Your CRM has “New York,” “NY,” and “N.Y.” in the same field. Bringing them together sounds straightforward. Then you find “NY office,” “New York or Boston,” and a blank value.
The first few records need consistent formatting. The others need a decision. Treating every variation as a typo is how a cleanup makes your data look better while making it less accurate.
Standardize values only when you can explain what the original means, what the replacement means, and why they are equivalent.
Separate formatting from interpretation
Formatting changes how a known value is represented. Interpretation decides what an unclear value means. Those jobs should follow different rules.
Removing accidental spaces from a region code may be safe once you understand the field. Converting every mention of New York into a single territory is a business rule. “New York or Boston” might describe two locations of interest, not a misspelled state.
Before building a workflow, write a definition for the property. State what it represents, who uses it, and which values are valid. A field called Location could mean a billing address, a service area, or a preferred office. The workflow cannot resolve that meeting for you.
Then group the observed values into three buckets: safe to standardize, already correct, and needs review. Leave the third bucket visible. It is useful information about where the process is unclear.
Write the mapping before editing records
Create a short mapping document with the original value, proposed value, conditions, and reason. Use exact examples from a review sample rather than inventing a rule from the property name.
For a hypothetical field that explicitly stores a US state, New York and NY could map to the same approved value. For a field that stores office preferences, that same mapping may throw away information.
- Equivalent values: Consolidate only when the business meaning is the same.
- Multiple meanings: Preserve the input and ask for clarification or a better field structure.
- Unknown values: Keep unknown distinct from a confirmed answer such as None.
- Unexpected values: Route them to review instead of assigning the most common category.
Have the person who uses the field approve the mapping. A sales territory owner, for example, should decide whether two labels belong together. Operations can implement that decision without becoming the accidental author of the territory policy.
Record the version of the mapping used for each cleanup batch. When someone asks why a value changed, “because it matched this approved rule” is a better answer than “the automation did it.”
Preserve the evidence before changing the field
Keep a dated export of the affected record IDs and original values before a bulk change. Record the intended replacements as well. This gives you something concrete to compare and a basis for investigating mistakes.
For a complex cleanup, consider writing proposed values to a separate review property first. That is an implementation option, not a requirement to add a second version of every field in your CRM.
Decide when the proposed value becomes the operational value and who approves that step. Also define when temporary review fields can be retired. Otherwise, the cleanup ends with three properties named something like Region final.
Check where the destination property is used before replacing values. A cleaner label can still change which records meet a workflow condition, appear in a segment, or group together in a report.
Use native formatting for rules that are clear
HubSpot's Format data documentation, checked on November 12, 2026, lists Data Hub Professional and Enterprise. The action prepares a formatted output; an additional action, such as Edit record, uses that output to update a property.
The editor can create that Edit record step when you choose to save the formatted value into a property. Check its destination carefully rather than assuming it writes to a separate review field. Custom mode also supports formulas and sample data for previewing results.
Start with those native options when they meet the requirement. A small, approved mapping may need straightforward branches rather than AI. Custom code is useful when you can name a requirement the standard actions do not cover.
Keep the rule and the write separate in your design even when the editor makes them convenient to configure together. Producing a plausible output is one decision. Applying it to a field other processes rely on is another.
Test the values that look inconvenient
Include ordinary inputs and exceptions in the test set. Do not stop after a clean example returns a clean result.
Names deserve particular care. HubSpot documents that its Capitalize first letter option lowercases the other characters and works with Latin-alphabet characters. That behavior may be unsuitable for an acronym, a brand's chosen spelling, or a person's name.
- Blank input: Confirm that missing information does not turn into an invented default.
- Existing standard value: Check that running the rule again leaves a correct value unchanged.
- Mixed content: Include a value with a code and explanatory text.
- Meaningful characters: Check punctuation, accented letters, and leading zeroes where they matter.
- Numeric ambiguity: Confirm the unit and scale before changing a number.
A postal code is not automatically a quantity because it contains digits. A value of 0.5 does not tell you, by itself, whether the source intended a ratio or half of one percent. Use the field definition and source specification, not the appearance of the value.
Compare the proposed output with the expected meaning for each example. Then inspect a controlled batch before extending the rule to the full population.
Fix the input path that keeps recreating the problem
Once the existing values are corrected, look at how new ones arrive. A form, recurring import, integration, and manual entry process may each use a different convention.
Apply the agreed definition at those entry points where practical. Give people understandable choices and explain what the field is for. Check that connected systems use the same mapping, or document the translation between them.
Keep an exception queue for values the rule cannot safely handle. Review recurring exceptions to decide whether the mapping needs another approved case or the property itself needs redesign.
For broader planning, our HubSpot database cleanup guide explains how standardization fits alongside duplicate management, integrations, and ongoing maintenance. A formatting workflow is one part of that work.
Measure accuracy alongside consistency
After rollout, review what changed, what stayed unchanged, and what still needs a decision. Sample corrected values with the team that relies on them. Fewer distinct values is useful only if the remaining categories still describe the business accurately.
If your HubSpot cleanup needs clearer rules before it needs more automation, talk with No Bounds Digital. We can help define the mappings, identify the native features that fit, and plan a controlled update.
