRM Data

Start With the Use Case

We consider CRM data cleansing to be effective when it provides the information you need to meet defined, value-adding use cases.
Take the automation, segment, campaign, or report you’re building. Work backwards to define the necessary fields. Then go searching for matches in unstructured and semi-structured data.

Decide Where Data Lives

Databases are designed to hold rigidly structured information in rows and columns. Numbers, inventory, user accounts – anything that needs to be instantly accessible and fits a strict format.

Data warehouses store a mix of current and historical data in fixed schemas. They’re more flexible than databases and are normally used for business intelligence.

Data lakes are more flexible and expansive. They’re designed to hold large volumes of raw, unstructured data that doesn’t need to be processed right away or retrieved in real time.
Each option suits a different use case. An enterprise organisation might combine all three or build a “data lakehouse”.

Just don’t let your lake become a swamp, as this Reddit thread calls out.

Clean Dirty Data

Triggered journeys and scheduled automations need clean data. AI systems too – you can’t build AI agents on patchy data.

This is where cleaning (rather than distilling) data is important:

• Delete duplicates
• Standardising field formats (phone number, address)
• Fixing typos
• Removing outdated information
• Archiving old records and unengaged leads

But only clean the data that’s obviously wrong or actively polluting your project. Tolerate the rest. You never know when it might be useful. Today’s messy field might be exactly what tomorrow’s mix-and-match task needs.

Cleanse for purpose, not for neatness. If a field doesn’t feed the task at hand, its mess may not be costing you anything today, and trying to clean it to fit a structure is likely to be more effort than it’s worth ~ Neil Hughes

Where Do Data Calculations Happen?

The problem usually isn’t the data itself. It’s where the calculations happen and how those tools or systems access the data they need.

Some platforms tolerate ingested mess better than others. For example:

• Braze was designed around common customer journeys; it expects the heavy lifting to be done upstream and the data delivered ready to use in the presentation layer.
• Bloomreach absorbs messier inputs and lets you create calculations and valuations based on customer interactions inside the platform.
• Adobe can pull and transform data from other platforms, where most tools require you to push pre-transformed data to them.

None of these is right or wrong. They’re merely built on different assumptions.
The takeaway is that platform choice and architecture influence how neat your data actually needs to be.

This idea that you don’t need to organise every piece of data might fly in the face of common sense. But once you start to think about the use cases for data that falls outside of rigid frameworks, it’s clear that you’re sitting on a goldmine. The key is how you refine that raw material to get what you need ~ Neil Hughes

So, Do You Actually Have a Data Problem?

CRM data cleansing can seem like an impossible task, especially at the enterprise scale.
Thankfully, your data problem is probably nowhere near as bad as you think.

With a bit of forethought about what matters and what doesn’t, messy data is not only tolerable but valuable.

That doesn’t mean bad data (duplicates, errors, or missing fields) aren’t an issue. But it does reduce the work needed to maintain a clean and usable CRM or CDP.

You can focus on cleaning dirty data rather than reformatting the unstructured information, which makes up the bulk.

If you’re looking to get more from your data, tech and team without scrubbing every record to a shine first, let’s talk.